Cecile Malaspina An Epistemology of Noise From Information Entropy To Normative Uncertainty

Copyright © 2018. Bloomsbury Publishing PLC. All rights reserved.
Malaspina, Cecile. An Epistemology of Noise : From Information Entropy to Normative Uncertainty, Bloomsbury Publishing PLC,
2018. ProQuest Ebook Central, http://ebookcentral.proquest.com/lib/ubc/detail.action?docID=5378035.
Created from ubc on 2018-06-06 11:13:45.
An Epistemology of Noise
Also Available from Bloomsbury
Noise Matters, Greg Hainge

Epistemology, Archaeology, Ethics, edited by Sebastian Luft and Pol Vandevelde
Speculative Realism, Peter Gratton
Genealogies of Speculation, edited by Armen Avanessian and Suhail Malik
Metanoia, Armen Avanessian and Anke Hennig
Introduction to New Realism, Maurizio Ferraris
Exceptional Technologies, Dominic Smith
An Epistemology of Noise
Cecile Malaspina
Foreword by Ray Brassier

BLOOMSBURY ACADEMIC
Bloomsbury Publishing Plc
50 Bedford Square, London, WC1B 3DP, UK
BLOOMSBURY, BLOOMSBURY ACADEMIC and the Diana logo are trademarks of

Bloomsbury Publishing Plc
First published in Great Britain 2018
Copyright © Cecile Malaspina, 2018
Cecile Malaspina has asserted her right under the Copyright, Designs and
Patents Act, 1988, to be identified as Author of this work.
Cover design by Catherine Wood

Cover image: ‘inter esse’, Berlin 1985–87 © Maria Sewcz
All rights reserved. No part of this publication may be reproduced or

transmitted in any form or by any means, electronic or mechanical,
including photocopying, recording, or any information storage or retrieval system,
without prior permission in writing from the publishers.
Bloomsbury Publishing Plc does not have any control over,

or responsibility for, any third-party websites referred to or in this book.
All internet addresses given in this book were correct at the time of going to press.
The author and publisher regret any inconvenience caused if addresses have changed or
sites have ceased to exist, but can accept no responsibility for any such changes.
A catalogue record for this book is available from the British Library.
A catalog record for this book is available from the Library of Congress.
ISBN: HB: 978-1-3500-1178-6

PB: 978-1-4411-6532-9
ePDF: 978-1-3500-1179-3
ePub: 978-1-3500-1180-9
Typeset by Integra Software Services Pvt. Ltd.
To find out more about our authors and books visit www.bloomsbury.com
and sign up for our newsletters.
To Andrea
Goya Series: They Do Not Agree, 1997 courtesy of John Baldessari
Contents
Foreword x
Acknowledgements xiv
Note on Text xvi
List of Abbreviations xvii
Introduction 1
Part 1 Concepts: Information Entropy, Negentropy, Noise 13
I How to Draw the Line between Information and Noise 15
II Entropy as ‘Freedom of Choice’ 23
III Information Entropy and Physical Entropy 27
IV The Idea of ‘Potential Information’ 29
V Physical Concepts of Information and Informational

Concepts of Physics 35
VI Information as Process Rather Than Content 39
VII To Think about Information as a Process of Individuation 43
VIII Redundancy and Necessity 51
IX Logic and Freedom of Choice 57
X Noise as Spurious Uncertainty 61
XI Negentropy 65
XII Complexity on the Basis of Noise 71
XIII The Astigmatism of Intuition 79
XIV The Path of Despair 85
viii Contents
Part 2 Empirical Noise 91
I On the Transduction of the Concept of Noise 93
II Accidental Information, Predictable Noise 97
III Ready-Made Information 103
IV Cosmic Background Radiation 109
V Noise in the Gap between Narratives 115
VI Noise in Finance 119
VII Statistics: The Discipline of the Prince 133
VIII The Man without Qualities 139
IX Noise Abatement: The Dawn of Noise 143
X Noise Pollution 149
XI Toxic, Viral, Parasitic 155
Part 3 The ‘Mental State of Noise’ 165
I The Crossroads: Mathematical, Technical, Empirical

and Subjective Noise 167
II Internal Chaos, Terror and Confusion 169

III The Vicious Whir of Sensations 179
IV Keat’s Negative Capability 181
V Closure to Noise and the Paradox of the Declining Life 187
VI The Catastophic Reaction to Noise 191
VII Anxiety 195
VIII Order 199
IX Control 203
Contents ix
X The Helmsman Metaphor: Kybernetes 207
XI The Helmsman in Plato’s Alcibiades Dialogue 213
Bibliography 219
Index 228
Foreword
To turn noise from an object of thought into ‘a variable within the process of
thought’: this is the goal of Cecile Malaspina’s philosophical investigation of
noise – philosophical because it entails transforming noise from an empirical
phenomenon into a condition for the possibility of empirical conceptualization.
Taking Claude Shannon’s notion of ‘information entropy’ as her starting point,
Malaspina shows how the phenomenon of noise harbours a profound philosophical
paradox. Information entropy is a measure of the degree of uncertainty or
‘freedom of choice’ about the state of a message. By aligning information
with unpredictability, Shannon aligns it with uncertainty. But uncertainty
implies ignorance. Thus the concept of information entropy entails this vexing
consequence: if uncertainty indexes information then certainty indexes noise.
But how could certainty, the apex of cognitive aspiration, be a symptom of lack
of information? In contrast to Shannon’s twinning of information with entropy,
Norbert Wiener’s characterization of information as the negation of entropy or
negentropy seems intuitively plausible. Wiener sidesteps the troubling affinity
between information and disorder by confirming our spontaneous identification
of noise with disorder. Yet the curious reversibility between information and
noise remains unaddressed.
Focusing on this reversibility, Malaspina shows how from its inception the
concept of noise as the obverse of information rests upon an equivocation
between order and disorder. This is not merely an equivocation but rather an
essential ambiguity, one that is symptomatic of a latent contradiction in the
concept of noise. Rather than seeking to expose it as a flaw, Malaspina sees
in this contradiction the clue to a deeper truth about noise. Her approach is
dialectical, and the contradictoriness of noise as a concept is the key to its
reality as a phenomenon. Working through this contradiction, Malaspina
patiently unravels the superficial oppositions of order and disorder, certainty
and uncertainty, knowledge and ignorance in all the theoretical contexts
where the distinction between noise and information has been deployed. Her
demonstration traverses information theory, cybernetics, thermodynamics,
biology, psychiatry and sociology, drawing upon such diverse thinkers as
Claude Shannon, Norbert Wiener, Gilbert Simondon, Michel Foucault
Foreword xi
and George Canguilhem. By engaging these sources, Malaspina achieves a

philosophical optic that is genuinely transdisciplinary. (Her achievement in
this regard is perfectly complemented by that of Inigo Wilkins, whose work
resonates beautifully with her own.1) This does not mean fashioning conceptual
hybrids from disparate theoretical discourses in an eclectic and ultimately
opportunistic fashion. Malaspina constructs a philosophical concept of noise
by pinpointing the decisive fault lines in the workings of the various theoretical
concepts whose functioning she carefully delineates. But the dialectical cast
of her thinking renders her stance constructive rather than deconstructive. It
allows her to integrate concepts from the mathematical and natural sciences
alongside those from social and cultural theory. This transdisciplinary
remit makes it possible in turn to articulate the epistemology of noise with
its ontology, which is to say that it encompasses both what noise is and how
it relates to knowing. What is important, from Malaspina’s perspective, is
that this does not so much subvert as politicize the conventional distinction
between epistemology and ontology:
The conceptualization of noise is thus no longer limited to the classical
philosophical problem of determining what we can understand of the reality
of noise ‘in itself ’ or even ‘for us’. It is irreversibly contaminated by a political
problem, which is the possibility of deliberate or accidental distortion also of our
critical faculties through noise. (p.162)
From this follows what is perhaps Malaspina’s most striking insight:
Noise, beyond the reference to unwanted sound, thus reveals itself to be

conceptually polymorphous because it has never been about types, classes or

measures of phenomena that qualify noise as a particular type of disturbance,
but about the relation between contingency and control. (p.203)
For Malaspina, ‘noise’ is not just the name for the force scrambling the recognizable
outlines of phenomena; it designates the anomaly from whence the distinction
between sense and senselessness originates. It is not merely a natural phenomenon
or kind because its co-articulation with information is the consequence of an act of
judgement, rather than the registration of a fact. Thus noise is a normative rather
than a natural category, which is to say that it is made not given. The empirical
discrimination of noise presupposes the normative establishment of its difference
from information within a given disciplinary framework. But this difference –
between control and contingency, determination and indetermination – follows
from what Malaspina calls ‘a suspension in indecision’ or ‘unthinkable freedom
of choice’ that is not of the order of structure or destruction. Precisely because it
xii Foreword
cannot be naturalized, objectified, catalogued or classified, this anomalousness,

from which judgement proceeds, can only be acknowledged by shifting from a
perspective that treats noise as object to one that treats it as subject, or in Malaspina’s
words, ‘the noise of cognition constituting itself, against the always looming crisis
of its dissolution’ (p.173). Malaspina distinguishes two ‘subjects’ of noise here: the
subject of cognition constituting itself and the experiential subject undergoing
the legislated difference between information and noise. The subject of cognition
enforces conditions of disciplinary regulation through which information is first
identified. The experiential subject is the locus for the psychological, cognitive
and affective ramifications of the experience of noise, which Malaspina explores
through a particularly inventive reading of Steven Sands and John Ratey’s seminal
1986 article, ‘The Concept of Noise’.
Ultimately, Malaspina’s book is an epistemic intervention. Knowledge is
‘of ’ uncertainty in both the subjective and objective senses of the genitive. It
is uncertainty that knows. The subject of cognition exercising the power of
judgement requires that the normative preconditions of cognitive judgement
determine the constitution of empirical facts. This is the grounding power of
judgement. But this grounding power is itself based on ungrounding as condition
of normative grounding. This ungrounding is a consequence of the ineliminable
role played by contingency in the exercise of determining judgement. This is
why, for Malaspina, ‘noise is, like disorder, an inconceivable freedom of choice’
(p.187). Knowledge is not the reduction of uncertainty because it is constituted
by it. Thus knowledge is neither solely predictive nor exhaustively fallible, no
more than it is either verifiable or falsifiable. It is ‘of ’ uncertainty because it is

rooted in this ‘inconceivable freedom of choice’.
Malaspina uncovers the productive, form-generating powers of epistemic
noise as the occluded source of cognition’s predictive capacities. Cognitive
invention is not rooted in the ‘negentropic’ negation of contingency; it proceeds
from the ‘negation of the negation of contingency’ (p.183). Regulation is not
the precondition for innovation; innovation gives rise to regulation through the
lawless collapse of regularity. Invention, whether cognitive, aesthetic or political,
is not the negation of disorder but the negation of its negation, which is to say,
the negation of order. Malaspina achieves a properly dialectical resolution of
the tension between the negative physical characterization of noise as form-
destroying entropy and the positive aesthetic valorization of noise as form-
generating novelty: radical transformation (whether cognitive, aesthetic or
political) can only arise through the unconditioned judgement that affirms the
irreconcilable tension between the destruction and generation of form. This
Foreword xiii
judgement, constituting cognition against the backdrop of its dissolution, is a

function of the noise that enables the process of thought.
Note
1 Inigo Wilkins, Irreversible Noise, Falmouth: Urbanomic, forthcoming.

Acknowledgements
My first expression of gratitude goes to Prof Alain Leplège and Dr Iain Hamilton
Grant (U.W.E.) for the freedom they granted me and for their unwavering
trust in supervising the doctoral thesis on which this book is based. Prof Ray
Brassier, Prof Emmanuel Picavet and Dr Matthieu Saladin are warmly thanked
for their questions and insightful comments. Frankie Mace at Bloomsbury,
as well as Deepakraj Murugaiyan have my heartfelt thanks for their support
and inexhaustible attention. I thank also the photographer Maria Sewcz for
generously putting her work ‘inter esse’ at our disposition for the cover of
this book and John Baldessari, for letting us use his ‘They do not agree’ as the
frontispiece. Catherine Wood paid particular attention to the artists wishes in
designing the cover.
The Reverberations conference organized by Dr Benjamin Halligan, Dr
Paul Hegarty and Dr Michael Goddard at Salford University in 2010 has been
determining for the transdisciplinary perspective of this book. I am grateful
for their editorial support going into the publication of my contribution to
Reverberations, the Philosophy, Aesthetics and Politics of Noise with Continuum
in 2012. The intense exchange about noise and art with Dr Michael Schwab and
all collaborators in the Data-Rush symposium organized in Vienna in 2016,
especially Prof Mauricio Suarez, and also Dr Paulo de Assis and Tiziano Manca
at the Orpheus Institute in Ghent have been immensely enriching. I am also
especially grateful for a host of new references and ideas I owe to the generous
suggestions of Prof Christian Walter and Prof Emmanuel Picavet at the Chaire
Ethique et Finance, College d’Études Mondiales, Fondation Maison des Sciences
de l’Homme, Institut des Sciences Juridiques et Philosophiques de la Sorbonne
(UMR 8103).
I am deeply grateful also to Dr Anne Lefebvre, for years of collaboration and
dialogue, for her invitation to speak at the Van Eyck Academy in Maastricht
in 2012 and for her initiative, to which I owe the opportunity of testing early
ideas at the transdisciplinary seminar at the Ecole Normale Superieure in Paris
in 2010, whose organizers Prof Claude Debru, Prof Jean-Charles Darmon and
Prof Frédéric Worms are also warmly thanked. Also the European Meeting for
Research in Systems and Cybernetics (EMCSR) in Vienna in 2012, 2014 and
Acknowledgements xv
2016, as well as the Schelling Grundlagen Seminars at the Institute for Design
Science, Munich, and have been important milestones. Prof Rainer Zimmermann
is warmly thanked, alongside Stefan Blachfellner, director of the Bertalanffy
Center for the Study of Systems Science, as well as Dr Jose Maria Diaz Nafria
and the BITrum consortium. Dr David Rousseau, editor at Systema: Connecting
Matter, Life, Culture and Technology, has my gratitude for his editorial advice on
the publication of my article on epistemological noise, which has enabled me
to articulate one of the core ideas going into this book. Not least do I thank all
those not mentioned here and not directly cited, whose thought has illuminated
the questions I could tackle, but also those questions that motivated me and that
remain in the undergrowth.
Amelie Mourgue-D’Algue, Sissi Taseva, my parents and brothers have
indefatigably supported, if not freed me to work on this book. To my sons
Federico, Paolo and Olivier I owe the greatest debt of gratitude, for enduring
the period of research and writing and for encouraging me at critical times.
To Andrea I owe the insouciance of beginning this work and the courage of
finishing it.
Note on Text
All quotations referencing texts with German or French titles are translated by
myself.
List of Abbreviations
ILFI Simondon, Gilbert. 2005. L’individuation à La Lumière Des Notions de

Forme et d’information. Grenoble: Editions J.r.me Millon.
METO Mode of Existence of Technical Objects, Gilbert Simondon, trans.

Cecile Malaspina and John Rogove, Univocal Publishing, 2017
MTC Mathematical Theory of Communication, Claude Shannon and

Warren Weaver, University of Illinois Press, 1964
NP The Normal and the Pathological, Georges Canguilhem, trans.

Carolyn R. Fawcett and Robert S. Cohen, Zone Books, 1991
WHO World Health Organization, http://www.who.int/about/en

Introduction
It has become commonplace to use the word noise, almost with inverted comas,
in a host of contexts unrelated to sound, often in opposition to information. It is
thus not the din of the trading floor that interests us when we talk about noise in
finance, but the uncertainty related to random variations in the stock exchange.
Noise has become a concept intrinsic to the statistical analysis of the variability of
data in almost every domain of empirical enquiry. Even acoustics can be argued
to have fully emerged only during the 1950s, when noise could be represented
as graphs of the frequencies and amplitudes of transitory signal changes over
time (Castellengo 1994). That these two dimensions of the conceptualizations
of noise, as sound and as random variation, speak to each other without being
reducible to one another is what this book is about.
This new, statistical meaning of noise is first and foremost the expression
of one of the most profound methodological transformations of the modern
sciences. Predating cybernetics and information theory, the source of today’s
understanding of noise as inextricably linked to variation or even error goes

back to the origin of the calculus of probability in games of chance, notably in
the work of Pascal and Bernoulli’s law of large numbers – also called the ‘law
of possibilities’. Interpreted by Laplace as an adequate representation of ‘errors
in measurement in astronomy’, the ‘law of possibilities’ has subsequently been
called the ‘law of errors’ (Desrosieres 2006). However, the definition of statistical
noise we inherit from the ‘law of errors’ must not obscure the fact that in modern
statistics precision itself has become a question intimately linked to noise:
Precision is a measure of random noise. (S. Smith 2002, 34)
The special sense of the word noise thus implies both a methodological
transformation and a new scientific status of the notions of uncertainty,
probability, and error in relation to statistical averages. Thus enriched, the
subsequent definition of noise in cybernetics and information theory has come
2 An Epistemology of Noise
to retrospectively encompass also the concept of physical entropy, and more

generally of uncertainty, statistical variation and error. No longer considered
only as a factor of disturbance, detrimental to information like ‘static noise’
in the channel of communication, the evolving concept of noise also becomes
constitutive of new forms of knowledge and of new ways of understanding
organization.
This new connotation of noise has entered ordinary language as a side effect
of the ‘information paradigm’ we have inherited from information theory and
cybernetics (Malaspina 2012a; Morange 2006). Although Raymond Ruyer noted
already in 1954 that cybernetics had failed to impose itself as a transdisciplinary
scientific paradigm, the notion of noise has nevertheless steadily gained
prominence over the past decades as a notion bridging disciplines: not only in
relation to computer science and even complexity theory, but across the natural
and human sciences and the arts, ideas of ‘order from noise’ or ‘complexity on
the basis of noise’ are steadily rising to greater prominence (Atlan 1979; Mersch
2013; Nunes 2012; Ruyer 1954).
The term inforgs has since been coined to emphasize the idea that we no
longer inhabit only an ecosphere, but also an infosphere (Floridi 2002). Yet
recent developments appear to suggest that, far from the dawn of an Information
Enlightenment, we seem to sleepwalk into an era of noise: levels of stress and
depression are rising as we pine under both noise pollution and information
overload (Bawden and Robinson 2009; Berglund and Lindvall 1995). Even
intelligence services suffer from too much information. Big data means that
noise can be harvested through data mining, but vast amounts of data once
more become noise as soon as we lack pertinent criteria to transform them into
information (Watkins 2011, 31).
A curious reversibility of information and noise thus becomes apparent: too
much information, and also the repetition of the same information ad nauseam,
becomes noise, whereas information that is radically new falls on deaf ears
when context and criteria of pertinence are lacking to adequately distinguish
information from noise.
Despite the ever more apparent complexity of the relation between
information and noise, the latter is often taken for granted as the mere opposite
of information, based on the intuitive analogy with acoustic noise disrupting
communication. What risks being overlooked in this simplistic opposition
between information and noise is a palimpsest, a rich layering of intuitive
notions of the still and the perturbed, the clear and the turbid (from Latin
turba: crowd), opaque or confused. This opposition is also rich in ideas that
Introduction 3
have a proud history at their heels – such as order and disorder, work and futility
(the latter indicating a leaking, untrustworthy vessel in medieval alchemy)
(Watkins 2011, 31) – and rich also in mathematically formalized concepts, like
Ludwig Boltzmann’s formalization of statistical entropy.
In this palimpsest of concepts, notions and ideas, noise always appears to
occupy the negative place of a dichotomy, be it in that of order and disorder, of
physical work and the dispersion of energy in the state of entropy, or of the norm
and the abnormal. In other words, noise is at best associated with the absence
of order, of work or of the norm – be it the statistical, moral or aesthetic norm –
and at worst, noise is identified as a threat to the norm and subversive of work
and order: a perturbation, a loss of energy available for work, a parasite.
Noise is thus a word that implicitly plays on the whole register of notion, idea
and concept and does so by mobilizing linguistic, historical, sociopolitical and
not least of all epistemological registers.
If we are to understand the new fortunes of the previously reviled and now
revisited idea of noise – from physics to information theory and cybernetics and
beyond – then we must not only disentangle these notions, ideas and concepts,
but also analyse the subtle ways in which new concepts of information have
rewired our conceptions of noise: starting with the concepts of ‘information
entropy’ and negentropy, which is what the first part of this book sets out to do,
before looking at some cases of empirical noise in Part II (from the discovery
of cosmic background radiation to noise pollution and the historical origin of
Statistik as the nomenclature of knowledge necessary for the sovereign) and
finally at the role of noise in the process of cognition itself, by focusing (in Part
III) on the idea of ‘the mental state of noise’, developed in 1986 by S. Sands and
J. Ratey to describe
an internally experienced state of crowding and confusion created by a variety of

stimuli, the quantity, intensity and unpredictability of which make it difficult for
individuals so afflicted to tolerate and organize their experience. Attempts to do
so may only add to confusion and psychotic phenomena. (Sands and Ratey 1986)
Yet before fanning out the whole spectrum of resonance of today’s notion of
noise, it is important to seize the precise moment noise erupts as a key concept
in science and technology.
Claude Shannon famously devised a mathematical theory of signal
transmission that paved the way for the effective elimination of noise from
the channel of communication. According to Claude Shannon’s Mathematical
Theory of Communication (MTC), information can be defined, in terms of
its probability, as a measure of ‘information entropy’. Let it suffice to say here

that ‘entropy’ indicates what Warren Weaver, in his introduction to MTC, calls
‘freedom of choice’, relative to the unpredictability of a message. What this means
is, critically, that a piece of information informs us only, if it is not redundant, in
other words, if it contains a margin of unpredictability and hence uncertainty.
The status of this physical concept, ‘entropy’ – coined by Rudolf Clausius to
describe the loss of available energy in thermodynamic terms – requires careful
examination, especially when we consider the inflection that information
theory has given to the way we use the word noise in the natural and the human
sciences. Today the dictionary defines entropy as a state of molecular ‘disorder’
(Larousse 2017). Ludwig Boltzmann’s great innovation during the nineteenth
century was to devise the statistical formulation of molecular entropy, which
became the basis for Shannon’s formalized concepts of both information and
noise. Yet despite the high degree of formalization, the intuitive idea of disorder
continues to colour our idea of entropy and, consequently, both Shannon’s
concepts of information and noise.
The persistence of a persuasive and intuitive idea of disorder when we refer
not only to noise but also to ‘information entropy’ may help explain why, despite
the striking clarity of Shannon’s idea – namely that information must tell us
something new, something we could not predict – his definition of information
as ‘information entropy’ has failed to impose itself outside the mathematical
theory of communication (MTC). What appears to have dampened the reception
of Shannon’s concept of ‘information entropy’ is that the correlation between
novelty of information and disorder also threatens the clear-cut conceptual

opposition between information and noise.
Norbert Wiener’s concept of information as, on the contrary, the negation
of entropy has been adopted across the natural and human sciences through
the neologism negentropy, coined by Leon Brillouin. A new understanding
of information imposed itself as the negation of entropy, and more generally
as the negation of disorder, meaning negation of everything contingent or
unpredictable. The value of entropy thereby becomes a measure of unwanted
variability, imprecision or error – in any case, a value to be eliminated for the
sake of efficiency and certainty: entropy henceforth becomes synonymous
with noise. Through the broad success of Wiener’s cybernetic theory of self-
regulating systems with feedback, the concept of negentropy has found its way
quite naturally into our thinking of organization and information in general.
Indeed, the very notion of a system, any system, can be put in cybernetic
terms as a set of organized constraints on contingency, in other words, as
Introduction 5
the organized negation of noise. As negation nestles at the very core of the
cybernetic concept of information, information comes to reflect the level
of organization of any system, insofar as it is apt to negate its spontaneous
progression towards entropy. Noise, in turn, becomes a metaphor for entropy
as the chaotic dispersion of energy, as disorder, if not as the entropic ‘death’ of
a system.
Such an impoverished view of information, impoverished because lacking
in the complexity that entropy contributes, of course fails to adequately
represent the theoretical wealth of Wiener’s own approach, and of the
subsequent development of cybernetics into second-order cybernetics (i.e.
the cybernetics of self-observing, self-regulating systems with feedback) as
well as of more recent developments in complexity theory. The point is, and
will be throughout, that concepts circulate through general discourse and
that general discourse in turn leaves its mark on the circulation of concepts:
as the metaphor of noise as ‘parasite’ in the channel of communication
started thriving, the idea of negentropy disseminated itself across the natural
and human sciences and general discourse, often without its mathematical
formulation, and frequently without being directly acknowledged. This early
formulation of one of the key concepts of cybernetics has thereby contributed
to polarize our epistemic field in its relation with the unpredictable and the
improbable.
Now widely diffused, what subsists in general discourse of Wiener’s idea of
information as negentropy subtly inflects our thinking about organization: from
the organism to the ecosphere, from sociopolitical to economic relations, from

networks to the idea of globalization. As a result, by emphasizing the negation of
contingency our idea of information has become tethered to predictability and
consequently antithetical to noise as the unpredictable.
And yet our narratives of prediction and self-regulation have failed both
spectacularly and catastrophically before the crises that have inaugurated the
twenty-first century (Walter and de Pracontal 2009) – heralding a ‘post-truth’
era of politics, catastrophic crises in finance, war and migration. The time
has come, it seems, to re-evaluate the epistemological import of Shannon’s
entropic idea of information, and to do so in light of this new protagonist
concept: noise.
A closer look at both information theory and cybernetics reveals that
the opposition between Shannon and Wiener’s mathematical approach to
information and noise is far subtler than it seems. But even then, the consequences
of this difference in thinking about the relation between information
and noise remain significant. They inform the widely diffused integration of the
concepts, both negentropy and noise, into theoretical contexts that are neither
fully mathematized nor reducible to the technical idea of ‘noise in the channel of
communication’ (Morange 2006).
To understand the conceptual ramifications of noise thus requires a careful
evaluation of the relation between physical entropy and ‘information entropy’.
Only when the moral and perhaps even ideological connotations of the notions
of ‘organization’, ‘work’ and ‘order’ are elucidated in their relation to predictive
certainty can we begin to understand how noise, alongside concepts like
‘metastability’ or ‘non-linearity’, could become common parlance in business
management and political lobbying alike. Take, for instance, David Cummings
reflections on the ‘Vote Leave’ campaign, which he directed leading up to the
Brexit referendum in June 2016. Cummings explains that the success of the
campaign was driven by new communication technologies and the targeted use
of social networks. His account of the success of the ‘Vote Leave’ campaign is
laced with the words ‘non-linear’, ‘interdependent’, ‘unpredictable’, ‘irrational’,
‘complex’ and even ‘noise’:
A news broadcast now contains much less information content and much higher
noise than reading. The only way to improve this is experimenting with formats
in a scientific way. (‘Dominic Cummings: How the Brexit Referendum Was
Won’ 2017)
The politics of information and noise are thereby elevated to a pseudo-scientific

status. However, what this pseudo-scientific smokescreen dissimulates is the
oldest trick of the trade: xenophobia, the fear of the other. To acknowledge and
analyse the interwoven nature of scientific, technological, moral and political
components of the conceptualization of noise is therefore indispensable.
It is all the more important, therefore, to stop and pause before the rapid
proliferation of the idea of noise across the natural, the human sciences and
public discourse. It is important to pause here for two reasons. As we have
seen, the noise metaphor endows discourse with a scientific aura, lending it
authority beyond the limits of its rational means. What is perhaps less obvious
are the moral and ideological inclinations that can, in turn, also affect scientific
discourse – for instance, when noise is associated with irrelevance, abnormality
or disturbance.
The second reason is more properly philosophical because it alerts us to an
aspect of the theory of knowledge that can no longer be sidelined as marginal.
The ambiguity that accompanies the flurry of conceptualizations of noise does
Introduction 7
not belong in the margin of error of scientific discourse, if we recognize instead

that ambiguity is the space that we must negotiate as part of the moving frame of
debate shared by contemporary science and general discourse.
It is no longer just the question of how we define empirical noise, but of how
the theory of knowledge handles ambiguity. In other words, we are looking at an
epistemological problem of noise.
The first thing that becomes apparent upon closer inspection is that a shared
formal definition of noise is lacking. This lack opens a space for metaphorical
reverberation within scientific discourse, and even more so in the straits between
the natural and the human sciences, technology and the arts. Inigo Wilkins notably
mounts a thorough critique of the metaphorical distortions of the concept of
noise in the humanities in his Irreversible Noise (Wilkins forthcoming). Wilkins
thereby aims to redress the fetishization of noise as randomness and chance,
and its reduction to the unintelligible. By situating the concept of noise within
the context of contemporary science, and by re-centring it around mathematical
definitions of randomness in the context of complex dynamical systems, Wilkins
builds a case for a concept of noise that can act, instead, as an index of intelligible
constraint. Thus re-enforcing the mathematical and scientific parameters of
the concept of noise, Wilkins emphasizes both formal (logical) and empirical
(scientific) criteria by which to assess the critical potential of conceptualizations
of noise in the humanities.
The conceptual building site of Wilkin’s approach could be said to coincide with
the present approach insofar as both problematize the multiple conceptualizations
of noise. However, the problem addressed here differs in perspective. Rather

than making metaphorical ambiguity the target of elimination, it is here taken
to be a relevant philosophical problem in its own right. The objective here is thus
not to eliminate ambiguity in the conceptualization of noise, but rather to make
it explicit so that we can get to grips with it as a form of epistemological noise
that accompanies the transformation of the epistemological field, especially in
the context of today’s increasingly translational and interdisciplinary platforms
for research.
Metaphor, of course, remains a dirty word in the context of scientific and
even of much philosophical discourse. Yet it is necessary to acknowledge that
metaphors are used abundantly, if artlessly, in scientific and philosophical
discourse. Being in denial about the critical role of metaphors, especially in the
communication of specialist knowledge to non-specialists, and in the context of
growing interdisciplinary consortia, is to submit uncritically to their rhetorical
power.
Metaphors of noise are not sufficiently subjected to critique, notably because

the humanities are kept at bay. The much talked about rapprochement of
the natural and the human sciences is currently still confined to making the
humanities, especially psychology and sociology, more empirical, weighting
analysis towards quantitative, rather than critical discursive, analysis. Yet
what we need is precisely a return to the critical lessons coming from the
literary end of the humanities: what we need is to learn how to use metaphors
critically, purposefully and artfully. Rather than allowing the metaphor of
noise to burrow its way into the cracks of scientific discourse, like a repressed
and unacknowledged fear or desire, we must learn to cultivate a critical use of
metaphors in public discourse within and on science and technology. We thereby
honour the epistemic humility already practised in the sciences but also in the
arts, where it has long been established that reality speaks to us through the art
of framing, as much as through what is framed.
At stake in this book is the shifting boundary between information and noise
and the sprawling of the idea of noise, in the guise of its many technical and non-
technical definitions. The epistemological, moral and political implications of
its impermanence are the red thread that connects the three parts of this book.
It is precisely the difficulty in its conceptual delimitation, especially in light of
the idea’s projection across the academic and professional boundaries of diverse
theoretical and experimental fields, that makes noise a privileged philosophical
problem. This restlessness of the conceptualizations of noise, the shifting
boundaries between what we consider to be information and what we discard as
noise, requires that we think about this restlessness as a form of epistemological

noise. Meaning quite simply that the communication between diverse theoretical
and experimental fields is not only subject to conceptual ‘noise in the channel of
communication’, but also generates epistemological noise. The very movement
of the idea of noise across disciplinary boundaries, the conceptual distortions
provoked by this movement, is a form of epistemological noise accompanying
its dissemination and transformations. In other words, the unstable concept
of noise is itself an example of epistemological noise in the communication of
concepts across theoretical boundaries.
The conceptual distortions this sprawling provokes, and the metaphorical
and ideological inclinations it reveals, could be said to act as what the French
epistemologist Gaston Bachelard called ‘epistemological obstacles’. Yet rather than
chastising interdisciplinary or transdisciplinary discourse for its imprecisions in
the use of the term noise, or even its relapses into proto-mythological thinking
(for instance, in the rapprochement between noise, disorder and chaos), the
Introduction 9
emphasis here is on the epistemological necessity of cross-fertilization between

the diverse theoretical and experimental fields, in other words: on the co-
constitutive role of noise in the formation of knowledge. This is what will be
called epistemological noise.
In order to understand the proliferation of ideas and concepts around noise,
the focus thus cannot be exclusively technical. This would be to overestimate
the fidelity of conceptualizations of noise to mathematical formalization, or
even to the technical applications the term noise has found in computational
logic and in new technologies of communication. However, also literary and
artistic evocations of noise are insufficient on their own, if they are limited
to a Romantic indulgence in noise as indicative of the incommensurability
of Being. If noise becomes the placeholder concept of a philosophical Other,
as that which does not submit to reason, then also hopes invested in its
revolutionary potential risk petering out in enthusiasm. What is more, the
temptation to indulge in noise as the mere negation of limit (embracing only
the Greek apeiron, the unlimited) or of established norms (as the abnormal)
squarely inhabits the conservative logic of negation. In short, it fails to subvert
the very logic that the idea of negentropy occupies in the cultural and scientific
imaginary.
The present conceptualization of noise owes much to the attention paid in
French epistemology and contemporary French philosophy to the ‘shifting
sands of emergent truths’, as Alberto Toscano aptly expresses it in his translator’s
introduction to Alain Badiou’s Being and Event:
For millennia, philosophy has attempted to ground itself on One Eternal

Necessity such as the prime mover, or the dialectic of history. Here it consciously
chooses to ground itself on the shifting sands of emergent truths. (Badiou 2005,
xxiii)
Readers may recognize echoes of Badiou’s insistence on the question

‘what counts as One?’ when we enquire into the question: what counts as
information, and what can be discounted as noise? Also a post-Cartesian
perspective on the subject, such as it animated a certain generation of
French philosophers and theorists, will come into play when we consider the
‘mental state of noise’ in light of John Keats’s definition of the poet’s negative
capability. However, the way noise will be problematized here, as a polyvalent
and polymorphous concept, will not base itself on equating mathematics
with ontology. Necessarily fanned across a wide range of topics, the problem
considered here, i.e. the moving boundary between information and noise,
will be treated more modestly like a minimal gesture in philosophical terms, a

way of reiterating the act of drawing a line from various angles, rather than an
epic tableau of noise in the grand genre of systematizing philosophy (Badiou
2005, 15).
This minimal gesture of repeatedly drawing the line between information and
noise in various contexts, has the advantage of revealing a margin of conceptual
indeterminacy between diverse fields of knowledge. To take stock of the
epistemological noise that arises from the inevitable sharing of concepts and from
the unavoidable recourse to common language, helps us understand the growing
complexity of the field of knowledge as a whole, in analogy to the way in which
the philosopher and bio-physicist Henri Atlan speaks of ‘complexity on the basis
of noise’ (Atlan 1979). Such an approach to the theory of knowledge, as engaging
all fields of knowledge in their plurality, takes in its stride the shift from an ideal of
knowledge without noise, indebted to the Cartesian Method, going towards an idea
of knowledge that gains in complexity by being exposed to epistemological noise.
To re-evaluate noise as a problem of epistemic complexity is to acknowledge the
functional role of uncertainty and ambiguity in the process of concept formation.
Michel Foucault’s introductory words to George Canguilhem’s re-edition of The
Normal and the Pathological here help to express the motivation for this book:
Error is at the root of what makes human thought and its history.
(Foucault 1989, 22)
This is how noise can be understood, ultimately, as a radical concept, in the

sense that Foucault and Canguilhem understand error, as touching the root
of human thought and of its historical irreversibility. At stake is the relation
between thought and contingency, which has become emblematic for a certain
way of thinking about the philosophy and history of the sciences.
Yet, while the swarming interest in noise makes it an imperative to engage with
it conceptually, the synthetic view that is called for is, by definition, condemned
to fail in making even a dent in any of the individual fields of knowledge and
practice that gravitate around the notion of noise: this book will not improve
stochastic models of noise, it will not resolve new problems of noise in big data,
nor will it improve propositions to tackle noise pollution – least of all will it
attempt to tell artists and musicians, or cultural and critical theorists, how to
conceptualize noise. In fact, it cannot even begin to do justice to the extent of
diffusion of the notion of noise to other disciplines, which is in a process of
active fomentation, expansion and dispersion.
Introduction 11
While this book owes everything to the growing wealth of literature on

noise, the conceptual movement I am after is not primarily concerned with the
knowledge of diverse phenomena understood as noise, but with the idea of noise
in the relation between the known, the unknown and the differently known.
This is why, unlike Greg Hainge’s journey through the many dimensions
of noise in his Noise Matters (Hainge 2012), this book cannot aim at what
he calls an ‘ontological taxonomy’ of noise, because this is predominantly an
epistemological enquiry, rather than an ontological one. Rather than aiming
at a phenomenology of noise (Cage et al. 2012; Voegelin 2010), and despite
benefiting hugely from the vast spectrum of literature on acoustic noise, on the
cultural (Bijsterveld 2001; Boutin 2015; Gibson and Biddle 2016; Schafer 1994;
M. M. Smith 2004) and even military history of noise (Volcler and Volk 2013)
and the impact of cybernetics and information theory (Bunz 2012; Mersch
2013) as well as the psychology of perception (Bawden and Robinson 2009;
Manson 2014; Shenk 1997), the problem of epistemological noise, as posed here,
is ultimately co-extensive neither with a phenomenology of noise, (because the
question posed here starts from Shannon’s counter-intuitive relation between
information and noise), nor with its cultural history, (insofar as it focuses on the
conceptual implications of thinking about noise). And although noise music and
noise art are what opened my mental shutters to the prospect of thinking about
noise, this book is about neither, leaving this avenue open for future projects
(Attali 1985; Brassier 2007; Hegarty 2007; LaBelle 2006).
This book looks also, albeit obliquely, at the emerging field of the
philosophy of information, and more specifically at its effort of reconciling

information concepts underlying science and technology with the humanities.
Information philosophy has recently established itself as a specialist discipline
in philosophy. It comprises new fields that have arisen from thinking about
electronically mediated information, dealing with the transmission, circulation
and conversion of one form of information into another (Dodig Crnkovic and
Hofkirchner 2011). Its teeming developments are so diverse that they can barely
be overseen and do not yet appear to constitute a coherent whole in the eyes
of its contributors – encompassing fields as varied as logic and computation,
cognitive and neurocognitive sciences, dynamical systems and actor network
theories, cybersemiotics and biosemiotics, information systems and epistemology,
and information culture and information ethics (Capurro and Hjorland 2003;
Diaz Nafria 2010; Floridi 2010).
Leaving many references, precious ideas and references out, indeed cutting
large chunks of the work that prepared this journey into noise was the necessary
sacrifice so as to allow the form of the argument to emerge. This cut is but the
performative aspect of the problem this book ultimately faces: how do we draw
that line that makes the form of an argument emerge, even an argument about
noise? What can we afford to exclude? How much variety, and hence how much
uncertainty can we retain, without dissolving the very movement of thought,
whose emergence we only begin to comprehend? In this sense, we will ultimately
come to think of maximum noise as an unthinkable freedom of choice.
While this book can claim none of the academic fields it visits as its own, it
seeks to understand the problem of the conceptualization of noise as a problem
that relates them, without reducing them to any single dominating view. The
oblique relation between these multiple domains requires that we understand
the resonance of the idea of noise as something that, like the reverse of a carpet,
reveals the messy connections that sustain the neatly separated forms of the
academic organization of knowledge. To look under the carpet no doubt implies
also a certain impertinence towards the well-established and well-deserved
boundaries of specialist knowledge, at the risk of necessarily exposing one’s
ignorance in comparison to those who have laboured hard to establish a more
secure basis of expertise in any one of these fields. Perhaps the risk implied
in being – not unlike noise – excessive of boundaries of discourse, resonates
with George Canguilhem’s lightly humorous concession that ‘the philosopher is
indiscrete everywhere’ (Canguilhem 1993, 19).
Part One
Concepts: Information Entropy,

Negentropy, Noise
I
How to Draw the Line between

Information and Noise
Draw a straight line and follow it.

La Monte Young, ‘Composition 1960 #10 to Bob Morris’
The conceptualization of noise takes a new turn in relation to Claude Shannon’s

definition of information as ‘information entropy’ – this much is certain.
Just what it means to rethink noise in relation to ‘information entropy’ is the
question posed in this first part of the book. The aim here is not to elucidate
the concepts of information and noise to the engineer who is in no need of
such speculations for practical purposes, but to understand how this new way
of thinking about information feeds into the broader scientific and cultural
understanding of noise. The latter turns out to be no passive receptacle of
technoscientific concepts, but feeds back into the new technologically inspired
understanding of noise, creating a new culture for theoretical and experimental
practices.
Shannon’s audacity consists quite simply in correlating both information
and noise with uncertainty. Both concepts are henceforth derived from the
statistical unpredictability he associates formally (mathematically) with
physical entropy. While information entropy clearly implies a degree of desirable
uncertainty, i.e. the novelty of the message, Weaver will say that noise can be
discarded as ‘spurious uncertainty’. Yet it is, in both cases, unpredictability that
is expressed via the calculus of probability and statistical analysis, constituting
what is called ‘entropy of the message’. As Weaver explains in his introduction
to the second edition of Shannon’s Mathematical Theory of Communication
(MTC) of 1964:
The quantity which uniquely meets the natural requirements that one sets up for
‘information’ turns out to be exactly that which is known in thermodynamics as
entropy. It is expressed in terms of the various probabilities involved – those of

getting to certain stages in the process of forming messages, and the probabilities
that, when in those stages, certain symbols be chosen next. (Shannon and
Weaver 1964, 19)
Before getting a better grasp of the status of the concept of ‘entropy’, both as a
concept in physics and as a metaphor in statistical analysis, we will compare two
statements further down, one by Warren Weaver in his introduction to Shannon’s
MTC and one by Norbert Wiener in his book Cybernetics (Wiener 1961). These
two statements show that there was no disagreement between those generally
acknowledged as the founders of information theory and cybernetics respectively,
regarding the method of calculating information probability; however, they
also reveal the fact that the same mathematical method nevertheless justifies
two diametrically opposed definitions of information: one of information as
‘information entropy’ and the other, on the contrary, of information as the
‘negation of entropy’. Noteworthy is that these radically opposed definitions of
information did not appear to constitute a problem even worthy of mention by
either Shannon or Wiener.
The introduction to Shannon’s MTC in fact begins by acknowledging Shannon’s
conceptual debt, not only to Wiener’s mathematical work, but to his philosophy
(Shannon and Weaver 1964, 3, n. 1). And yet, it proceeds to define information
positively as a measure of entropy, and entropy as a ‘measure of one’s freedom of
choice’. Contrary to Wiener’s definition of information as the negation of entropy,
for Shannon, greater information goes hand in hand with greater uncertainty. A
completely predictable message, by contrast, has only one possible outcome and
is therefore redundant; it tells us nothing new. In Warren Weaver’s words,
[I]nformation is a measure of one’s freedom of choice (p.9) […] in these
statistical terms the two words information and uncertainty find themselves to
be partners (p.27) […] entropy (or the information, or the freedom of choice
[…]) (p.13). (Shannon and Weaver 1964, 9–27)
In other words, the redundant message presents no ‘freedom of choice’,

because it contains no ‘information entropy’. Information is null, if there is no
uncertainty about the state of the message. Norbert Wiener, too, acknowledges
the shared origin of the statistical conception of information in his own and
Shannon’s work, amongst others:
This idea [of developing a statistical theory of the amount of information, in
which the unit amount of information was that transmitted as a single decision
between equally probable alternatives] occurred at about the same time to
Concepts: Information Entropy, Negentropy, Noise 17
several writers, among them the statistician R. A. Fisher, Dr. Shannon of the Bell
Telephone Laboratories, and the author. (Wiener 1961, 10–11)
However, Wiener and Shannon arrive at diametrically opposed ideas of what

information is, because Wiener defines information precisely as the opposite of
‘information entropy’, namely as the negation of entropy (which the physicist
Leon Brillouin later entrenches as the dominant technoscientific definition of
the concept of information, by inventing the neologism negentropy):
The notion of the amount of information attaches itself very naturally to a
classical notion in statistical mechanics: that of entropy. Just as the amount of
information in a system is a measure of its degree of organization, so the entropy
of a system is a measure of its degree of disorganization; and the one is simply
the negative of the other. (Wiener 1961, 10–11, emphasis added)
For Wiener information is precisely the reduction of freedom of choice, and

thus the reduction of uncertainty. Information, in Wiener’s cybernetic theory,
is a measure of increased constraint, associated with ideas of organization and
order, bound to decrease entropy. Here entropy is not a measure of information,
as with Shannon, but, on the contrary, a measure of its presumed opposite, i.e.
disorder or noise.
There is a startling matter-of-factness in the way both mathematicians
provide diametrically opposed definitions of information, without mentioning
their fundamental divergence. This may be indicative of the low priority that
discursive definitions have for the two mathematicians. The real emphasis
is instead on the mathematical innovation, which both share without
disagreement, even complimenting each other. The concepts of information

and noise are treated as theoretical tools that must be not only fit for purpose,
meaning the communication of mathematical theories to a broader public,
but also tailored to different needs, be it the transmission of a trans-Atlantic
telephone conversation, or a successful targeting of a self-directing missile, to
name just two of the most frequently cited examples of information theory and
cybernetics.
What this tacitly accepted dichotomy between ‘information entropy’ and
the ‘negation of entropy’ reveals is, first of all, that there is freedom of choice
in the discursive interpretation of mathematical formalization. This freedom of
choice is nothing other than the ambiguity of non-mathematical concepts. It
is thus not at the level of mathematical stringency itself that ambiguity arises
about the conceptualization of information and noise, but at the level of freedom
of choice in its discursive interpretation. This source of discursive ambiguity is
important to underline, since it is at the level of discourse rather than at the level
of mathematics, that the concepts of information and noise are translated into
other scientific domains – notably biology (Morange 2006), and from biology
to sociology and economy etc. – often via a tacit adoption also of the cybernetic
paradigm of self-regulating machines with feedback (Mersch 2013).
It is here, at this crossroad of conceptual circulation, that we must be most
attentive, because concepts reveal themselves to be more than just theoretical
tools: they are prisms through which we see and discover the world at the same
time as being the tools with which we transform the world. Their consequences
go well beyond mere functionality in a theoretical apparatus for this or that
technological or scientific purpose: concepts contribute to shape cultures and
precondition value judgements, while being in turn also imbued with cultural
preconditions and slanted by pre-existing value judgements.
Claude Shannon’s definition of information as ‘information entropy’ has the
singular merit of having prepared the ground for a philosophy of noise that
evades the Manichean opposition between information and noise, echoing that
between order and disorder, life and death. It also evades the mere relativism
according to which what we define as information or noise is a question of
individual perspective. To demonstrate the cultural relevance of this conceptual
feat, we will tackle the difficulty that arises when the concept of noise is no
longer applied only to the channel of communication, but also to other domains,
where the distinction between information and noise is not a given. In vivo,
rather than in the well specified and controlled situation of the channel of
communication, the distinction between information and noise is never ready-
made, but always presents itself as a vital decision or as an epistemological

problem.
At stake, hence, is the difference between information and noise in the
making, i.e. the moment when information must be selected prior to the
transmission of a message, necessarily implying a decision, an act of selection
whereby information can stand out from noise. Is not the challenge of every
form of research the problem as to how we can identify what counts as
information and what, in turn, can be discounted as noise? The dividing line
between information and noise is so fundamental to all forms of enquiry and
experimentation that the consequences of Shannon’s ‘entropic ideas’ vastly exceed
any technological framework, making the conceptualization of information and
noise philosophical problems in their own right.
Shannon’s ‘entropic ideas’ require us to rethink our most basic attitudes
concerning information and noise. Rather than opposing noise to information,
as the presence of entropy to its absence, he divides the presence of ‘information
entropy’ from the presence of ‘noise entropy’. The dividing line between
information and noise now runs within entropy, rather than between entropy
and its negation.
This is a subtle but fundamental shift that effectively challenges the
principle of the excluded middle, according to which a proposition is either
true, or its negation is true, and which implicitly underscored the analogy
of the information/noise opposition with that of sense/non-sense, and even
organization/chaos. A new division between desirable and spurious uncertainty
now competes with the classical opposition between truth and error or, as in the
excluded middle, between the truth of a proposition and its negation.
The philosophical consequences are profound, for the process of information
can now also be understood as a cut across the fabric of uncertainty. Information
becomes the process whereby this cut progressively gives rise to a form of
measurable uncertainty.
Shannon’s ‘entropic ideas’ thus have a profound philosophical and, more broadly,
cultural importance, if only we are willing to consider their conceptual relevance
beyond the technical realm. Common criticism instead holds that Shannon’s
concept of information applies only to electronic signal transmission, and is utterly
misleading in any other context. Complicit with this criticism is the equally common
position that opposes culture and technology. Endowing only ‘cultural’ artefacts
with signification, this view reduces all aspects of technology and even of science to
their mere utility. In its extreme form it represents a technophobia that pits culture,
and even nature, against science and technology in a relation of hostility.
The widespread cultural condescension towards the mere utility of the
sciences and technology corresponds in kind to the technocratic dismissal of

culture as mere recreation (or product of consumption). In fact, both attitudes
are but two sides of a coin. Both fail to recognize the cultural potential of
Shannon’s deconstruction of the traditional opposition between information
and noise and the revaluation of uncertainty that it entails.
Just as technocracy implies an aberration of cultural values, so technophobia
fails to rescue culture, because it is itself the symptom of a redundant, conservative
idea of culture. It is redundant because it wilfully ignores the fundamental
role that forays into mathematics and more broadly science and technology
have always played in the visual arts, literature and music. The fascination
with mathematics, science and technology has characterized the art of Greek
Antiquity no less than the metrics of Arabic poetry, it has fuelled the European
Renaissance five hundred years after driving Bagdad’s cultural prominence,
finally it has been an indelible aspect of twentieth-century art, literature and
music, and is even more so of the artistic practices of new millennium.
The redundant opposition between technology and culture atrophies not only
the quality of engagement between the arts, the sciences and technology, but in
turn also atrophies the status of creativity attributed to science and technology,
by denying it its cultural relevance beyond its utility. French philosopher Gilbert
Simondon was right to speak of an enslavement of technology and to see in it a
factor for mutual alienation in culture.
To place Shannon’s ‘entropic ideas’ within this cultural frame of debate
thus means overcoming the consensus that there is an opposition between
technology and culture. The first task is to work against this alienation, so
that we can recognize Shannon’s as a minimalist definition of information
and noise of the highest cultural relevance. It is minimalist insofar as it deals
with the conditions of possibility of information, precisely by bracketing out
signification: it separates out signification from both the means and the process
of transmission – thereby revealing the structural and procedural conditions of
information processes, much like minimalist art did with artistic expression in
an industrialized world. Taken outside the narrowly technical context of signal
transmission, we can begin to see that Shannon’s ‘entropic ideas’ also offer an
iconoclastic definition of information and noise, one that breaks the spell fusing
signification with the means and process of its transmission as if they were one.
Some of the most beautiful words regarding the reconciliation of culture with
science and technology have been written by Gilbert Simondon in On the Mode
of Existence of Technical Beings (Simondon, trans. Malaspina and Rogove 2017,
15–16):
Culture has constituted itself as a defense system against technics; yet this
defense presents itself as a defense of man, and presumes that technical objects
do not contain a human reality within them. […] The most powerful cause of
alienation in the contemporary world resides in this misunderstanding [caused]
by its absence from the world of significations, and its omission from the table of
values and concepts that make up culture.
Guiding us here is the ethos, rather than the method deployed by Simondon
in METO (On the Mode of Existence of Technical Objects), where he gives a
genetic account of the modalities of technicity across what he calls the evolution
of technical individuals, technical elements and technical ensembles. It is not
the objective here to construct a genetic analysis that would be in any way
comparable to what Simondon did for the concept of technicity. No comparable
historicizing claim will be made about the splitting of the idea of noise across
technics and religion, and across theory and ethics. Nor will there be an attempt
to determine the role of aesthetics in mediating such a split.
The objective here is more modest. It is to test two widely held presumptions
about noise, and to do so in a number of different contexts, so as to reveal their
intrinsic relatedness. The first is the implicit presumption that we can rely on
an intuitive notion of noise, in order to bridge its definitions across different
techno-scientific and cultural settings. The second presumption is that, rather
than intuition, it is a formal, i.e. mathematical, definition that presides over the
multiple uses of the concept of noise across the spectrum of scientific discourse.
What emerges instead are far from uniform conceptions of noise, some of which
profoundly counter intuitive. Although ubiquitous, both the idea of noise and
information reveal themselves to be conflicted, both displaying a fundamental
ambivalence towards novelty and change, as signaled by Shannon and Wiener’s
mathematically identical, yet discursively opposed definitions of information.
II
Entropy as ‘Freedom of Choice’
To equate information with unpredictability is intuitive enough, if it is to tell

us something new, something that does not follow automatically from what
came before. It is equally easy to accept that a message we can fully predict
is redundant, if it gives us no new information. What is much less intuitive
are the consequences Shannon and Weaver draw from this unpredictability.
By aligning the concept of information with uncertainty and by quantifying
it as such and without concessions, we arrive at the apparently paradoxical
conclusion that more information means more uncertainty. This appears
paradoxical, in the sense that it contradicts the equally common assumption,
even the doxa (opinion or dogma), that information is what reduces uncertainty,
rather than increasing it.
Quantifying information according to the degree of uncertainty it presents,
according to the ‘entropy of the message’, has therefore caused alarm (Janich
2006). Shannon appeared to have fundamentally misunderstood what we mean
by information. The fundamental role he gives to contingency in information

contradicts what we commonly associate with the purpose of information,
namely that information is reliable only if it reduces uncertainty and makes
experience less contingent. Shannon’s definition of information thereby appears
dangerously close to that of noise.
In his article ‘What Is Information?’ José Maria Diaz Nafria’s makes the
implications of this difference between the ordinary sense of information and
Shannon’s definition of information entropy abundantly clear. If the only criteria
for the quantity of information is its unpredictability, expressed in terms of
entropy, then a critical signal consisting of few bits would be discounted as low
in information, while the high entropy of irrelevant background noise would
measure the greatest quantity of information:
Just one binary digit may tell us if the universe is about to collapse, thus being
very informative, and all millions of terabits on the web could just as well
be generated by the whim of electrons in a rheostat, being thus completely

uninformative. (Diaz Nafria 2010)
It must be clear from this example that Shannon’s entropic ideas about
information are not a mere extension or deepening of the ordinary notions of
information and noise, but a challenge to the ordinary conception of information
of the highest order, since, as Diaz Nafria’s example makes plain to see, nothing
distinguishes outwardly ‘information entropy’ from what we would ordinarily
call noise.
While information is, as a matter of course, meant to tell us something new, the
logical consequence that this novelty decreases predictability and thus increases
uncertainty appears to be going one step too far. Shannon’s quantitative measure
of information has since been interpreted almost as a form of sacrilege against the
‘true’ understanding of information, which ought to increase certainty. It is also
discounted as incapable of telling us anything about what matters, which is not
quantity, but quality of information, and which is a prerogative of its signification.
We could say, on the other hand, that Shannon’s understanding of the relation
between information and contingency is indeed paradoxical, but not because
of the misplaced conceptual ambition. It is paradoxical in the sense that it is
free of cultural pre-conceptions and therefore offends such pre-conceptions,
transgressing their doxa: in this sense the conceptual innovation inherent in
Shannon’s concept of ‘information entropy’ indeed acts as a form of conceptual
noise, when it is exported from its technological application to other domains.
Let us be clear, Shannon’s definition of information as an ‘uncertainty relation’
does not contradict itself, but the doxa according to which one ought to obtain
from information simultaneously both novelty and a reduction in uncertainty.
Shannon’s definition of ‘information entropy’ instead frustrates this paradoxical
need (novelty and certainty) and thereby enables us to think about contingency
as belonging to the conditions of possibility of all processes of information,
including but not only of those processes we associate with signification in the
semantic communication between sapient beings.
What, then, is the relation between uncertainty and information, and hence
also between information and noise? The answer to this question is not as obvious
as it might at first seem and unfolding it may change the way we think about
both noise and information. It is this question that is posed, in mathematical
terms, by Shannon’s MTC. Shannon gives an engineer’s answer to this question,
which Warren Weaver translates for a broader readership, expressing it in the
following way in his 1964 introductory essay to the MTC:
[I]nformation is a measure of one’s freedom of choice […]. (Shannon and

Weaver 1964, 9)
This definition is intentionally shortened here in order to indicate the importance

we must attribute to it in the context of our enquiry into the conceptualizations
of noise. As a technical definition it is, first of all, abstracted from the habitual
association of information with signification. The common criticism of Shannon’s
concept of information is therefore that it is a purely quantitative measure that
is indifferent to the signification of a message, which consequently ignores the
inherently qualitative aspect of information, which is its signification.
The criticism levied against Shannon’s ‘entropic ideas’ is certainly important
and valid at the level of interpretation and evaluation of a message, but it has
also detracted attention away from two very subtle philosophical gestures that
Shannon’s approach implies: one concerning information as a process rather than
a given, the other concerning the role of contingency, and hence of uncertainty,
in this process. It is necessary to render these explicit, as the concept has de facto
been translated and applied to a great variety of disciplines in the natural and
human sciences, often as a language that facilitated profound methodological
upheavals, like the shift from classical to non-classical mechanics, from classical
physics to quantum physics, from classical biology to molecular biology and
biophysics.
Shannon’s definition of information as ‘entropy of the message’ offers us a
contribution to our understanding of contingency, which challenges both the
doxa according to which information is a correlate of certainty and the certainty
with which we can distinguish information from noise when the concept is taken
outside the technical paradigm of the channel of communication, for instance in
biology or economic theory, where the scenario of the engineer who transmits a
ready-made message as information no longer prevails. Shannon’s contribution,
which follows from the fundamental realignment of information and uncertainty,
is fundamental insofar as it enables us to place information and noise on an equal
footing, where both represent a measure of ‘entropy’ or unpredictability, prior to
the assignment of signification, purpose or representation; prior, in other words,
to the levels of decoding, interpretation and evaluation.
If we follow through with Shannon’s ‘entropic ideas’, our fundamental
assumptions about information must be rethought, taking contingency and
hence noise into account, not only as that which impinges on the fidelity of the
message, not only as that which obstructs the decoding and interpretation of
information, but as an uncertainty fundamental to the process of information
itself. This attention to contingency and uncertainty is what will enable us to

rethink the definition of noise, to take it outside the channel of communication,
in other words to think about noise in vivo, where the distinction between
information and noise is always a process in the making.
The emphasis is here on the conceptual consequences that must be drawn from
Shannon’s alignment of information with contingency, and more specifically
concerning the less Manichean, less oppositional relation between the concept
of information and that of noise. We can now ask, in light of the indelible aspect
of contingency in the information process, on what grounds can we draw the
line between ‘information entropy’ and noise? Meaning quite literally that the
distinction between information and noise is a problem of ground or foundation
of knowledge. If both noise and ‘information entropy’ are measures of ‘entropy’,
understood as in purely probabilistic terms as ‘freedom of choice’, then how can
we be sure which measure of choice informs and which exceeds and deforms?
If uncertainty increases with both information and noise then, in this greater
uncertainty, with what certainty do we draw the line between information and
noise?
III
Information Entropy and Physical Entropy
The conceptual operator upon which the idea of information uncertainty

hinges, is entropy. But how are we to understand the idea of entropy, when it is
no longer a concept bound by the theoretical and empirical constraints of the
field in which it arose as a key concept: thermodynamics? Shannon’s theory of
information is said to have emerged from his work on Boolean logic, applied to
electrical switches. It owes more, in fact, to Norbert Wiener’s use of the calculus
of probability in cybernetics, than to a direct engagement with Boltzmann’s
statistical theory of physical entropy (Atlan 1979). The resulting formalism,
however, is not just metaphorically, but formally analogous with the statistical
expression of physical entropy.
Let us see how Shannon’s designation of information as ‘information
entropy’ builds entropy as an indispensable metaphor into this ontologically
arbitrary concept of information. Shannon’s choice of expressing the
unpredictability of the message as ‘information entropy’ (H) reflects the fact
that its mathematical formulation in information theory, is indeed almost

identical with Ludwig Boltzmann’s statistical formulation of molecular
entropy in thermodynamics (S):
H = – Σ pilog pi,
S = – K Σ pilog pi
Both information [H] and the physical system [S] measure the number of possible
states, either as a message or as a physical entity. This probability, attached to the
number of possible states, is the sum of probabilities of the ‘presence’ [p1, p2, …,
pn] of ‘signs’ or particles [i], multiplied by a logarithm [log]. [H] is thus a measure
of the uncertainty over the occurrence of one amongst all possible events H(p1,
p2, …, pn). If all probabilities [pn] are equal, then the greatest possible ‘freedom
of choice’ corresponds to the greatest possible uncertainty regarding the actual

state of either system or message, with respect to all its possible states.
More simply put, in the state of maximal entropy the movement of particles
is determined by nothing but their random collision. The more random the
movement of particles is, the greater is the number of their possible positions,
speed and direction. Maximum entropy thus corresponds to the equal
probability of all possible states of the system, which is often illustrated with the
idea of the random collision of molecules in a canister of gas. All configurations
of free particles can thus be said to occur with equal probability. Conversely, the
probability of predicting the state or behaviour of the physical system, meaning
the exact configuration of its micro-complexions, including the positions, speed
and directions of its particles, is at its lowest.
Entropy and noise indeed become identifiable, and to this extent more
predictable, thanks to a better knowledge of the statistical framework which
encompasses them, which we owe notably to the important work of Boltzmann
and Shannon, no less than Wiener and many others. Nevertheless, it would be
wrong to overstate the statistical determination of noise and call it predictable,
because ultimately entropy and noise remain a measure of ‘freedom of choice’,
characteristic of an undetermined state of the system or message. The quantity
of information in Shannon’s sense is analogous to the probability with which the
observer of a physical system can predict what Max Planck called the micro-
complexions of a given system, and the probabilities of finding the system in any
of these complexions:
Maximal disorder corresponds to the greatest number of possible complexions

with equal probability for all […]. (Atlan 1979, 31, n.1)
Disorder here means that no external order imposes any form of constraint that
would compel particles to behave in one way rather than another. Both entropy
and ‘information entropy’ must thus be defined by sophisticated statistical
measures expressing the receiver’s uncertainty as to the determination of the
system, message or event. Increased quantity of information, in this sense of
‘information entropy’, is thus not the equivalent with increased certainty about
the system, even if certain forms of noise have identifiable and reproducible
characteristics in statistical terms.
IV
The Idea of ‘Potential Information’
To predict the probability with which signals occur in a message, Shannon

uses a mathematical expression that is almost identical to Boltzmann’s. What is
significant, however, is that he leaves out the term ‘k’. ‘k’ is the physical constant
that expresses the calorific value of flows of energy, understood as displacements
of thermal charges, wherever a disparity exists between energy levels, for instance
in electrical currents or in flows of matter. This algorithm, ‘k’, is what anchors
Boltzmann’s formula in physical reality.
Physical potential arises from unequal energy levels that compel a process of
equalization. Potential therefore expresses a form of constraint on the system,
which is forced by the disparity of energy levels to evolve in such a way as to
equalize this difference. Potential thus increases the probability of a system to
evolve in the direction of this equalization of energy levels, and its way of doing
so will depend on the interaction between its constituent elements. Potential
thus effectively reduces the number of possible states of a physical system:
compared to the state of maximum entropy where all possible states occur with
equal probability (for instance, the molecules randomly bouncing off each other
in a canister of gas), potential obliges the system to actualize an equalization of
energy levels.
Now, if potential reduces ‘freedom of choice’ by compelling a process of
equalization of energy levels, then the greater the disparity of energy levels – the
greater the potential – the more powerfully the system is entrained to evolve
in a particular way, as for instance in the flow of an electric charge. Even if a
margin of indeterminacy persists as noise, potential is what reduces the number
of possible states of the system, forcing it to evolve according to its constraints.
To transpose the idea of potential information to Shannon’s ‘entropic ideas’ thus
runs us into difficulties, if we want to preserve the idea of ‘freedom of choice’.
To understand potential as a form of constraint no doubt offends common
sense. It forces us to pause before the usual idea of potential as synonymous
with opportunity in the sense of ‘freedom of choice’. When we say that an

event has the potential to occur, we usually mean that it may or may not occur:
potential thus represents an added option and hence greater ‘freedom of choice’,
as when one says ‘he has the potential to become a great pianist’. When we say
a child has potential, what is meant is that it has the opportunity to develop
certain capabilities. On the contrary, a child seen as lacking in such potential
is presumed to have fewer opportunities and thus a lesser ‘freedom of choice’.
The child’s perceived aptitude, however, is better understood as an optimum fit
with given requirements. The potential to become a great pianist, for instance,
compels a child onto a certain path of development. Because potential is valued
as increased ‘freedom of choice’, we ignore the constraints that turn perceived
potential into a constraint: the child with perceived potential is compelled to
succeed.
But what is success, if not the growing constraint to proceed from one level
of achievement to the next? It becomes an almost mechanical sequence of
expectation and compliance – which ultimately feeds into the social reproduction
of relations of power. If to succeed, more often than not, increasingly narrows the
noose of the expectation not to fail, then is the ‘freedom of choice’ we attribute
to potential not ultimately greater for the one who is not pressed into the mould
of expectations, corresponding to a perceived potential, or who deliberately
confounds these expectations? Art critic Martin Herbert, for instance, in his
collection of essays Tell Them I Said No, looks at the work and lives of ten artists
who, in different and sometimes extreme ways, refused to play the game of
celebrity that enchains artists to an ‘overly educated’ and ultimately conservative
audience (Herbert 2016; Judah 2017).

In a market driven society it is undoubtedly heretical to question the link
between potential and ‘freedom of choice’. It is only when we turn the idea of
potential around and express a negative potential that the compelling nature
of potential as constraint becomes more evident, and we can begin to think
about potential as reduction of ‘freedom of choice’. If we say, for instance, that a
certain group of underprivileged children will potentially fail to thrive in society,
according to available sociometric parameters, then the idea of ‘potential’ reveals
its negative characteristic of constraint more readily. In both cases, however,
expectations are present, be they valued positively or negatively, and potential
can be said to reduce ‘freedom of choice’ insofar as it represents criteria,
norms and structural conditions that compel a child, or any other observable
phenomenon with potential, to evolve according to its greatest probability. What
needs to be retained here is simply that the physical concept of potential, with its
noble Aristotelian heritage, is by far not an anodyne synonym for the possible,
when it comes to understanding the relation between probability and ‘freedom
of choice’.
It is thus important to stay alert when considering whether ‘information
entropy’ must be understood as ‘potential information’. Physical potential
implies that an event is more likely to occur, thus in fact reducing the number
of possible events, as when one switches on a light circuit and the electricity is
compelled by the physical potential to rush through the wire. Both potential
and freedom of choice are manners of speaking about a possible event, yet the
difference of inflection between potential, perceived as an option, as greater
‘freedom of choice’ and potential as greater probability of occurrence, hence
reduction of choice, is not without consequence.
In the state of maximal entropy, on the contrary, initial differences in
energetic potential have equalized through interaction, until the system as a
whole finally reaches a state of energetic equilibrium, where flows of matter
or energy from one part of the system to another are highly improbable, at
best random effects, because the micro-constituents of the system are no
longer exposed to the tension of discrepancies between energy levels, no
longer compelled by the physical potential that arises from these differences.
Consequently, each state of the entropic system occurs with equal probability
or, differently put, with the greatest ‘freedom of choice’. Coming back to
Shannon’s formal mathematical definition of information, this means that to
define ‘information entropy’ as potential information is to inverse it completely:
potential, strictly speaking, would be a negentropic factor, negating entropy.

In Boltzmann’s definition the algorithm [k] serves to indicate flows and
displacements of thermal charges. Yet by leaving out the reference to this
physical aspect, Shannon transforms Boltzmann’s mathematical expression
of entropy into an ontologically arbitrary measure of probability. Shannon
thereby unmoors probability from Boltzmann’s empirical measure of calorific
conversion of energy and work related to thermal displacements in a physical
system. Although Shannon himself applies this formula to the problem of
electronic signal transmission, his concept of ‘information entropy’ and hence
also of noise is now devoid of any ontological reference: it could inform us about
the probability of occurrence of any phenomenon involving large numbers,
be it the flow of signals, flows of people, of goods or unicorns – in short it is
ontologically arbitrary.
The abstraction of Shannon’s quantitative measure of information is

undoubtedly what facilitated its translation with great ease into every imaginable
field of research involving mass phenomena. It lends itself to statistical analysis,
not only in communication technology, but also in economical or biological
systems or any other domain. This gives Shannon’s concept of information an
eminently analogical, if not paradigmatic function. Shannon’s definition is lower
in theoretical constraints than Boltzmann’s formula, giving it greater polyvalence,
but by the same token also increasing ambiguity of its interpretation: with
respect to Boltzmann’s definition of entropy, Shannon’s concept of ‘information
entropy’ is thus itself a prime example of what ‘information entropy’ does, when
it increases the number of possible interpretations, namely increasing also
uncertainty.
The analogy between information and entropy, nevertheless, remains
paradigmatic in the denomination of information as ‘information entropy’.
The application of the term information to communication technology also
reinforces this analogy with physical processes, because the engineer must deal
with the effect of physical entropy in order to ensure the message is sent without
loss due to perturbations, such as thermal noise, during the transmission of an
acoustic or electric signal. The physical analogy thus persists, but as we have
seen on a metaphorical rather than formal, mathematical level (Morange 2006).
As a result the notion of physical potential remains, despite the obliteration in
Shannon’s concept of the physical referent [k], a key feature of the concept of
information.
What persists is also a margin of ambiguity, when one speaks of ‘information
entropy’ as the information one lacks, or as ‘potential information’, as the German

physicist and philosopher Carl Friedrich von Weizsäcker does, using the term
entropy here with specific reference to Shannon’s ‘information entropy’:
Positive entropy is potential (or virtual) information. (Weizsäcker 1994, 167)
However, the idea of possibility which is implied in both physical ‘potential’

and the ‘virutal’, risks blurring a distinction that is perfectly clear to the physicist
and much less clear in ordinary language. What von Weizsäcker means is that
entropy, which denotes the number of possible states of a system, corresponds
to virtual information and he specifies this in the parenthesis. Von Weizsäcker
thus speaks of ‘potential’ information in order to make the idea of ‘information
entropy’ more accessible. However, the virtual refers to the number of possible
states, which increases with ‘information entropy’, while physical potential, as
we have seen places a constraint on the physical system and thus decreases the
number of possible states by making one event more likely than another.
Where the notion of information ‘potential’ is introduced, it is thus in fact re-
introduced as an extrinsic criterion for the evaluation of ‘information entropy’,
more specifically of its hoped-for use-value as information in the traditional
sense of certainty and constraint. Better put, the idea of potential information
introduces the idea of the capacity of ‘information’ to perform work, to make
sense, which in turn is specific to the recipient of this information and the use
s/he can make of it. What remains ambiguous and unspoken is the necessary
conversion between the uncertainty that ‘information entropy’ introduces as
‘freedom of choice’, as under-determination, and the implied sense of potential
information leading to negentropy, i.e. of increased certainty and constraint.
Implied is that the actualization of potential information is equivalent with
this conversion of uncertainty into certainty. And nothing could be further from
certain than the spontaneous consolidation of uncertainty into certainty. For
what this requires, is also that the nature of the boundary between information
and noise changes, from being a limit that curtails the uncertainty of the ‘entropy
of the message’ vis-à-vis the unlimited uncertainty of noise, to a border that
opposes information and noise as certainty and uncertainty.
There is thus continuous ambiguity at the level of conceptualization when
the notion of physical entropy is transformed into the pure probability of
‘information entropy’, at once untethered from the physical paradigm, yet
indelibly tied to it through metaphor and philosophical tradition. It is this
ambiguity that constitutes ‘epistemological noise’ when Shannon’s concepts of
information and noise are exported, alongside negentropy and often without
distinction, to other domains, like biology, sociology and economics, where the
physical paradigm risks becoming prematurely reductive.
V
Physical Concepts of Information and

Informational Concepts of Physics
The controversial aspect of Shannon’s definition of information is that it is

one of randomness and unpredictability, for which Shannon uses not only the
mathematical expression but also the term entropy. Shannon in fact continued
the line of enquiry into the mathematical treatment of signal transmission begun
by H. Nyquist and R. V. L. Hartley at the Bell laboratories. He openly declared
his debt to Norbert Wiener’s work, who in turn pointed out the innovation that
Shannon’s ‘entropic ideas’ represented for information theory (Shannon and
Weaver 1964, 3). As Weaver points out in his introduction to MTC, the notion
of entropy was already associated with the notion of information in physics,
notably in the work of L. Szilard, and had proven useful in quantum mechanics
and particle physics, notably in the work of von Neumann (Neumann 1932,
chap. V). Yet albeit being fundamental to the physical sciences and engineering,
it is easy to see how the notion of entropy becomes a source of confusion when
introduced into common language.

There is of course no obvious reason why the concept of information should
be introduced in this way into our understanding of physical processes, or why
concepts from physics should pertain to our understanding of information1. The
biophysicist Henri Atlan already takes this controversy into account, referring
to Leon Brillouin’s 1959 Science and Information Theory, but also to the more
general problem of the use of intuitive concepts in physics, such as ‘energy’,
‘force’ and ‘speed’, which has been discussed notably by Cornelius Castoriadis,
G. Hirsch and J.-M. Levy-Leblond (Balibar, Lehoucq, and Lévy-Leblond 2005;
Brillouin 2013; Hirsch 1976; Lévy-Leblond 1976).
Outside of the purely scientific engagement with the notions of entropy and
information, Shannon’s entropic definition of information also provoked and
still provokes controversies as being excessively technical and alienating, if not
contradictory (Capurro and Hjorland 2003). The term ‘information entropy’ evokes
the paradoxical notion that information is reduced to disorder, if not chaos or, on the
contrary, that entropy corresponds to the idea of homogeneity and un-differentiation,
which is the opposite of what one would normally associate with the idea of a signal
or message that stands out against the indifference of background noise.
The mathematical formalization Shannon uses is, as we have seen, almost
identical to the way in which Ludwig Boltzmann first formalized the statistical
measure of entropy in a physical system, as expressing the average of all its possible
microphysical configurations, occurring with equal probability under specified
constraints. It is understandable that this notion of ‘information entropy’ is
incompatible with what one ordinarily calls information, if ‘information entropy’
evokes simultaneously the ideas of disorder and of homogeneity, and which to
boot becomes a measure of the information we lack:
Dr. Shannon’s work roots back, as von Neumann has pointed out, to
Boltzmann’s observation, in some of his work on statistical physics (1894),
that entropy is related to ‘missing information’, inasmuch as it is related to the
number of alternatives which remain possible to a physical system after all
the macroscopically observable information concerning it has been recorded.
(Shannon and Weaver 1964, 3, n. 1)
Yet how can the quantity of ‘information we lack’ correspond to the ‘quantity
of information’ we receive? The natural answer to is to say that it is precisely the
opposite that is the case, that information is the opposite of the ‘information
entropy’, namely its negation, and to explain this with the minus sign that
precedes the symbol Σ in the equation used to measure ‘information entropy’:
H = – Σ pilog pi
This definition of information as negation of entropy has become core to

the now dominant definition of information as negentropy. But to accept this
neologism, without enquiring into the theoretical conversion it implies, risks
discarding too hastily the philosophical potential of Shannon’s ‘entropic ideas’,
thereby risking to transform the concept of information into one of redundancy.
To negate entropy, is to negate all possible alternatives, and hence to affirm an
identity that cannot change.
Another way of solving this dilemma is to say that the more improbable
the occurrence of a particular sign is a priori, the more informative it is
a posteriori (conversely, if its occurrence was certain a priori, then it would
bring no new information a posteriori) (Atlan 1979, 33). While this appears
to be a good compromise, it leaves us with the abyssal question: how do we
turn the unexpected, and hence that which we could not anticipate or know a
priori, into something we know a posteriori? The difference between a priori
and a posteriori is a little more complicated than a mere before and after the
fact, if we accept that these terms have been irreversibly conditioned by Kant’s
critical philosophy. The idea that we can turn the a priori unknown into what
is known a posteriori implies an epistemological conversion that isn’t entirely
straightforward.
Let us recapitulate the idea by which the paradox of Shannon’s information
entropy could be brought back into the fold: information entropy is what is
unknown a priori, but known a posteriori and, crucially, the more unknown it is
a priori (i.e. the more unexpected it is), the more knowledge it procures (i.e. the
more it informs us in the traditional sense of the word information) a posteriori.
Now, if the a priori is a critical term that designates the conditions of
possibility of cognition, i.e. the concepts without which there is no coherent
unified experience, and the a posteriori designates that which is experienced
on the basis of these concepts, then I am not sure what such a conversion of a
priori uncertainty into a posteriori certainty could mean. It could mean making
the absence or indistinctness of concepts (or our uncertainty about the a priori
conditions of thought) the prerequisite for our certainty about experience. In
other words, it would mean a rejection of Kant’s critical legacy, and a return to
dogmatic intuitionism, where experience, if not irrationalism, supplants reason.
This, it appears to me, is not a solution to the paradoxical relation of information
and uncertainty, but an even greater paradox.

If, on the other hand, the terms a priori and a posteriori here do not refer
to critical philosophical concepts, but are simply used as erudite terms for
‘before’ and ‘after’ the fact, then we are still faced with a difficult epistemological
conversion, namely of the virtual (the purely possible) into the actual. In this
case, the more uncertain we are about the virtual possibilities inherent in a
situation, in other words, the more unexpected the evolution of this situation
is, the more knowledge this transformation will have imparted on us once it has
occurred. This inverse relation of the virtual and the actual may indeed provide
a fruitful conceptual framework for thinking about information. However, what
is certain, is that such a way of thinking about information is revolutionary by
default, if by revolution we mean the radical and unexpected transformation
of a situation, (and not a sudden reversal understood as a return to something
pre-existing). Such an approach, whereby the maximal value of information is
the most revolutionary, implies an epistemological attitude to information that

could not be further from the idea of negentropy, if the latter is understood as
the negation of alternatives.
VI
Information as Process Rather Than Content
The conversion of uncertainty into certainty is implicit and therefore taken

for granted, when ‘information entropy’ is defined as potential information,
or as information we ‘lack’. This conversion, however, cannot be the same as
a simple actualization of a potential, neither in the metaphysical, nor in the
physical sense. The problem of uncertainty and ‘freedom of choice’ and its
transformation into the opposite, into information as certainty and constraint
takes as its starting assumption what still needs to be explained, namely how
knowledge constitutes itself in the face of contingency and what role uncertainty
plays in the constitution of knowledge.
What is taken for granted is thus the fundamentally dynamical problem of
information at the heart of epistemology: information can only be understood
as a process rather than a given, a factum or a datum. Intuitively the idea of
quantifying information as ‘bits’ suggests a simple encounter of form and
content, as if one could quantify a certain amount of information as when one
measures how full a cup is. This risks obscuring one of the most important
aspects of Shannon’s entropic concept of information, which quantifies not the
individual signal or message, but its relation of probability with the set of all
possible messages given particular constraints – such as for instance a string of
letters in relation to a finite number of possible letters in an alphabet: a message
can be composed of a selection of discrete symbols, which could be letters,
words, musical tones or any imaginable other signal, each however belonging to
a set of symbols or a spectrum within which there is a certain ‘freedom of choice’
in terms of probability.
Each choice furthermore stands not on its own, but always in relation to
previous choices having already occurred in a discrete or continuous transmission
of information. The previous state is thus factored into the probability with which
the next symbol is chosen as the most likely, in what is called the Markoff process.
It is this progressive relation of probability, which turns out, as Weaver says, ‘to
be exactly that which is known in thermodynamics as entropy’ (Shannon and

Weaver 1964, 12). ‘Information Entropy’ is thus a measure of the probabilities
involved in progressing through stages of selection, indicating the probabilities
with which, at each stage, certain symbols will be chosen next. It is thus never,
the individual message that is carrier of information, but its relation with the set
of all possible messages under equivalent constraints, a relation that changes as
the transmission progresses:
The concept of information applies not to the individual messages (as the
concept of meaning would), but rather to the situation as a whole, the unit
information indicating that in this situation one has an amount of freedom of
choice, in selecting a message, which it is convenient to regard as a standard or
unit amount. (Shannon and Weaver 1964, 9)
When Weaver says the ‘unit of information is called a “bit”’, what is thereby
quantified is not a signal or message, but a changing relation between the actual
and the possible, within a given frame of constraints. To predict the actual
symbol or even message on the basis of the set of all possible symbols or even
messages, is to anticipate the relation between a set of n independent symbols
and the probability of choice p1, p2, …, pn. It is this relation of probability that
finds mathematical expression in Shannon’s formula:
H = – Σ pilog pi
The calculus of probability therefore measures how rich in entropy information

is, in terms of the progressive relation between our ‘freedom of choice’ and
our capacity to predict. This means that the quantity of information is never
measured as content or amount of the transmitted message alone, but as a
function of the relation between this message and all possible messages with
equivalent constraints. Information is thus understood as a dynamical relation
of probability that measures a process rather than a content. It is this progressive
sequence of probability between the actual and the possible that becomes the
raw material of communication, quantified in terms of ‘freedom of choice’ prior
to any possible interpretation and evaluation of the message as being significant
or not within a semantic context.
The habitual sense, in which information is considered like a vessel, a carrier
of a certain amount of signification, is thus transformed by Shannon into a
measure of the relation between the set of all possibilities, allowing a certain
‘freedom of choice’ in terms of probability, and the probability of prediction
based on already actualized choices:
The significant aspect is that the actual message is one selected from a set
of possible messages. If the number of messages in the set is finite then this
number […] can be regarded as a measure of the information produced when
one message is chosen from the set, all choices being equally likely. (Shannon
and Weaver 1964, 31)
For one, this implies that information is never a given, because it characterizes
a progressive modulation of certainty and uncertainty. Information presupposes
as essential the structural and operational synergy between context and
individual message, as between the uncertainty of ‘freedom of choice’, and the
progressive modulation of certainty during the evolution of the individual
message. Information, then, is the progressive unfolding of this relation between
uncertainty and certainty.
VII
To Think about Information as a Process

of Individuation
The philosophical significance of Shannon’s shift of emphasis, from individual

signal to process, can be appreciated if we look at it through the lens of a medieval
problem that was once known as the problem of individuation.
The idea of individuation must ring unfamiliar to the contemporary ear,
evoking at best a vague impression of scholastic disputes, perhaps echoing faintly
around the names of Aquinas, Scotus or Ockham. Readers of Leibnitz or Wolff
will recognize its gradually effacing traces in early modern philosophy. However,
with Descartes and the modern empiricists, the problem of individuation
appears to slide into oblivion, and what we are left with, qualifying the object
of experience, is the idea of ‘All Things, that exist, being Particulars … ’ (Locke
1975, 409 in Barber and Gracia 1994, 2). The individual has since become the
starting point of critical reflection and even the axiom of any possible rationality,
be it as the cogito or as the touchstone of empirical investigation. By the same
token it has become the stronghold of what we now call information according
to common sense. The individual, as object of experience, is what informs us
on ourselves and on the world (constituting either a bundle of faculties or an
aggregate of attributes).
Even though science has proceeded to dissolve individuality all the way
down to quantum fields, and has dismantled any residual faith in its intuitive
givenness through the neurocognitive sciences, the individual has nevertheless
ossified into a tenacious idea of personhood. The concept of the individual has
congealed into a political and moral sine qua non. We like to flatter it, when we
qualify the individual as a subject, paying no mind to the pejorative connotation
of subjection, which implies that we are subject to other powers, and that we
thereby glorify what the Ancients considered a passive substrate to an active
principle. A narcissistic investment in the idea of the individual thus makes it
difficult to render this notion unfamiliar once more, or even to recognize the
impact it has on our way of thinking about the world, and about what can inform
us and how.
To question the legitimacy of the individual’s primordial role in the
contemporary Zeitgeist is perhaps even threatening to some, as it touches the
centre piece of contemporary humanism: the individual and its identitarian
reclamations. Yet, what is left of the idea of humanity appears to be a fragmented,
hedonistic individualism. It has the merit of keeping the economy alive with its
voracious need to accessorize individuality and to soothe its fear of dissolution
with consumption. However, by the same token, the idolatry of the individual
also heralds the potential demise of humanity. Biologist Eugene F. Stoermer and
atmospheric chemist Jozef Crutzen even proposed to call our current geological
era the Anthropocene, indicating that the presence of humans on earth now has
the power to catalyse a process of such magnitude that the planetary survival of
all forms of life is put in doubt. (Crutzen 2002).
If the question of individuation appeared to belong to the Middle Ages, it
may yet acquire a new urgency in light of the consequences of today’s unbridled
individualism. It is the singular merit of Gilbert Simondon (1924–1989), to
have put the ossified concept of individuality back into motion, by reviving
the question of individuation. An atom, a biological cell or, indeed, a person
is no longer considered a given, either in the form of a monadic entity or as
an always already constituted whole. Instead, whichever entity or term we call
‘individual’ is seen as the end product of a process of individuation, whose most
final stage of individualization is but the exhaustion of its potential for further
individuation.
Simondon thus proceeds to put the metaphysician back on the school

bench. He takes us on a theoretical tour de force through modern physics,
biology and psychosocial theory – without neglecting a critical appraisal of
information theory and cybernetics, within a wider philosophical analysis of
technical reality.
He postulates that any theoretical problem or even existential crisis can be
said to reflect a field of tension such that, in analogy with an electromagnetic
field, whatever discovery, concept or idea is inserted as a new element is at
once seized by this field and polarized. The process of individuation is thus
compelled by the potential that characterizes this field of tension. This impetus
is comparable, yet subtly different from Aristotelian entelechy, because form
no longer predetermines the outcome, but enters in a recurrent causality
of form and matter. In an ambitious synthesis of Plato’s concept of form and
Aristotle’s hylomorphism, which he updates with the scientific concept of the
‘field’, Simondon revives the formal power of the idea, but embeds it in a revised
hylomorphic schema: the structuring or organizing principle of form enters a
reciprocal, mutually determining relation with a field of tension.
More simply put: what Plato considers a superior reality, the idea or form,
is no longer aloof of matter, but seized in a hylomorphic relation. In turn also
Aristotle’s hylomorphism is reformed. Form and matter are no longer abstractly
linked, as active and passive principles. The field of tension that receives
structuration, comparable to Aristotle’s matter, is itself active: it polarizes
and affects the idea or form, as much as it is structured by it. There is thus an
embeddedness of the formal power of ideas and concepts in an empirical field.
The idea becomes constitutive of this field and its process of transformation, but
is also polarized by it.
Simondon’s account of the process of individuation thus not only comprises
the emergence of form, i.e. the gradual or sudden structuration of a domain,
but implies also the concurrent transformation of the field itself, whose pre-
individual state gives rise to a milieu associated with the process of individuation.
Individuation co-evolves with its own milieu. Both the final individual and its
associated milieu are thus seen as by-products of a same process of differentiation.
Rather than being the first object of consideration, a given, the individual is thus
what comes last. Conversely, the milieu is not what precedes individuation – in
other words, it is not simply that to which the emerging individual adapts – but
is itself a correlate of individuation (Simondon 2005a).
Crucially, what qualifies the genesis of form, for Simondon, is information.
Information is not an aspect of the individual alone, such that one could compare
the content of a piece of information to the complexity of an individual entity

(or signal). Rather, information is whatever catalyses a process of differentiation,
comparable to the effect that a crystalline germ has on an oversaturated solution,
but information also qualifies whatever modulates this process, amplifying or
regulating it. In other words, information is both what triggers a process of
differentiation and what acts as an organizing principle. Whatever catalyses
and modulates the process of differentiation, as a resolution of tensions or the
solution of a problem, can thus be qualified as information.
Information is thus, for Simondon, what links the epistemological and
the empirical aspects of individuation. It is the idea or concept (or form) that
catalyzes a process of differentiation, that modulates this process and that,
hence, informs empirical reality, which in turn polarizes the idea. However, the
coupling of reason and experience is also subject to a formal analogy between
thought processes and empirical processes, whereby
no norm, no system detached from its content can be defined: the individuation
of thought alone can, by accomplishing itself, accompany the individuation of
beings other than thought; it is thus not an immediate nor a mediate knowledge
that we can have of individuation, but only a knowledge that is an operation
parallel to the known operation; we cannot, in the habitual sense of the term,
know individuation; we can only individuate, individuate ourselves, and
individuate within ourselves; […] an analogy between two operations, which
is a certain mode of communication. (ILFI, 36. Emphasis in the original; my
translation)
Information thus becomes a key concept for individuation. The new

inflections that the concept of information receives by way of information
theory and cybernetics are present throughout Simondon’s two main works:
his major doctoral theses entitled L’individuation à la lumière des notions
de forme et d’information (published as a whole only in 2005 by Éditions
Jérôme Millon (Paris) and as yet untranslated),2 and his secondary thesis,
written in accordance with academic requirements at the time, entitled
Du mode d’existence des objets techniques, which was published as early as
1958 in by Aubier, Flammarion, and whose official English translation has
been published by Univocal Press only in 2017 (Simondon, Malaspina and
Rogove 2017).
It is in this key role given to a processual understanding of information that
we find an affinity with Shannon’s mathematical approach to information as
an evolving relation of probability. The philosophical magnitude of Shannon’s
processual understanding of information can be gleaned by comparing it to
Simondon’s metaphysics and epistemology of individuation: there is no ‘piece’

of information whose quantity could be determined in and of itself, any more
than there are individuals that can be abstracted from a process of individuation
without rendering them sterile and lifeless. The point that can be made here,
without imposing too violent a reading on Simondon’s work, is that information
is fundamentally misunderstood, if it is taken to characterize an entity (a piece
or information, a message in and of itself) rather than a process, from which the
transformation of context cannot be dissociated.
Shannon and Simondon both operate a Copernican revolution, replacing the
individual (being or message), traditionally at the centre of the attention, with
a careful attention to the co-evolution of both individuation and its context or
milieu.
However, Simondon’s own critique of the insufficiency of any quantitative
concept of information must be born in mind. It would be wrong to
suggest that Simondon’s theory of individuation lends itself to an appraisal

of Shannon’s concepts of ‘information entropy’, and especially to the
philosophical revaluation of the concept of noise, such as it is argued for here.
Yet without seeking to harmonize the difference between Simondon’s concept
of quality of information and Shannon’s ‘information entropy’, we can still
find resources in Simondon’s theory of individuation which enable us to shed
new light on Shannon’s conception of information as a progressive relation
of probability.
For one, Simondon enables us to grasp the philosophical enormity of a
concept of information that puts the individual (message) last, and brings to the
foreground a progressive relation of probability between the individual state of
the message and the set of all possible messages under a given set of constraints.
To think with Simondon, without thinking exclusively in Simondonian terms,
thus helps to critically redress our understanding of information, and notably
to reject the common conflation of information with ‘data’, understood as
something given.
However, Shannon’s definition of both information and noise as entropy,
distinguishing only desirable from spurious uncertainty, no doubt strains
the analogy with Simondon’s theory of individuation. For Simondon the role
of information remains essentially one of differentiation and structuration. It
is ultimately based on an opposition to entropy, understood, as Wiener and
Brillouin do, as the final state of equilibrium or the death reached by a closed
system. Although, technically speaking, Wiener’s concept of negentropy is based
on the same mathematical model as Shannon’s, imposing an understanding
of information as process rather than entity, in cybernetics it is always in the

service of an already constituted and correctly functioning entity, a machine
or an organism, that information is required to counteract entropy, notably
through feedback processes.
This is not to say that Simondon settles for Wiener’s definition of information
as negation of entropy, any more than Shannon’s. Both fall short of providing a
qualitative definition of information, such that it could encompass individuation.
What Simondon finds lacking in the quantitative definition of information
is the notion of potential, the tension that polarizes and thus gives a sense, if
not a signification to information. Simondon rejects the idea that information,
measured in bits, could in any way encompass what we must understand by
the quality of information, which is what characterizes not only the capacity to
inform or regulate reality, but, crucially, its capacity ‘to illuminate new domains’
(Simondon 2005, 549).
Simondon does, however, incorporate a great number of aspects into his

theory of individuation, which would appear to lend themselves to a revaluation
of various aspects of noise, such as we are aiming at here, by relativizing its
opposition to information. He has a subtle understanding of demodulations
of structure, even of crisis, which he sees as a necessary reload of potential for
novel structuration and reorganization. He mentions, for instance, processes
of dis-adaptation in developmental psychology, as necessarily preceding
reorganization at a higher level (Simondon 2005, 545). Several key terms
in his philosophy express a deep appreciation for indeterminacy or even
dissolution of structure or form. In his On the Mode of Existence of Technical
Objects he values margins of indeterminacy in the functioning of so-called
open machines, as necessary for their capacity to respond to input from the
environment. In his lecture Communication et Information Simondon even
delves into the motivating or bonding aspect of social noise (Simondon,
Simondon and Chateau 2010; Simondon 2010).
Furthermore, Simondon’s notion of a pre-individual state lends itself
to an analogy with noise if we think of the pre-individual state of being as
characterized by the equal probability of all possible states, and thus rich in
the greatest possible ‘freedom of choice’. As such a primordial noise could now
be thought, in light of Simondon’s concept of the pre-individual, as a positive
ground for differentiation, in other words, as ground for the emergence of
form and its transformation. This would mean that a primordial maximal
uncertainty not only grounds but co-evolves with the process of individuation.
Noise in the ordinary sense, understood as interference, would thus be the
remaining uncertainty, tethered to the process of individuation by its associated

milieu.
What Simondon calls the ‘pre-individual’ state of being is thus not entirely
alien to our idea of a subtler difference between ‘information entropy’ and noise
entropy, both understood as a margin of ‘freedom of choice’ or uncertainty.
However, even this limited analogy requires an important proviso. The analogy
works only if both concepts, the ‘pre-individual’ and ‘maximum entropy’, are
treated as regulative ideas, acknowledging that no situation is absolutely closed,
in the sense that the absolute absolves from any relation to an outside. There
are approximations of absolute closure of a system (for instance, in
nanotechnology or quantum technology) that come at the cost of a technological
apparatus and of mathematical constructions of great sophistication. However,
no process of individuation and no process of information ever takes place
on absolutely blank slate: nowhere in the empirical world is a closed system
realized in absolute terms, i.e. without having to take into account the role of the
observer or the impermanence of this closure. Every system known to mankind
is always already situated in a reality that is densely packed with pre-existing
processes of individuation. To acknowledge this is to admit that empirical reality
is always already a noisy mess of competing processes of individuation, involving
dissolution of form and a wrangle of certainty and uncertainty.
However, even if we acknowledge that no process of individuation is absolute
and that multiple processes of individuation may compete noisily, it must also
be acknowledged that Simondon explicitly stops short of an affirmation of noise,
(or of phenomena involving de-differenciation, indeterminacy or even chance),
as constitutive of information processes.
Simondon compares for instance the tension of form, which he sees as a
precondition for the quality of information, to social phenomena such as
pre-revolutionary tensions. It is conceivable that in such situations, he says, a
‘thought coming from elsewhere’ (le fait qu’une idée tombe d’ailleurs) triggers a
sudden structuration (ILFI, p. 550). Just as a chance correlation of molecules
may set off the process of crystallization, so a ‘chance encounter’ may set of a
revolutionary process. However, and this is crucial, ‘it is very difficult to admit
that chance has a value of creation of good form’ (ibid.). This is because the
quality of information is more than a fortuitous aggregation: its structuring
effect must be more than just fleeting, it must sustain a structuring power,
and sustain what in French is called sens and which we can only partially
translate as both signification or direction. The quality of information mediates
information’s power of structuration and the tension that characterizes a
domain capable of receiving information. This mediation is quite literally the

sense that information makes, the signification or organizing power it catalyses.
A purely fortuitous process, in turn, would be subject to an equally fortuitous
dissolution.
In other words, what we could call a noise phenomenon, and which Simondon
here characterizes as coincidence or chance (hazard), may act as a trigger for
spontaneous structuration but – and this is the crux – the process of information
itself remains a process of structuration for Simondon and ultimately a negation
of entropy.
While this appears to cut short any analogy between Shannon’s conceptual
audacity and Simondon’s return to the problem of individuation, it must be born
in mind that Simondon is already several steps ahead of the problem we are
addressing here. His emphasis on the quality of information is already a problem
of signification. For us, on the other hand, what is at stake is a rather limited
problem that does not yet encompass the question of signification, but only the
presence of uncertainty among its conditions of possibility.
What Shannon enables us to think is not an absolute value of noise as novelty –
which one could provocatively call ‘pure information’, if one were to attribute a
maximal information value to maximal entropy. It is, rather, the fact that we can
now think of information as a subtler difference than that between organization
and chaos or sense and non-sense, a difference that takes place within the
conceptual space of entropy, within the space of uncertainty: if information can
be thought as qualified uncertainty, then noise too can be released from the
theoretical exile of negation into which it was thrown. Noise can become possible
information. In other words, unqualified uncertainty can be understood as one
of the preconditions of qualified uncertainty and, hence, of information.
VIII
Redundancy and Necessity
Information entropy, understood as ‘freedom of choice’ or as equi-probability

of events, is thus never absolute, especially when we consider that this term
qualifies the quantity of information not merely as ‘entropy’, but as ‘entropy
of the message’. ‘Entropy of the message’ designates a variability that is always
already limited by the condition that the source of information continues
to employ the same set of symbols and that this set is finite. It is a ‘relative
entropy’ that implies a certain amount of redundancy, in other words of
repetition (frequency) giving it a head-start on the margin of predictability.
As Weaver puts it:
If the relative entropy of a certain source is, say 0.8, this roughly means that this
source is, in its choice of symbols to form a message, about 80 per cent as free as
it could possibly be with these same symbols. (Shannon and Weaver 1964, 13)
If the relative entropy of a source (of continuous signal transmission, like for
instance a radio transmission) is given a value of 0.8, then the remaining 0.2
corresponds to constraints that are placed upon the message, in other words,
to what makes this message minimally predictable and hence, to what will be
redundant within it:
One minus the relative entropy is called the redundancy. (Shannon and Weaver
1964, 13)
Constraints on the entropy of the message can be, for instance, statistical rules
governing the use of symbols, or the set of letters in an alphabet or syntactical
rules. The predictable part of the message is what can be reconstructed and is
therefore considered to be inessential to the novelty of the message, and in this
sense ‘redundant’. It is what separates the ‘entropy of the message’ from complete
randomness or noise. Interestingly Weaver goes so far as to call the redundant
part of the message unnecessary, which appears to suggest that the message can
still be a message without it:
[T]his fraction of the message [that] is in fact redundant in something close to

the ordinary sense; that is to say, this fraction of the message is unnecessary (and
hence repetitive or redundant) in the sense that if it were missing the message
would still be essentially complete, or at least could be completed. (Shannon and
Weaver 1964, 13)
Weaver is of course right in the sense, for instance, that most vowels can be
left out of a typed message, without making it impossible to reconstruct the
message. The journalistic convention of replacing letters in offensive words
with the symbol * (i.e. f***) is indicative of the fact that what is redundant need
not be reiterated. Redundancy is nothing other than the predictable part of a
message. Weaver indeed goes on to note that language has a very high level of
basic redundancy:
It is most interesting to note that the redundancy of English is just about 50 per
cent, so that about half of the letters or words we choose in writing or speaking
are under our free choice, and about half (although we are not ordinarily aware
of it) are really controlled by the statistical structure of the language. (Shannon
and Weaver 1964, 13)
The conceptual presence of necessity here slips into Weaver’s expression

nonchalantly, in the form of a negation: redundancy is the part of the message
that is not necessary. However, to define redundancy as unnecessary, meaning
inessential or accidental, could lead to a misunderstanding of far reaching
theoretical consequences. While Weaver appears to say something obvious, the
concept of necessity is one that cannot leave philosophical analysis of information
and noise indifferent. Weaver’s way of putting it, namely that redundancy is
the part of the message that is not necessary is potentially misleading, not least
because his introduction seeks to lay the conceptual foundations for a new
understanding of the broader theoretical relevance of MTC. It is a far from
negligible slippage of logic to describe redundancy as unnecessary, because it
shows and even performs that the necessary, that which cannot not be and which
constrains ‘freedom of choice’, is what, as self-evident, can be left unsaid and
hence un-thought.
What does it mean for the redundant part of the message to be unnecessary?
The necessary is, in simple terms, whatever is absolutely indispensable and
hence of utmost importance (as for instance in the expression of the ‘bare
necessities’ for survival). Necessity can also be understood as a constraint, such
that its stringency or unavoidability is recognized in law even where a necessity
contravenes the law, as in the expression ‘state of necessity’: the necessity to
safeguard the interests of a person may, before the law, result in the impunity of
an incriminating act (dir. Jaen-Marie Pierrel et al. n.d.).
Necessity thus designates what is required by a situation (material, practical,
technical or vital necessities), but more fundamentally, in the philosophical
tradition, that which cannot not be, or which cannot be otherwise. In other
words, necessity is the mother of all philosophical concepts: a categorial, logical
or metaphysical necessity is what reason posits as valid in any circumstance and
whose contradiction is an impossibility. For reason, necessity is nothing less
than the axiomatic starting point of rational thought. Everything else has been,
since Greek Antiquity, attributed to the order of opinion, of mere phenomena or
appearance. What is not necessary is contingent: either absolutely contingent or
contingent upon a necessity that we may or may not know.
The introduction of probability into reasoning is, therefore, a significant event
in the history of thought. That something can be said to be 0.2 per cent certain
and 0.8 per cent uncertain introduces the possibility of nuance and process:
genesis and corruption are no longer excluded from the realm of reason. It is, in
epistemological terms, the metaphorical equivalent of introducing colour into a
black and white vision of truth. All the more reason to take note that necessity
is what Pascal, one of the founding fathers of the calculus of probability, called
a ‘state of constraint or restraint that annuls freedom of choice’ (dir. Jaen-Marie
Pierrel et al. n.d.; Pascal and Guern 1987).
Now, if redundancy is the part of the message that imposes a constraint, that
reduces ‘freedom of choice’ in terms of the message’s probability, then it is hard
to see how it could be unnecessary to the message. Redundancy is what in the
message remains constant, what is stable and not subject to degradation through
noise. Redundancy, in other words, is the very state qualified by the Latin root of
necessity: non cedens, that which does not give in and which, in its regular form
necée, is close to the idea of chastity: untouched by genesis and degradation.
Is redundancy in the message not precisely that which remains untouched by
‘freedom of choice’, by entropic degradation of the message, by contingency, in
short, by noise?
Redundancy, without which ‘information entropy’ would be indistinguishable
from noise, is thus not only necessary to the message, it is what, as self-evident,
becomes the invisible or unthinkable a priori of information. The consequences
of underestimating redundancy as unnecessary are far from trivial, if we
acknowledge that every form of organization is based on constraints that
introduce redundancy. Every system is informed by constraints that discriminate
‘freedom of choice’ according to given (hence redundant) criteria of pertinence.
To euphemize redundancy as unnecessary thus risks losing from sight what

founds every system: necessity, in other words that without which a system
cannot be what it is. By becoming redundant in the sense of self-evident,
necessity is what can remain unspoken and hence un-thought. Necessity, that
which makes a system what it is, is thus also what most easily subtracts itself
from critical analysis, as an a priori of this system, of this organization, of this
way of functioning, that is always already taken for granted.
In linguistic terms, the redundant part of language is what can be taken for
granted, what can be ignored. But we must not overlook the fact that it can afford
to appear unnecessary, it can afford to be ignored, only as long as it operates as
the condition sine qua non of communication. Redundancy, contrary to Weaver’s
claim, is necessary – and only therefore is it not necessary to reiterate – just as
good manners do not need spelling out where they can be taken for granted, and
just as the work of illegal immigrants can remain invisible to the public, while
operating as a condition of possibility for an economy hungry for cheap labour,
unregulated by workers’ rights. The question could be transposed onto the idea
of work redundancy. It is especially relevant in view of the digital revolution,
which exposes entire segments of professional activity to potential redundancy.
The question, as with redundancy in language, is this: which a priori are being
taken for granted? What goes without saying? A critical approach would be to
seek the blind-spot in redundancy, in order to debunk mere preconceptions
masked as necessity.
When the necessary conditions of the information process, of communication
more generally and of organization (including social or political) are what is
redundant, then they are nothing less than the a priori of our way of thinking
and acting. Yet if we fail to address the informational value of redundancy by
minimizing it as ‘not necessary’, then it becomes increasingly difficult to ask:
when is an a priori a necessity, a sine qua non of being thus, of thinking and of
communicating thus? And when is it mere prejudice?
The a priori restriction on the ‘freedom of choice’ in the message, is nothing
less than the condition of possibility of communication, also because, without
it, nothing would offset the uncertainty that is a correlate of the novelty of
information. In other words, without redundancy the pure novelty [entropy]
of information would be absolutely incomprehensible and equivalent with
noise. It is only on the basis of redundancy that novelty demarcates itself from
what is already certain. Redundancy is, furthermore, in this sense, also an
essential concept for our understanding of physical entropy. For the measure
of entropy in a physical system is a direct correlate of the knowledge we already
have of it and the knowledge we lack. The knowledge we have of a system, for
instance of chemical rules of interaction between elements, acts as a constraint
in epistemological terms: it reduces the entropy of the system. Without this
knowledge the behaviour of the system is absolutely random to us. Conversely,
what we call complexity is a correlate of low redundancy, in other words, of a low
level of pre-knowledge about the system.
By complexity we must understand the degree of indeterminacy of a system,
rather than its level of structural complication. Biophysicist and philosopher
Henri Atlan defines complexity as the measure of the observer’s ignorance as
to the precise determination of a system. Greater complexity of information
denotes greater uncertainty. What he calls the ‘maximum maximorum’ of
ignorance is the state of greatest complexity. It corresponds to the most basic
measure of information in Shannon’s sense (H), which informs us only about
pure multiplicity, nothing but the number of elements in a system (H= log N).
Atlan calls this the first, ‘trivial and maximal’ measure of complexity. It
corresponds to the observer’s maximal ignorance of other factors, such as variety,
frequency and other constraints. The second measure of information takes into
account statistical distribution and frequency (H= Σ p log p). Its quantitative
value is therefore smaller than the first, as its complexity is reduced. The third
measure of information, finally, introduces redundancy through the addition
of constraints [H = Hmax (1 – R)]. This corresponds to the least complex level
of information, as determining factors carve away at the complexity of the pure
multiplicity that characterized the first and maximal level of complexity. (Atlan,
1979, p. 80).
IX
Logic and Freedom of Choice
In logic, deductive redundancy is achieved when each proposition is tightly

correlated with the propositions deduced from it. Step by step, each proposition
becomes resonant with the others and no element can be modified without
compromising the whole (Blanché 2009, 10). The philosopher of mathematics
Robert Blanché thus describes the process of deductive thinking as propagating
a structure of constraint where, like the emerging lattice structure of a crystal:
[S]tep by step, a tight network is constituted where, directly or indirectly, all
propositions communicate. (Blanché 2009, 9–10)
This is why one speaks of the deductive resonance of a mathematical theorem if,
like a crystalline structure, it achieves a rock-solid correlation between each and
all terms, where nothing is left to chance and no ambiguity can arise. The logical
coherence of the whole can then be called isomorphic, like the lattice structure
of a crystal.
This resonance is a form of redundancy in the logical chain. It eliminates

ambiguity, just as redundancy in the transmission of a message serves to reduce
noise. It was long hoped that logic would provide a language without ambiguity,
in other words without noise. If the elimination of ambiguity could ensure the
truth of all statements in a logical chain, by eliminating ambiguity and hence
error, then deductive redundancy was deemed necessary in order to stabilize
philosophical discourse.
If, however, all statements are not bound by necessity then they retain a
trace of ambiguity. This is the case for unproven postulates, in other words, for
all statements that are not yet proven axioms. A single postulate in a system
of axioms becomes the gateway to a potentially radically diverging axiomatic
system. A paradigm shift thus lies dormant in each postulate, in each trace of
ambiguity and openness. This ambiguity and openness is what we will call noise,
because it persists as a margin of uncertainty and freedom of choice, until it too
is axiomatized. The fifth postulate in Euclid’s Elements long remained open to

such uncertainty, because of the impossibility of proving it. According to this
postulate, also called the ‘parallel postulate’:
If a line segment intersects two straight lines forming two interior angles on the
same side that sum to less than two right angles, then the two lines, if extended
indefinitely, meet on that side on which the angles sum to less than two right
angles.
The impossibility of proving this postulate despaired mathematicians for

centuries. The uncertainty it represented revealed itself as a freedom of choice
only when, in 1829 Nikolai Ivanovich Lobachevsky could prove that the fifth
postulate does not hold true in a geometry of infinite dimensions. The search
for the impossible proof of Euclid’s fifth postulate thus paved the way for the
emergence of so-called hyperbolic or ‘non-Euclidian’ geometry. The classical
three dimensions of Euclid’s geometry opened up to the infinite dimensions of
Lobachevsky’s geometry, and as a consequence, also the space-time of classical
physics could eventually open up to the possibility of special space-time relativity
(Blanché 2009, 47).
Eric Temple Bell called Lobachevsky the ‘Copernicus of all thought’, and
saw in Lobachevsky’s work an incentive for mathematicians and scientists
to ‘challenge other “axioms” or accepted “truths”’ (Bell 1986). For us, the
ambiguity of Euclid’s fifth postulate can be thought as a borderline case, where
uncertainty becomes freedom of choice, in other words, where noise becomes
information. If Euclid’s fifth postulate represented a source of uncertainty for
centuries, Lobachevsky transformed this uncertainty into freedom of choice,

opening up three dimensions to infinite dimensions, and opening up the idea of
mathematical truth to the emergence of new paradigms.
In the early twentieth century, Georg Cantor set out to harmonize the domain
of mathematics and to rescue the unity of mathematics in light of the plurality
of geometries. In the attempt to axiomatize set theory, Cantor encountered
foundational paradoxes that shook mathematics to its core no less than
Lobachevsky’s achievements in geometry, and thereby irreversibly unmoored
its theoretical capacities from its last anchorage in intuition. Mathematics and
geometry were henceforth wide open to the prospect of infinite or transfinite
sets of infinite numbers, chaperoned only by the paradox that the set of all stets
cannot ground itself. In response to the crisis in the foundation of mathematics
that was provoked by Cantor, logic, it was hoped, would reign mathematics in
and prevent its speculative excesses.
Robert Blanché gives an account of the axiomatization of mathematics that

is valuable in this regard, because it helps understand both the necessity and
the absurdity of deductive redundancy, when taken to its extreme and when
erected as the sole pillar of truth. Logic, it was hoped, would rescue mathematics
from the paradoxes of axiomatic set theory, which appeared to have turned
mathematics into a
science where one never knows what it is that one is talking about, nor whether
it’s true. […] By proposing to ground […] the entire edifice of mathematics on
logic, Frege and Russell’s ‘logicism’ aimed further than returning to its principles:
it intended bringing it to its term, reaching the rock, the ultimate foundation.
(Blanché 2009, 70)
It is history, by now, that rather than eliminating the antinomies that had sprung
up within axiomatic set theory, disagreement about the validity of logical
principles were, in turn, to shake also the foundations of logic to their core,
ultimately demolishing ‘the idea of an absolute, unique and universal logical
legislation’. Blanché reconstructs how the axiomatization of logic finally led
to the ‘disintegration of logic from within’, issuing forth into a pluralization of
logics.
Even if the intra-logical and axiomatic problems raised by a plurality of logics
and of axiomatic systems could be set aside, what remains problematic is thus
the ‘fit’ of logical redundancy and reason’s intrinsic complexity. Its formalized
terms, albeit submitting to the utmost criteria of necessity, no longer represented
anything but the mediation between ‘simple tautologies’. Although perfectly
harmonized in a deductive redundancy, without what we may call the noise

of ambiguity, they ‘say strictly nothing about the real, but […] for this reason,
remain valid whatever content one applies’ (Blanché 2009, 71).
In a different context the philosopher of biology Marjorie Greene struggled
with the limitations of Logical Positivism in conceiving of the problems specific
to biology. Its incapacity to deal with the imprecision of the empirical world
eventually turned Greene away from Logical Positivism, in which she saw a
sterility when it came to the life sciences:
In the Anglophone tradition (which I derive partly from the Germano-Austrian
tradition) that which one called the received view dominated until recently. I
participated personally in Carnap’s seminar in Chicago during the year 1937–
1938. Having previously studied zoology, I was rapidly disappointed. It seemed
impossible to treat the praxis of zoology with a purely extensional logic. I tried to
explain this difficulty to Hempel, who was Carnap’s assistant in this seminar and
he replied: ‘We only say what we can explain with precision.’ […] twenty-five
years later […] logical positivism was taught under its new less aggressive name
of logical empiricism. It dealt with laws, theories, the deductive relation between
theories and laws, the problem of confirmation etc. […] Today however, this
old orthodoxy is, if not entirely buried, then in a – how shall I say – catatonic,
vegetative state. (Grene 2007, 24–25)
Greene’s disappointment in articulating the deductive power of Logical

Positivism with the open field of biological complexity is directly relevant
to our interest in the question of ‘epistemological noise’, to the unavoidable
ambiguity and distortions of concepts and theories, when we seek to cross-
fertilize fields of knowledge and praxis. It shows that ‘epistemological noise’ is
not merely a deplorable side effect of interdisciplinary communication, or of the
transdisciplinary formulation of scientific problems, but a positive requirement,
without which specialized discourse becomes not only sterile, but worse,
redundant.
X
Noise as Spurious Uncertainty
The question that has emerged from the previous sections is: how do we draw
the line between constraint and ‘freedom of choice’? We cannot avoid complexity
and ambiguity entirely without risking sterility of information, but we still need
to impose a boundary between a level of complexity relevant for the formation
of knowledge and infinite complexity, which consigns us to the power of oracles,
not reason. Here lies the difficulty in distinguishing between the ‘entropic’
understanding of information and the entropy of noise. Yet this is precisely the
question that is suspended when the engineer transmits a readymade message,
regardless of whether it is the rambling telephone conversation of someone’s
mother-in-law, an encrypted message, or Schönberg variations. The a posteriori
evaluation of what is received, as either spurious or significant, literally doesn’t
come into the equation. Any value judgement that discerns a message with the
mark of distinction, namely that it is informative, is either pre-given, a priori, in
the decision to transmit a message as information, or it occurs a posteriori, as the
result of a process of evaluation and interpretation.

The relatively restrictive scenario, where an already selected signal is
transmitted as information has this particularity: the decision upon which
we base the distinction between information and noise has already been
made when the message is transmitted. The engineer treats the message as
ready-made, but in order for information to be transmitted, in order for the
entire technological apparatus of signal transmission to exist, a decision must
have preceded transmission: namely the decision that information must be
transmitted, and the decision that what is transmitted is information. Even
if this is not a decision that concerns the communication engineer, it is a
decision that needs to be made, repeatedly, every time a message is selected for
transmission as information.
However, the pre-established distinction between information and noise
prevails only in the narrow technical context of communication, presupposing
the decision has been made and information has been selected for transmission.
Outside the fully specified function of transmission, the question ‘what counts as
information?’ is wide open. When faced with new experiences, our ready made
distinctions between information and noise, are more often than not contested.
The redundancy of pre-established constraints on information’s ‘freedom of
choice’ is only a temporary fix before the ultimately unpredictable course of events
in our human, geological and cosmological history. The distinction between
information and noise is stabilized by way of scientific methods that were once
innovative and that become conventions and even traditions. Enshrined in the
institutions of knowledge, almost iron-cast into the transmission of knowledge,
methods become habits of thought, so deeply engrained that they inhabit
the blind spot of redundancy, rendering indistinguishable the conditions of
possibility of knowledge from the discovery of what we find informative.
What we are left with is thus a tenuous, historically contingent line separating
the entropy intentionally selected as information and the entropy spontaneously
adding itself to information as noise. How to draw this line becomes an
epistemological problem once the concepts of information and noise are taken
outside the context of electronic signal transmission, as when they are translated
into molecular biology or systems analysis more generally – and a fortiori when
they enter general discourse.
This is perhaps why Shannon’s audacious levelling of information and
uncertainty is attenuated by Weaver, as a way of curtailing the radical
consequences of aligning information with uncertainty, revealing a reluctance
perhaps to let go of the clear-cut distinction between information and noise.
Weaver indeed acknowledges the bond between information and uncertainty,

but only to rescue what he can of the ordinary understanding of information, by
expelling noise as ‘spurious’ uncertainty:
It is generally true that when there is noise, the received signal exhibits greater
information – or better, the received signal is selected out of a more varied set
than is the transmitted signal. […] Uncertainty which arises by virtue of freedom
of choice on the part of the sender is desirable uncertainty. Uncertainty which
arises because of errors or because of the influence of noise is […] spurious and
undesirable.
Weaver’s assumption therefore is that noise tells us nothing new, because there
is no telling at its origin, no intention: what noise tells us is ‘spurious’ because
accidental. Yet, as we have seen above, noise cannot be accidental in the sense
that it is less necessary than ‘information entropy’, since both are and remain
a measure of uncertainty whose positive quantity is negatively correlated with

redundancy. Noise can be spurious in Weaver’s sense only if it is accidental vis-
à-vis intention
Intention, however, pertains only to semantic communication, where some
form of consciousness can be presumed. But information theoretical concepts
of information and noise have proven their relevance in fields that far outstrip
problems of intentional communication. Consider for a moment just how
broad the field of conceptual relevance is for the concepts of information and
noise: if we accept that the very concept of system relies on communication
between its parts and the whole, then all systems can be thought of in terms
of processes of information. All empirical systems imply communication,
which can be understood as ‘causal relation’, or with Ross Ashby as relations
of ‘non-independence’ between its parts and the whole (Ashby 1962) and even
more cautiously as merely statistical correlation. From the quantum level to
the cosmic level, all correlations of events can thus be formulated more or less
loosely in terms of information processes.
Yet if we accept this premise, which vastly broadens the concepts
of information and noise beyond the scope of verbal or technical
communication, then intention can be considered to be relevant to only a
small number of in principle infinitely many possible systems that operate
through processes of information. Conscious intentional communication,
which we perhaps too hastily attribute to human beings as a mark of
distinction, becomes a limited domain, the only domain where the
distinction between desirable and ‘spurious’ uncertainty pertains. We may
have to concede that the centrality of human communication, understood

as a semantic and culturally saturated information system is, at least in
principle, neither the first system in which information processes occur,
nor necessarily the most efficient.
Systematic communication, understood broadly as non-independence
between parts and the whole, can be assumed to have preceded our cognitive
faculties to communicate, via physical, chemical, evolutionary processes of
information. Even if we concede that we can have knowledge of any such
system only in the cultural domain and via semantic mediation, that the very
concepts of system, information and noise are nothing but a mental construct,
we cannot account for mental constructs without assuming that there is genesis
and transformation of form that has preceded, as a condition of possibility,
the forms of cognition and the processes of information we call our own.
How apt a handle the concepts of system, information and noise provide, as
mental constructs, to think any such processes in themselves, is another matter

altogether.
When Weaver narrows the conceptual reach of Shannon’s ‘entropic ideas’, by
consigning Shannon’s concept of information to intentional communication,
he thus generalizes the pre-given distinction between information and noise.
And yet, intentionality is not a concept comprehensive enough to encompass
the difference between information and noise on the large spectrum of
theoretical and experimental domains in which Shannon’s concepts are
relevant.
The distinction between intentionally and accidentally transmitted ‘entropy’,
between information and noise, is precisely what is not pre-given in most of the
contexts into which the information-theoretical concepts have been translated.
Most areas of scientific research, on the contrary, explicitly set out to redefine
the boundary between relevant and irrelevant phenomena, between uncertainty
productive of new scientific insights, and uncertainty that is a mere distraction
or loss of focus. The very core of scientific endeavour is to establish what, in
the vast contingency of interrelated empirical events, is relevant as information
and what, on the contrary, stands in the way of the identification of phenomena
relevant for enquiry. The difference between information and noise is thus rarely
ready-made, and always provisional in the context of research.
The difficult elaboration as to what, in a research question, must be isolated
as a relevant phenomenon, and what, on the contrary must be discarded as
irrelevant is ultimately a methodological decision. It is always threatened with
the potential error that what is excluded as noise may lengthen the path of
erring towards understanding. Even acquired certainties are subject to this rule,
whereby history unravels scientifically held beliefs as ‘epistemological obstacles’,
as Bachelard would say, often prompted by phenomena of perturbation or errors
that reveal a flaw in those certainties, and thereby turn information (in the sense
of acquired certainty) into noise and noise into information (in the sense of a
productive uncertainty).
Weaver’s idea of noise as by definition ‘spurious’ uncertainty thus risks
unnecessarily limiting the relevance of Shannon’s open definition of information
as ‘information entropy’, by artificially hardening a boundary between the
measure of entropy that is apportioned to ‘information entropy’ and the measure
of entropy that is considered to be an excess, i.e. noise.
XI
Negentropy
There is a peculiar tension between on the one hand Shannon’s very minimal
definition of information as pure probability, shaved even of Boltzmann’s
reference to physical processes (k), and on the other hand of the cluster of
concepts relating to purpose and organization that characterize negentropy
and even Weaver’s reliance on intention. Peculiar because it opposes different
epistemic frameworks for the conceptualization of information and noise, that
don’t reduce well to mere opposition, such as the idea of the simple negation of
entropy suggests. The notion of negentropy in fact implies a multilayered concept
of information. No longer a pure concept of probability, theories of organization
and order are tacitly aligned with the concept of negentropy. The question this
concatenation raises is the following: how can the negation of entropy increase
the complexity of information, which we commonly associate with organized
forms of life?
The neologism negentropy first established itself first in light of Erwin
Schrödinger’s conception of life as that which distinguishes itself from the

proneness of physical processes to entropic dissipation of energy and death
(Schrödinger 1945). Leon Brillouin’s seminal article ‘Life, Thermodynamics and
Cybernetics’ (Brillouin 1949) has further cemented this and Norbert Wiener’s
use of the term information as a measure of prediction of future behaviour in the
domain of control and communications systems. Information here answers the
question, reiterated by Brillouin:
[W]hen we possess a certain number of data about the behaviour of a system in
the past, how much can we predict of the behaviour of that system in the future?
(Brillouin 1949, 554)
This conception is now cemented in current usage with Brillouin’s neologism

negentropy. The degree of organization of a system thus comes to represent its
quantity of information, which contrary to Shannon, stands in inverse relation
to entropy. This leads Brillouin to make observations about the relation of
information to time and memory, which we cannot pursue here, but which
would be of great interest to every Bergsonian or Deleuzian.
As we have seen earlier, Warren Weaver also signals this link between
organization and the negation of entropy in his introduction to MTC, but on
the contrary explicitly states that, in light of Shannon’s entropic ideas, the state
of organization is one where ‘information […] is low’:
Thus for a communication source one can say, just as he would also say it
of a thermodynamic ensemble, ‘This situation is highly organized, it is not
characterized by a large degree of randomness or of choice – that is to say, the
information (or the entropy) is low’. (Shannon and Weaver 1964, 13)
Weaver and Brillouin’s shared assessment of the role of organization as constraint

of ‘freedom of choice’, in other words, as negation of entropy, must not lead us to
overlook that two very distinct or rather opposed definitions of information are
nevertheless at stake depending on whether we define information as negentropy
or, with Shannon, as ‘information entropy’. In Science and Information Theory
Brillouin makes this difference explicit:
Entropy is usually described as measuring the amount of disorder in a physical
system. A more precise statement is that entropy measures the lack of information
about the actual structure of the system. This lack of information introduces
the possibility of a great variety of microscopically distinct structures, which
we are, in practice, unable to distinguish from one another. Since any one of
these different microstructures can actually be realized at any given time, the
lack of information corresponds to actual disorder in the hidden degrees of
freedom.
[…] The origin of our modern ideas about entropy and information can be
found in an old paper by Szilard […]. The connection between entropy and
information was rediscovered by Shannon, but he defined entropy with a sign
just opposite to that of the standard thermodynamic definition. Hence what
Shannon calls entropy of information actually represents negentropy. […] To
obtain agreement with our conventions, reverse the sign and read negentropy.
(Brillouin 1949. Emphasis added)
What does Brillouin mean, when he says that Shannon defines entropy with
a sign opposite to the standard thermodynamic definition? When Brillouin
says ‘he [Shannon] defined entropy with a sign just opposite to that of the
standard thermodynamic definition’ then presumably he does not mean
Boltzmann’s definition (S = −K Σ pi log pi.), which as we have seen is preceded
by the same minus sign as Shannon’s definition (H = −Σ pi log pi). By standard
thermodynamic definition Brillouin must mean Wiener’s definition of

information as negation of entropy. To follow Brillouin we must ‘reverse the
sign and read negentropy’, in other words, the definition of information as
negentropy is based essentially on the negation of a negation.
Not surprisingly this convoluted relation between information and entropy
lends itself to a certain amount of confusion when the terms information
and noise enter the mainstream. The concept of negentropy is in fact not
rarely justified with the minus sign that precedes the symbol for entropy [Σ]
in the mathematical expression for ‘information entropy’ [H = − Σ pi log
pi]. This apparent subtraction of the value of entropy from the quantity of
information [H] appears to chime with Wiener’s statement that information
is simply the negative of entropy or disorder: entropy consequently can be
said to decrease as information [H] increases. In other words, the idea is, the
smaller the quantity of entropy, the more reliably we can predict. Raymond
Ruyer, in La cybernétique et l’origine de l’information, explains this point of
view:
Entropy goes in the direction of the most probable states; information, with the
opposite sign, is thus an ‘anti-probability’ or, to use Edington’s expression, an
‘anti-contingency’ [anti-hasard] […[ the formula that expresses it is exactly the
formula of entropy, the logarithm of equal probability, but with the contrary
sign. Information is negative entropy. (Ruyer 1954, 114–15)
What results is an apparent paradox that has riddled the interpretation of

Shannon’s use of the term information ever since. How are we to interpret the
minus sign that precedes Σ, in both Boltzmann’s and Shannon’s formalization

of ‘information entropy’? Warren Weaver demystifies this difficulty for non-
mathematicians in his introductory essay to MTC, by spelling out the role of
this minus sign:
Do not worry about the minus sign. Any probability is a number less than
or equal to one, and the logarithms of numbers less than one are themselves
negative. Thus the minus sign is necessary in order that H be in fact positive.
(Shannon and Weaver 1964, 15, n. 5)
The minus sign in Shannon’s mathematical expression of ‘information entropy’

thus expresses in positive quantitative terms the entropy, hence ‘freedom of
choice’ of a message. Clearing up any potential confusion, we can now say that the
minus sign in the standard mathematical definition of entropy, and in Shannon’s
definition of ‘information entropy’, expresses the positive value of entropy or
‘information entropy’. It does not express the negation of entropy or negentropy.
It is the reversal of this minus sign, which reduces ‘freedom of choice’ in terms
of probability, and hence ‘entropy’. The minus sign whereby any probability below
1 is transformed into a positive number, meaning the entropy of the message, is
what must be reversed, in order to align the positive quantity of information
not with entropy but with the negation of entropy. Hidden in this reversal of
a reversal is thus a major redefinition of the concept of information, but also a
subtle shifting of ground from information as pure probability to information
as index of organization (still understood straightforwardly as negation of
entropy) and from organization to order (as the negation of disorder, but also as
instruction or function).
Where Shannon freed the concept of information from the reference to physical
potential, as we have seen earlier with reference to the algorithm ‘k’, Brillouin
reintroduces the idea of physical potential as a precondition for the organization
of vital processes. Information is therefore put under the helm of Schrödinger’s
nascent theory of emergent biological organization, which is tacitly aligned with
the idea of purpose that underlies Wiener’s cybernetic definition of information.
In short, negentropy becomes an index of organization, which in turn becomes
tethered through cybernetics to the telos of predictable functioning in manmade
systems.
Brillouin rightly recognized that highly organized systems delay the entropic
dissipation of energy. In this sense it is true that Shannon’s notion of ‘information
entropy’ cannot suffice on its own to qualify the informational content we associate
with highly organized systems and that redundancy, (understood as frequency
or constraint which reduces entropy), must be taken into account as one of the
conditions of possibility of organization. It can be agreed that what qualifies highly

organized systems is the capacity to sustain conditions far from equilibrium.
However, by reversing Shannon’s ‘entropic ideas’ of information, negentropy
cannot help but become nothing but a degree of predictability. Paradoxically, the
greatest measure of information then is what is most predictable and consequently
what tells us nothing new at all. Equating information with negentropy means that
the maximum of information is, ultimately, what is completely redundant.
Negentropy, by negating entropy, renders equivalent the idea of ‘information
entropy’ and noise: both negated as a mere measure of entropy. It thereby
surrenders novelty, unpredictability and hence also complexity of information
to the imperative of certainty at any cost. Yet what is lost when Shannon’s
definition of ‘information entropy’ is simply negated, is whatever occurs with a
lower probability than the redundant message. As a result, the idea of growing
information complexity makes no sense in terms of negentropy. The unintended
consequence of this negation is that, strictly speaking, information and indeed

organization can learn nothing new. They cannot become more complex, but
only less complex.
There is thus a paradox inherent in Norbert Wiener and Claude Shannon’s
radically divergent evaluations of the same mathematical formulation of
information entropy. And yet, it is important to stress that the mathematical
formulation based on Boltzmann’s definition of entropy ultimately remains
unchanged. What changes is only the discursive, conceptual and theoretical
conversion of entropy into its negation, of ‘freedom of choice’ into constraint, of
pure probability into conceptions of order and disorder.
The fidelity to the mathematical formula shows that it is the need to interpret
and translate mathematical concepts into general discourse that becomes a source
of what we can call ‘epistemological noise’. The multiple conceptualizations of
information, and implicitly of noise, that have sprung forth from cybernetics
and information theory across the natural sciences, but also the humanities,
thereby come to confront different ideals of information and organization: on
the one hand an ideal of information as ‘freedom of choice’ and on the other
hand an ideal of organization as increase of constraint. While ‘freedom of choice’
must take uncertainty in its stride, the functional constraint of the cybernetic
information model must ensure control through noise cancelling feedback.
Where Shannon’s concept of information risks becoming indistinguishable from
noise, Wiener’s cybernetic definition risks producing a radically conservative
form of information that can learn nothing new and can only preserve what is
already there.
The conceptualization of information, and as a consequence of noise, is caught

in this field of tension. Yet a satisfactory definition of either information or noise
cannot be reduced to either of the two, entropy or its negation, except at the cost
of becoming self-contradictory when taken to its extreme as a maximum value,
i.e. by becoming either pure noise or mere redundancy.
XII
Complexity on the Basis of Noise
Neither absolute uncertainty nor complete redundancy, on their own, thus

suffice for a notion of information that does what the word information says,
which is to inform. A viable concept of information must satisfy the criterion
of genesis of form, whether we understand form in Platonic terms as idea, or
in morphological terms as the emergence of a topological difference. Gilbert
Simondon recognizes this tension when he defines information as
an activity and an irradiation, the capacity to enlighten new domains. (Simondon
2005b, 541)
A concept of information that satisfies the requirement for this capacity to

sustain a process of genesis of form and transformation of diverse domains, must
be capable of sustaining the conceptual tension between the two opposed terms.
Wiener and Brillouin’s negation of entropy does not suffice to characterize
highly organized systems, if by ‘highly organized systems’ we mean those systems
that maintain themselves far from equilibrium, like a biological organism or an

ecosystem functioning on the basis of homeostasis, i.e. of self-regulation of an
internal milieu. Yet it is precisely to this understanding of self-regulation that
the cybernetic model extends itself. The negation of entropy, paradoxically,
substitutes one form of equilibrium for another, negating entropic equilibrium by
substituting it with structural equilibrium. To be far from equilibrium, however,
means that a system maintains itself far not only from entropic equilibrium,
but also far from the equilibrium of structural redundancy. As a consequence
neither the concept of negentropy, nor that of ‘information entropy’ can account
for a median term on their own, i.e. for a dynamical equilibrium or metastability.
We have already seen that the state of maximal entropy is one in which all
exchanges between unequal energy levels have taken place, all energy potential
has been exhausted, until no energy potential further propels the system to
behave or move in any particular way. However, also its negation, for instance
the lattice structure of a crystal is (macroscopically at least) at rest when all

possible molecular reactions have resulted in stable molecular concatenations.
The process of crystallization is one example whereby energetic equilibrium is
attained through structural equilibrium, by means of a process of molecular
interaction that diffuses entropy in the form of heat into the environment.
The crystalline structure itself has more or less exhausted the chemical and
physical potential for transformation that was present in the saturated solution
from which it emerged. It remains at rest, unless it still contains potential for
further crystallization if catalysed, for instance, by a change in temperature.
Consequently, the process of crystallization, too, reaches energetic equilibrium,
as the final (relative) absence of entropy falls short of a self-perpetuating
dynamic or metastability, an instead results in the rigor mortis of a system
with energetically spent structures. Paradoxically, the ultimate conclusion of a
concept of information based on the negation of entropy would be that a two-
kilogram crystal contains more information than the average human brain, on
the grounds that it more effectively negates entropy. While it certainly makes
sense to qualify an organized system as rich in information according to its
self-regulating capacity, the simple negation of entropy fails to do justice to the
generative and transformative qualities we associate with information.
In both forms of equilibrium, structural and entropic, no physical potential
is left to propel transformation further (Atlan 1979, 31; Tonnelat 1996). The
difference between structural and entropic equilibrium is that one form of
energetic equilibrium is bound up in structure, while the other allows for the
random movement of free particles. Neither form of equilibrium on its own,
however, can be said to retain the potential for the genesis or transformation of
form. As a consequence, if by information we seek to characterize processes of
organization capable of sustaining themselves in a changeable environment, if
we agree that information is what qualifies organized systems that are both far
from equilibrium and capable of growing complexity, then neither ‘information
entropy’ nor its negation can suffice on their own to account for the genesis of
organized form and its dynamical and transformative potential.
It becomes clear that neither the maximal state of entropy nor that of structural
equilibrium can fully encompass the notion of information, whose maximal
value would in both cases correspond to the loss of potential for the genesis and
transformation of form. Fully redundant ‘information’ and maximum uncertainty
of ‘information entropy’ alike fail to maintain the potential to inform. The
dialectic of negation between ‘information entropy’ and negentropy thus leads
us to a dead end. What characterizes the resilience of a dynamical and adaptable
system with organizational capacity thus cannot be the mere negation of entropy,
which would ultimately mean structural equilibrium rather than being far from
equilibrium. It must be, instead, the persistence of a metastable state. Metastability
is the dynamical suspension of a system between two forms of equilibrium,
between entropic dispersion and structural inertia. It was Gilbert Simondon’s
merit to have introduced the concept of metastability to the philosophical corpus,
by making it the cornerstone of his theory of individuation.
Insofar as our definition of information and noise must answer this tension
between both forms of equilibrium, it is not only the physical systems we can
describe as rich or poor in either ‘information entropy’ or negentropy, but
the very concept of information itself must be characterized by metastability.
Neither entropy nor its opposite, redundancy, can act as the seat of information:
if information can inform only when it is far from equilibrium, then it cannot rest
in either of the two forms of equilibrium. The process of information must be,
like the act of walking, a controlled way of falling or, as Henri Atlan formulates
it, a succession of ‘recuperated disorganizations’:
On can conceive of the evolution of organized systems or of the phenomenon

of self-organization more generally as a process of increasing structural and
functional complexity, as a result of a succession of recuperated disorganizations,
followed each time by re-organization with a higher level of variety and a lower
level of redundancy. (Atlan 1979, 49)
Information, in other words, must be sought in a cycle that expends and reloads
its potential for transformation, through repeated cycles of acquisition and loss
of equilibrium: it must fluctuate between the two extremes, between entropic

dispersion and structural rigidity, between uncertainty and certainty – without
succumbing to the temptation to seek rest in either of them. As a consequence,
the role of entropy in systems with potentially complex organization cannot
be dismissed as nothing but noise. Entropy cannot be eliminated without
undermining the very principle of organization as that which resists both
entropic dispersion and the rigor mortis of structural rigidity. Nor can entropy
constitute, on its own, a viable concept of information that can encompass
‘freedom of choice’ beyond its purely mathematical value of probability. The
line we draw between information and noise thus becomes the point of highest
tension: it is here that information can arise as genesis of form, as new idea, and
as transformation of forms of knowledge.
Consequently, if the general concept of information is to express the idea of
a sustained process of genesis and transformation of form, in a formal analogy
with physical processes, then it must include a form of disparity that is in some
way comparable to the disparity between energy levels in a physical system.
But how do we translate this understanding of tension created by disparity of
energy levels in a physical system, into a more general concept of disparity
or tension of an epistemological kind? The epistemological tension between
‘information entropy’ and negentropy could be defined as the tension between
on the one hand an a priori uncertainty, which gives us a measure of ‘freedom
of choice’ and, on the other hand, the empirical need to predict, decide and
act, which requires a reduction of this ‘freedom of choice’. What is needed is
a concept of information that places a relative uncertainty in the context of
existing knowledge and problems that constitute such a field of tension, in
other words, within which the antagonism between a priori uncertainty and
a posteriori reasons to believe creates the potential to propel a process of
information further.
Our conceptualization of noise in turn makes sense only when introduced
into our understanding of processes that must maintain themselves far from
equilibrium, and hence far from both absolute uncertainty and complete
redundancy. The need to address this tension as the very core of the concept
of information helps us to devise new criteria also for our understanding of
the relation between information and noise. Just as the sustained genesis and
transformation of form, which we call information, cannot be reduced to the
mere negation of entropy, so the role of noise, in other words of unintentional
or accidental increase of entropy, cannot be discarded offhand as having no
informational value. It calls for no less than a methodological revival of the
docte ignorance. At stake, however, is no longer the Cusean wisdom of knowing

the finitude of knowledge before God, but the knowledge we gain from the
mathematical specification of our uncertainty. Noise can thus be said to have a
positive epistemic value, if it is understood to specify the uncertainty in which
we are about the unfolding of a system.
Noise, as a result, is no longer an entirely negative nor an entirely unspecified
form of ignorance, but on the contrary is what specifies complexity and informs
on what remains to be known. Information and noise in turn become more
refined concepts, than the conflation of information with data, signification
or knowledge suggests, as for instance in the definition of information as
‘facts provided or learned about something or someone’, as in the Oxford
Dictionary (‘Information’ 2014a). The Merriam-Webster dictionary further
emphasizes this epistemological aspect of information as ‘the communication
or reception of knowledge’ (‘Information’ 2014b). These definitions suppose an
equivalence of information and knowledge, which is relayed also in the Penguin
dictionary, where information and knowledge form a whole, composed of facts,

data, signification and even ideas:
Information/[…]/noun 1a knowledge obtained from investigation, study, or

instruction […] b facts that represents data […] esp. regarded as significant […]
2 communication or reception of facts or ideas […].
The participation of noise in a positive aspect of uncertainty, closely welded

to the concept of ‘information entropy’, thus no longer indicates only lack of
data, loss of signification and the margin of error. Uncertainty instead becomes a
measurable function of information as possible knowledge, stratified according to
entropy of the message and spontaneous noise in the channel of communication.
Knowledge in turn, is not necessarily obtained only by elimination of uncertainty
and noise, but by cultivating noise as both reservoir and dynamo for learning,
allowing for the dual role of information, as both increasing ‘freedom of choice’
through the novelty of ‘information entropy’ and as reducing uncertainty
(information as negentropy) through a process of analysis and interpretation.
The idea that ‘the creation of information can only occur on the basis of noise’
was in fact already commonplace in research on artificial intelligence (Atlan
1979, 63–64; Cowan 1965; Neumann 1956). Henri Atlan has been amongst
the first to argue explicitly for a theory of ‘complexity on the basis of noise’
from the early 1970s. In order to adapt Shannon’s concept of information
to biophysics, Atlan refers to Ross Ashby’s theory of self-organizing systems
(Ashby 1962), in particular his ‘law of requisite variety’.
For Ashby a ‘function’ denotes a correlation and in this sense ‘communication’
between subsystems. Whether an organism can match the number of its

possible states to the alea of the environment, depends on the variety of its
subsystems. While noise in the ‘channel of communication’ between subsystems
is detrimental to a particular function, making the correlation it establishes
insecure, it can still be viewed positively, as an aspect of increased variety of the
system as a whole. Noise may, Atlan argues, ultimately contribute to what Ashby
defined as the ‘requisite variety’ of the system’s possible states. By increasing the
system’s number of possible states, its ‘freedom of choice’ in probabilistic terms,
noise can thus be said to contribute positively to the resilience of a complex or
hyper complex system in an unpredictable environment – provided it does not
precipitate the system’s break down. Any positive understanding of noise as a
factor of increased variety must therefore respect the difference between noise
affecting the channel of communication, and noise that is transversal to the set
of all channels of communication, constituting the system as a whole, which
it may affect positively as increased variety. This is where it begins to make
sense to think both Shannon’s entropic conception of information and Wiener’s

negation of entropy alongside one another.
Insofar as any system relies on the stable communication of its subsystems,
noise at the level of the subsystem is always going to be detrimental to its
predictable functioning. The very concept of organization is, if we follow Ashby,
premised on a mode of communication, understood as a more or less stable
set of correlations between subsystems. If the level of noise in each subsystem
is such that signal transmission breaks down, then the correlation between
subsystems no longer pertains, making them ‘independent’ from each other.
The more independent individual subsystems are, the more vulnerable the
system as a whole is to a break down. Functional constraint indeed negates
‘entropy’ in the sense that Schrödinger, Brillouin and Wiener, but also Shannon
and Weaver saw and clearly argued: organization requires redundancy. Let us
not forget that, despite our emphasis on the conceptual importance of Shannon’s
‘entropic’ definition of information, he is remembered first and foremost for
his contributions to the high fidelity of signal transmissions without noise. Yet
while noise can never add information at the level of the individual channels of
communication, where information is essentially negentropic, in the system as a
whole it can be seen to increase the variety of possible states, and consequently
improve its capacity to respond to a changing environment, allowing for a
greater number of possible states.
A system that is constituted by a hyper-large number of subsystems is less
likely to break down, if only some subsystems are affected by noise. Systems such
as those which von Neumann called ‘extremely highly complicated’ because of
the hyper large numbers of components and connections, and which Edgar
Morin baptized ‘hyper complex systems’ (Atlan 1979, 48), can thus afford a high
degree of noise without collapsing. The quantity of information considered at
the level of the organism as ‘information entropy’ (variety) can thus be seen to
increase on the basis of noise, in Shannon’s sense of ‘freedom of choice’, rather
than in Wiener and Brillouin’s sense of negentropy. Noise between subsystems,
although detrimental to their channel of communication, may thus even be
considered as creating information (as ‘information entropy’), in other words,
as increasing variety in Shannon’s sense of increasing ‘freedom of choice’ at the
level of the system as a whole.
Despite the now dominant definition of information as negation of entropy
(negentropy), Shannon’s positive value of ‘information entropy’ as a measure of
uncertainty has thus increasingly fed into various theories of ‘complexity on
the basis of noise’: in game theory, computing, artificial intelligence, but also
in theories of physical and biological complexity. Just as physical entropy is no

longer understood only as loss of available energy, as it was in classical mechanics,
but has become a core concept in our understanding of non-classical physics
and mechanics, so the role of noise is set to continue to enjoy re-evaluation in
light of the ‘information-entropy’ of complex systems.
XIII
The Astigmatism of Intuition
The positive appreciation of the informational value of entropy now belongs to

the mathematical dexterity, without which empirical reality can no longer be
understood beneath and beyond the threshold of intuition. The counter-intuitive
nature of mathematical reasoning, however, also leads to the pulverization of
the idea of simplicity: the apparently simple becomes maximally uncertain. The
analytical method of the modern sciences, based on the Cartesian method of
analysis, which consists in progressing from the simple to the more complex,
here it encounters a fundamental difficulty, as apparent simplicity reveals itself
to be an epistemological stumbling block, rather than constituting than the rock
of foundation of knowledge. Not only does the apparent homogeneity of the
entropic mix reveal uncertainty and unpredictability, but even the atom, the
god-father concept of all individualism, the simplest building block of empirical
reality since Democritus, is pulverized into complexity by quantum physics.
Noise and uncertainty are no longer the prerogative of (large or hyper-large)
aggregates, but even of simple elements: simplicity itself becomes complex.

The conversion of highly formalized concepts of information and noise to
those which are less formalized, and therefore more speculative, is not aided by
the fact that the mathematized sciences have for a long time, and certainly since
Lobachevsky, Boltzmann and Cantor, become accustomed to proceed in a counter-
intuitive way, to trust in mathematics where the senses can no longer follow. As a
result, one of the difficulties in correlating noise with complexity is the common
perception of entropy as the homogeneous murmur of ‘white noise’. Apparently
bland and predictable, noise becomes a blanket background against which a signal
appears to stand out as what is least predictable. The common perception of entropy
is one of lacking difference, of a homogeneous and therefore presumed simple and
predictable state. If we are to see how entropy is related to low probability and high
uncertainty, it is therefore indispensable to deal with this tension between counter-
intuitive and intuitive modes of reasoning as a veritable clash of cultures.
If we want to understand what is at stake in the definition of information

as negation of entropy, or negentropy, we must tackle this paradox, whereby
the greatest novelty of information, that which is maximally unpredictable,
corresponding to ‘information entropy’ in Shannon’s sense, appears to be
incompatible with the image we may have of entropy as a state that is of no
interest at all, homogeneous and unchanging like a fully diffused drop of ink
in water. This image of maximum entropy as a bland homogeneity from which
nothing new can arise, implies that nothing can surprise us: the entropic system
no longer has any physical potential to further transform itself. From this point
of view maximum entropy would appear to be perfectly redundant. Following
this logic one could thus say, arguing against Shannon’s concept of ‘information
entropy’, that it is entropy that tells us nothing new, that entropy is redundant
like the image of a homogeneous and undifferentiated whirr of ‘white noise’.
Nothing interesting will happen, nothing will surprise us in a closed system that
has reached the state of entropic equilibrium.
Raymond Ruyer insists on this point and also Gilbert Simondon argues in
this direction (Ruyer 1954, 114; Simondon 2005, 541). What is thereby taken
for granted are two things: firstly, that information can only be defined as the
negation of entropy, as Wiener did in the context of self-regulating systems with
feedback; and secondly, in Simondon’s case, that the concept of information
negates entropy, insofar as information corresponds to the novelty of what
may still happen within a system that has not yet reached entropic equilibrium.
Schrödinger’s conception of life as that which delays the spontaneously
entropic dissipation of physical processes, reinforced by Brillouin’s neologism
negentropy, has led to a conception in the humanities whereby information is

what occurs against all probability, meaning against the greater likelihood with
which physical processes are said to evolve spontaneously towards the complete
entropic dissipation of energy. This is a well disseminated view that appears to
feed on the philosophical and literary imaginary of an infinitely unpredictable
world, imagined as an open system, where anything can happen, so long as it
has not exhausted itself and reached its entropic death. This idea of an open,
complex and changeable world is contrasted with the image ‘white noise’ in a
closed system, symbol of the decomposition of all possibility.
What this understanding of information as negation of entropy implies,
however, is that the second law of thermodynamics, which stipulates that every
closed system invariably evolves towards maximum entropy, is in fact contingent
rather than necessary: it can be resisted. This, however, requires that we leave
physics behind and consider the second law of thermodynamics as a mere
image, where the necessity of a natural law is no different from the contingency
of probability: a law that is no law because it does not always apply with the same
vigour, but is contingent upon the particular type of empirical reality which may
submit to it or resist it. It makes no sense to speak of the probability with which
a system evolves towards entropy, unless there is at least one possibility that it
will not – and in this case either the second law of thermodynamics is no law
at all, or information is a miracle. It is thus difficult to maintain that the second
law of thermodynamics is contingent rather than necessary, while holding onto
it as a core element in the definition of information. What is certain, however, is
that the second law of thermodynamics applies to closed systems. Open systems,
on the other hand, which are called metastable when they maintain themselves
far from equilibrium, must answer a different set of theoretical requirements,
to which the second law of thermodynamics still applies, but requiring explicit
theoretical conversion.
Consequently, the idea that information corresponds to the negation of
entropy, commonly understood as what occurs against all probability, cannot
be applied without reservation to the notion of information in an open system.
This conversion is in fact the object and originality of theories of so-called ‘self ’-
organization, such as those initiated by Manfred Eigen’s work on the origin of
life in chemical cycles and hyper-cycles, Ilia Prigogine’s work on dissipative
structures and complexity or Katzir-Katchalsky work on chemico-diffusional
coupling in biological flow-structures (Atlan 1979, 99–128; Atlan and Katzir-
Katchalsky 1973; Eigen 1971; Prigogine and Stengers 2009). These approaches,
however, are not representative of an interpretation of the notion of information
as negation of entropy, but on the contrary render possible theories that consider
the positive role of entropy in organized systems with increasing complexity and
indeed theories like Atlan’s ‘complexity on the basis of noise’.
To associate information (negentropy) with unpredictability insofar as it
negates entropy thus relies on the conflation of multiple levels of analysis: a
system that has reached a state of energetic equilibrium, hence of entropy,
will indeed have no predictable transformation going on at the molar, visible
level. But this apparent ‘death’ of the system, which is no longer compelled by
physical potential to transform itself, belies the lack of constraint that physical
potential exercises on a system: the compelling nature of physical potential is,
as we have seen earlier, what makes the system’s transformation more rather
than less predictable. The higher the potential, i.e. the energy differential, the
more predictably a system evolves. This is why we tell children not to play with
electrical sockets. What remains maximally unpredictable, on the contrary, is
the movement, position, speed and direction of microparticles in the entropic

mix, like the whirring atoms in a gas canister.
The failure to see complexity and unpredictability in the apparently simple
homogeneity of the entropic mix comes from the fact that the macroscopic, molar
level of observation tacitly steps into the foreground, pushing the analysis of the
molecular and statistical level into the background. The macroscopic presentation
of a system in the state of maximal entropy is indeed often characterized by
homogeneity, as in the example of the diffused drop of ink. Although strictly
speaking, the term entropy refers to the homogeneity of energy levels, which
does not necessarily preclude macroscopic heterogeneity, as is the case of an
entropic mixture of water and oil, which is not morphologically homogeneous.
In any case, it is also true that in a maximally entropic closed system no dynamics
are likely to arise spontaneously (even if localized and transitory random
oscillations exist). The entropic mix appears completely redundant, like the
homogenous murmur of white noise. It appears to harbour nothing of interest,
no informational potential can be detected in it: no directional movement,
no flows, no noticeable difference. Viewed like this, it is indeed entropy that
appears redundant and predictable. This apparent simplicity, however, diverts
attention from the complexity of its microphysical presentation. The absence of
constraints, which distinguishes entropy from redundancy, and which entails
equal probability of all its possible states, is still what characterizes entropy in
mathematical terms as the greatest possible ‘freedom of choice’. In other words,
a misunderstanding can arise if we conflate the molar and the molecular level
and a fortiori if we conflate the role of entropy in a closed and in an open system.
Norbert Wiener’s definition of information as the negation of entropy does

not deny this, it merely performs a reversal of Shannon’s concept of ‘information
entropy’: information is no longer understood as what increases the ‘freedom of
choice’ in terms of probability, but as its negation. Wiener’s functional, machine
oriented concept of information as the reduction of ‘freedom of choice’ and hence
as an increase in predictability simply means that information as negentropy
is what is ultimately redundant. To say otherwise is to induce a curious short
circuit, whereby the negation of entropy, negentropy, would result in a concept of
information that expresses not frequency, repetition, or deductive redundancy,
but randomness – transforming the trajectory of a self-directing missile into
the erratic loop of a swallow. There is thus a certain amount of confusion,
when the concept of information as negentropy is seen as increasing ‘freedom
of choice’ and unpredictability vis-à-vis the state of the apparent redundancy of
entropy or ‘white noise’.
Negentropy has as its main purpose the reduction of unpredictability,

notably in the cybernetic context, where the requirement is the prediction for
instance of the trajectory of a self-directing missile, which certainly cannot
tolerate any increase in unpredictability. When we narrow down the options,
decrease the number of possible states of a system and thus increase probability
of accurate prediction, we generate ‘information’ in the sense of negentropy,
i.e. a greater prediction. If we now go on to say that the negation of entropy
serves to reduce the margin of error of prediction, by narrowing down the
number of possible scenarios, then what Shannon calls ‘information entropy’, in
opposition to negentropy, becomes even less clearly distinct from the concept of
noise, if noise like ‘information entropy’ is what makes reality less predictable.
Both ‘information entropy’ and noise must be understood as what increases
uncertainty. ‘Information entropy’ and noise thus stand together in opposition
to what the negation of entropy designates as information, which is the accuracy
of prediction and negation of possible alternative scenarios.
Entropy, finally, corresponds to the unpredictability of the micro-complexions
of the system, which is frequently represented as the random collision of
particles. The apparent simplicity of entropic homogeneity merely masks an
indiscernible complexity, which is revealed only at the microscopic level, where
there is increased uncertainty, if one were to predict the movement of its micro-
constitutive elements. Equal probability thus corresponds, despite the apparent
simplicity of ‘white noise’, to the greatest possible uncertainty in which the
observer is as to the prediction of the micro-complexions of a physical system.
Despite its apparent simplicity this uncertainty corresponds to the greatest
‘freedom of choice’ and to what Shannon calls ‘information entropy’. Apparent

simplicity thus merely masks its indiscernible complexity at the microscopic
level. When entropy is seen as ‘simple’ rather than complex, in accordance with
the appearance of ‘white noise’, it is thus a case of phenomenological astigmatism.
XIV
The Path of Despair
The two dominant interpretations of the term ‘information’, having taken

hold in the natural and human sciences, carry significantly different implications:
one values information as a specific form of uncertainty, implying increased
‘freedom of choice’ (‘information entropy’), and the other negates the latter
in favour of prediction (negentropy). Shannon’s ‘entropic ideas’, whereby the
information containing the greatest degree of novelty is the most difficult to
predict, and Wiener’s cybernetic definition of information in control theory,
where the objective is reliability and prediction, appear incompatible yet are more
often than not diffused across scientific discourse without clear demarcation.
Yet both are based on the same mathematical expression of entropy. Precisely
because they offer a purely quantitative value, their definition of information is
not determined by epistemic, aesthetic or moral values: the engineer does not
judge the quality of the information s/he transmits – whether it is a redundant
phone call of your mother-in-law or a transmission of Schönberg’s Variations
for Orchestra, only the quality of transmission. Nevertheless, Shannon’s entropic

definition of information opens a chasm of uncertainty at the heart of our
understanding of information. It allows for a positive epistemic appreciation of
uncertainty (‘information entropy’), whose generalization beyond the technical
problem of signal transmission enables a renewal of the Socratic maxim, ‘I know
that I know nothing’. Such knowledge of our uncertainty, however, is no longer
an intuition, but measures the ratio and hence limitation of what we can predict
with the greatest possible precision.
Throughout its history philosophy has been prone to a perpetual high-jacking
of certainty. This high-jacking of certainty is perhaps not as different in nature to
Shannon’s conceptual audacity as the ‘merely’ technological status of his concept
of information suggests.
Philosophy indeed never rests on its laurels for long. Radical doubt and new
concepts make established certainties redundant, or recover and transform
defunct certainties that were left for dead. Whatever certainties philosophy thus
acquires or rehabilitates in its own right (rather than, say, mathematical proof
or mystical faith), they are by default exposed to critique’s systematically erosive
force of doubt and renewal. From Greek scepticism to Hume, from De Cues’s
docte ignorance to Pascal’s wager, from Descartes’s tabula rasa to Kant’s epistemic
humility and from Schelling’s speculation about the absence of ground to Hegel’s
‘path of despair’, the cycles of liquidation of illegitimately held grounds for
certainty are inseparable from the production of new, reconfigured foundations.
Wittgenstein’s paradoxes, Feyerabend’s radical critique of method and Popper’s
falsificationism are not the end of the tether for philosophy, but the flowering of
its many cultures of systematic uncertainty. So-called ‘continental’ philosophy,
from post-structuralism to post-modernism, has been ridiculed for preferring
the deconstruction of our certainties to the plain language of constructive or
analytic objectives of philosophy. Yet both the ‘continental’ and analytical
traditions display a radical distrust of certainties, which they see as exposing
reason to the risk of naivety, if not ideology. Both the philosophers associated
with ‘continental’ philosophy and those associated with the analytical tradition
thus belong together in the abstract tableau of philosophy’s great etching of
imprudent convictions.
Looking at the diverse philosophical cultures of uncertainty, one might be
tempted to see in philosophy the vocation of overcoming dogma with ever
more recalcitrant cultures of uncertainty. If philosophy can be said to inform
us, in a way that participates in the natural and the human sciences, then it is
fair to say that Shannon’s entropic conception of information is better suited to
qualify such information as specified uncertainty than the traditional correlation

of information with certainty or even the negation of uncertainty implied in
negentropy.
However, every self-respecting philosopher must surely object that the
discipline of philosophy is, on the contrary, to put the systematic use of reason in
the service of solid foundations for certainty. What is the vocation of philosophy,
if not to surpass the unreliability of the senses, to overcome the changeability of
opinion and to reveal the unfoundedness of doxa? In other words, what else is
the vocation of philosophy, than to dispel uncertainty – just as Shannon puts
mathematics at the service of a noise-free channel of communication? From
this point of view, it would seem justified to say that philosophy doubts in the
service of certainty and its ultimate victory over uncertainty. Is not the objective
of philosophy to reduce the margin of error in reason in order to arrive at an
ultimate victory of reason over empirical contingency? To make philosophy
duty-bound towards certainty, however, would be to neglect what drives

philosophy, the divine mania of Platonic inspiration, the esprit de finesse with
which Pascal complements the reason of the geometer, Kant’s acknowledgement
of the vertigo of the philosopher before the question of ground, Schelling’s
notion of ground as anti-idea and even Hegel’s path of despair. Michel Foucault’s
famous expression that ‘error is at the root of what makes human thought and its
history’ here reveals its true relevance for the conceptualizations of noise, which
is the normative relation between reason and contingency (Foucault 1989, 22).
Foucault’s appraisal of ‘error’ in his introductory statement to George
Canguilhem’s 1966 edition of The Normal and the Pathological stands in
apparent contrast with the admiration he proclaims for a philosopher who
imparted the utmost exigency of rigour to an entire generation of philosophers
in France, for whom rigour was elevated to an ethos of philosophy. But the
counter-intuitive consequences of this rigour produced an offence to common
sense that contributed to the apparently unsurmountable distance between
French and Anglo-Saxon philosophy. The distance has since been caricatured
as opposing the presumed literary and imprecise bent of so-called ‘continental’
philosophy, to Anglo-Saxon philosophy, which in turn is downplayed by many
a ‘continental’ philosopher as wallowing in pseudo-scientific rationality.
What should in itself arouse suspicion about the geographical labeling is
that much of the so-called ‘Anglo-Saxon’ philosophy pledges critical fidelity
to a moment of great conceptual inventiveness taking place in Europe at
the turn of the twentieth century around the Vienna Circle. Continental
philosophy is deeply entwined with this critical moment in the history of
philosophy; some of its most crucial developments during the 1960s, like
for instance the journal Cahiers pour l’analyse, created an intense dialogue,
notably with Wittgenstein’s thought and also via translations, like philosopher
of logic Claude Imbert’s translation of the logician Gottlob Frege into French
(‘Cahiers Pour l’Analyse (An Electronic Edition)’ 2017; Hallward 2012a,
2012b).
French epistemology and post-structuralism’s interest in the ruptures,
bifurcations and unforeseeable turns in the historical unfolding of scientific
theories has, in turn, reached well beyond France, and beyond the theory of
knowledge, into the Anglo-Saxon humanities, taking back alleys through the
fields of literary, cultural and political theory and aesthetics. And while its
ramifications into the Anglo-Saxon theory of knowledge are less evident, they
are no less significant, if we consider for instance the importance attributed
to Thomas Kuhn’s theory of scientific revolutions, who in turn pays tribute
to French epistemology. To this day the lack of translation of key works of

what we now call ‘French epistemology’ – such as those of the philosophers
of mathematics, Jean Cavailles or until very recently Albert Lautman, into
English – testifies to the obstacles that must still be overcome before these two
approaches can dialogue in earnest.
Nevertheless, it can’t be denied that at a time when the Vienna Circle made
it the objective of philosophy to eliminate ambiguity from discourse, espousing
the formalization of language through logic and an emphasis of scientific
empiricism, an alternative approach to the philosophy and the history of the
sciences emerged in France. Without further distorting or exaggerating the
divergence between ‘analytical’ philosophy and French epistemology, it can
be acknowledged that analytical philosophy made it its objective to stabilize
philosophical discourse and its history by emulating scientific methodology
and emphasizing logic, while French epistemology espoused the unpredictable
historical unfolding of mathematics and of scientific theories more generally.
Its emphasis on the historical singularity of the ‘event’ in thought opened
philosophy to an understanding of uncertainty as constitutive of the question of
the foundation of truth.
Jean Cavaillès thus enquired into the consequences for philosophy implied
by Georg Cantor’s unmooring of axiomatic set theory from intuition. Even
the Kantian proposition of an unchangeable structure of our transcendental a
priori found itself shaken in Cavaillès analysis, and thrown into the historical
becoming of mathematics (Cavaillès 2000). Gaston Bachelard, in turn, called
for a philosophy of complexity in light of developments in quantum physics and
chemistry (Bachelard 2003), and George Canguilhem’s revaluation of the concept

of error in his study on The Normal and the Pathological (Canguilhem 1991) was
driven by the realization that reason is not only subject to norms, but is itself a
source of the norms of reason: this normativity of reason accounts also for the
unpredictability of science’s historical unfolding, revealing its necessity only
retrospectively.
From a different perspective to that of Logical Positivism, French epistemology
thereby reopens the question of ground each time reason is caught in the act
of generating new norms, thereby outgrowing and trespassing the limits of
established norms:
Almost all progress in algebra has come from the desire to accomplish forbidden
operations (negative, rational, imaginary numbers etc.). (Atlan 1979, 229;
Canguilhem 1993; Collectif 1979, 508)
This quotation is a curious case of palimpsest, where different texts and different
authors transpire and appear to speak in one voice. Georges Canguilhem here
quotes the mathematician René Thom, a Fields Medal winner known for
catastrophe theory. What is interesting is that Canguilhem cites this passage
from Henri Atlan, who himself cites René Thom’s intervention at the Colloque
de Royaumont in 1975, in his book Entre la fumée et le crystal, 1979. It is the
quotation of a quotation through which transpires a philosophy that values
non-linear thought positively as a vector for scientific striving, as an impulse for
reason’s normativity, be it in different fields of theoretical enquiry.
It is significant that Canguilhem here cites Thom in a conference paper
entitled ‘Le cerveau et la pensée’ (Canguilhem 1993), a now-famous attack on
reductionist theories of the brain and thought, which Canguilhem associates with
cybernetics and a market-driven obsession with computers. For Canguilhem
one name is emblematic for the cybernetic ‘ironmongery’ of thought:
[L]et it suffice to cite a name: that of Leonid Pliouchtch, and an emblem: that of
I.B.M. (Canguilhem 1993, 11)
It may thus appear particularly ill-fated to seek in Canguilhem’s thought a

motive to unpack the philosophical relevance of Shannon’s ‘entropic ideas’.
Shannon’s quantitative definition of information has not unfrequently been
blamed, alongside Wiener, for the reduction of thought to the mere calculating
power that stands accused in Canguilhem’s paper. The reduction of thought to
what Canguilhem calls disparagingly (quoting Thom), the ‘ironmongery’ of
thought,3 could indeed be associated with the successful application of Shannon’s
information theory to informatics alongside Wiener’s cybernetics.

To correlate the idea of noise with Canguilhem and Foucault’s revaluation
of error as the ‘origin and history of thought’ could thus appear nothing short
of heretical. The now common notion of ‘noise pollution’ furthermore evokes
the idea that noise is the acoustic waste generated by industrial development,
rather than the late flowering of the Enlightenment. Rather than pointing to
the origin of thought, which is traditionally sought in the peace and quiet of
contemplation, nothing could seem less conducive to human thought and
its history than noise, and nothing more readily the source of litigation in
increasingly densely populated urban spaces than noise (Berglund and Lindvall
1995; Bijsterveld 2001, 2003). Even Pascal, who may be credited with having
invented the calculus of probability in games of chance, which later informed
the concept of noise, saw in noisy entertainment and distraction an avoidance
of thought (Pascal 1966, 71–73). Schopenhauer even defined noise as ‘the most
impertinent of all interruptions, since it interrupts and even shatters our own
thoughts’: even a great mind is powerless if dispersed by noise, like a scattered
army or a splintered diamond it loses its power and incisiveness (Schopenhauer
1851, 517).
And yet, if two ways of reading ‘information entropy’ are possible, then
the fundamental ambiguity in Shannon’s definition of information for which
we have argued thus far, namely the ambiguity between ‘information entropy’
and noise, enables us to take Canguilhem and Thom’s defence of thought’s
normativity as an argument for the defence of Shannon’s ‘entropic ideas’. The
cultural significance of the ‘freedom of choice’ inherent in Shannon’s entropic
ideas can then be measured against Wilhelm von Humboldt’s words, quoted by
Canguilhem in this same text, according to which
[i]t [language] must therefore make infinite use of finite means (Sie [die Sprache]
muss daher von endlichen Mitteln einen unendlichen Gebrauch machen …).
(Canguilhem 1993, 26; von Humboldt 1903, VII, 98–99; 1836, 106)
Notes
1 I thank Emmanuel Picavet for his insightful comments regarding this point.
2 Only the second part of the main thesis was published in 1964 as L’individu et sa
genèse physico-biologique by Presses universitaires de France, while the first and
second parts were published together by Aubier as L’individu et sa genèse physico-
biologique and L’individuation psychique et collective.

3 ‘The traders of electronic ironmongery [or junk; French quincaillerie] would like to
make us believe that with the distribution of computers a new era will open up for
scientific thought and for humanity’ (Atlan 1979, 228–29).
Part Two
Empirical Noise
I
On the Transduction of the Concept of Noise
The new understanding of noise, brought about by the mathematical approach

to information, has transformed modern society beyond recognition, both
from non-classical mechanics and quantum physics to molecular biology and
across the natural and human sciences wherever statistical reasoning holds
sway. New communication media have sprung forth, which have in turn
generated new flows of information but also new structures of information
distribution, and therefore new dimensions for noise, ambiguity and error.
Shannon’s ‘information entropy’, but also Wiener’s negation of entropy, in
this sense carry forward the leap in modernization that had been accelerated
during the nineteenth century by mathematical physicists, notably by James
Clerk Maxwell’s theory of electromagnetic radiation and Boltzmann’s statistical
expression of entropy. The increasing energy efficiency of machines, capable
of harnessing entropy, together with the emergence of the electric grid are the
direct result of the molecular understanding of entropy in electrodynamics,
which in turn helped lay the conceptual foundations of what Floridi has
called ‘the information revolution’ (Floridi 2014) which ushered us from an
industrial to a post-industrial modernity. Noise no longer characterizes only
entropic processes related to mechanical work, but increasingly conditions
information networks, and even, if differently, the co-emergence of cognitive
labour, characterized by information overload and even the ‘mental state of
noise’.
Boltzmann’s statistical expression of entropy and Shannon’s generalization of
the concept of entropy as a problem of probability and statistics thus represent
not only the relay of a profound conceptual innovation, but also a leap in
technological innovation. The scientific and technological mastery of noise
implies novel scientific views, provoking what Thomas Kuhn called paradigm
shifts: the shift not only from classical to non-classical mechanics, but also from
the logic of transformation of energy (for instance, mechanical to electrical) to a

logic of information, amplification and control through feedback.
How can we understand this transfer of the idea of noise from mechanics to
information, and from its antithetical relation with work in classical mechanics
to self-regulating systems with noise, bearing in mind also that the general notion
of noise is derived from an aesthetic and moral connotation of acoustic events?
Gilbert Simondon, once more, provides a key for understanding this conversion
of noise from its aesthetic/moral connotations of sound to mechanics, and from
the industrial era of mechanical work to the post-industrial era of regulative
information networks and cognitive labour: the concept of transduction.
The term ‘transduction’ commonly describes the conversion of one form
of energy into another (for instance, kinetic into electrical energy) or the
transmission of a signal from one system into another (as when a signal is
‘transduced’ across the membranes of a biological cell). Simondon uses the term
transduction to describe the process of individuation, using the example of a
crystal that grows in all directions, by transducing its structure such that ‘each
molecular layer already constituted serves as a base for the layer in the process
of formation’ (Simondon 2005, 33). He also establishes an analogy between
transduction in empirical reality and transduction occurring in the domain
cognition, which consists in ‘following being in its genesis, in carrying out the
genesis of thought’ in analogy with ‘the genesis of the object’ (Simondon 2005,
34). In other words, the domain of cognition, and the fields of knowledge it
establishes, is said to come about through a process of transduction analogous
to the processes of transduction observed in empirical reality. In fact, it is this

process of conceptual transduction that enables Simondon to apply the concept
of individuation to a great variety of empirical domains, from quantum physics
to psychosocial reality.
Simondon’s epistemological appropriation of the idea of transduction is
helpful if we want to understand the transformation of this cluster of concepts
(work-order-entropy-information-regulation-organization) around noise. The
transduction of the concept of noise would imply not only the transformation
of one domain, for instance, of information theory, but also the conceptual
transduction from one domain to another. Applied to the domain of knowledge,
with its increasingly specialized and diverse areas of expertise, this means not
only that transduction is how an area of knowledge is progressively structured
according to guiding principles or concepts, but that the structuration of one field
of knowledge also transduces its guiding principles, concepts or problems, across
academic divisions and institutional boundaries, into other fields of knowledge.
Empirical Noise 95
When Simondon says, ‘by transduction, we mean a physical biological,

mental, or social operation, though which an activity propagates from point to
point within a domain’, we must therefore hear also the transduction a concept
or idea developed in one domain of knowledge into others. It is in this sense
that we can say that that Boltzmann’s statistical concept of entropy has been
transduced by Shannon, Wiener and other mathematicians, into the statistical
concepts of noise and information, which we now find across the natural and
human sciences and even in the arts.
To be sure the idea of transduction of concepts from one domain of knowledge
to another, like that of the concept of noise, is an appropriation of Simondon’s
own appropriation of the term. It is taylored here to help us to think about
epistemological resonance and epistemological noise.
The effect of Boltzmann’s statistical understanding of entropy can now be
understood in its significance beyond that of innovation in the specific field of
thermodynamics and mechanical engineering. The profound consequences it
had as a conceptual innovation illuminated the domain of new communication
technologies and progressed from there to all other domains of knowledge,
thereby transducing the concept of entropy from mechanical and thermodynamic
domain to the concepts of information and noise in communication technology
and beyond.
The transduction of the concept of noise is at once conceptual and
technological, because it goes hand in hand with the material recording and
transmitting information. Conceptual transduction requires more than just
translation of a term (noise), from acoustics to thermodynamics, or from

statistics, say, to molecular biology. There is an irreducibility between these
domains. There is a lack of one-to-one correspondence between the diverse uses
of the term noise. In other words, they are not isomorphic as two sets of axioms
could be. The consequences of thinking in terms of noise as a result differ
from domain to domain. As a consequence, the transduction of the concept of
noise could be said to scatter, more than it resonates. It hits obstacles, as reason
struggles to find a universal key of conversion from one domain of knowledge
to another.
In other words, there is no transduction in the epistemological sense without
noise. This noise increases ‘freedom of choice’ in the form of ever greater
complexity of the field of knowledge as a whole. In this sense the transduction
of concepts is irreversible, like entropic dispersion, which means that it does not
lead to a reductive one-to-one correspondence of concepts between one field
and another.
As a result, any concept and any problem that becomes relevant across
theoretical or experimental boundaries must accept a certain metaphorical
warping, unless a strictly reductive logic is applied. Indeed it is doubtful that
many domains of knowledge currently submit to a reductive logic. It seems,
rather, that many reductive theories today simply postulate that the bridging
between levels of complexity is merely a question of time and a matter of filling
in the gaps. Yet we have seen how in an axiomatic system even one postulate
can become a gateway to a radically different axiomatic system, as was the case
with Euclid’s unproven fifth postulate and the subsequent development of non-
Euclidian geometry with infinite dimensions. For any conceptual transfer to
occur without noise, and without metaphorical distortion, both fields would
therefore have to be fully axiomatized, any divergence arising from mere
postulates fully accounted for, in all its amplitude.
However, to accept metaphorical warping, and to understand that concepts
are often transduced not only in the form of abstract formalization but also in
the form of mental images, must not mean to accept the intrusion of concepts
coming from other fields of knowledge uncritically or without precision. On the
contrary, the currently germinal relation between the conceptual creativity in
the sciences and in the arts could provide a new opportunity to enable scientific
discourse to engage in metaphor and images of thought critically and creatively,
but most of all consciously. In other words, if we want to understand how the
concepts of noise and entropy have contributed to restructuring the epistemic
field as a whole, transductively and with the epistemological noise arising from
this transduction, then the necessary metaphorical latitude must be engaged

with actively rather than passively, creatively and critically rather than through
denial or convention.
II
Accidental Information, Predictable Noise
To think about the scattering that necessarily accompanies any transduction

of concepts in the form of epistemological noise, no doubt implies a return to
Weaver’s distinction between information and noise as the difference between
desirable uncertainty (information entropy) and accidental, hence spurious
uncertainty (noise, or just entropy). And yet a problem immediately arises,
when the difference between information and noise is reduced to the difference
between intention and spontaneity or accident, because this difference is almost
never as clear-cut in vivo, as it is in the situation of the engineer, who transmits
a ready-made message. Research, for instance, occasionally happens upon
the spontaneously informative as an instance of serendipity. The element of
accidental discovery implies a shifting of the boundary between the informative
and noise, which must not remain unthought. Moreover, in statistics it is not
only serendipity (which every researcher must hope for), but the systematic
use of randomness that requires us to place the accidental at the heart of the
information process. As Andrew Gelman and Cosma Rohilla Shalizi put it in

‘Philosophy and the practice of Bayesian statistics’:
[t]he statistician begins with a model that stochastically generates all the data
[…]. (Gelman and Rohilla Shalizi 2013, 8)
In other words, data must be generated randomly, even if the goal is to arrive
at an understanding of the systematic relation between variables, between
variables and their parameters, and between variables and ‘noisy aspects of the
data’, meaning its ‘contingent, accidental, irreproducible’ (my emphasis) aspects.
Bear in mind that randomness is a mathematical problem, and that the word
‘stochastic’ was introduced in early probability theory from the Greek, meaning
to conjecture, to take aim or guess. The definition of noise is thus only slightly
distinct from stochastically generated data insofar as the stochastic implies
taking aim, while noise implies missing the target. The accidental thus partakes
in the way a research question is formulated against the backdrop of uncertainty,

in the design of a statistical model or scientific experiment, and in the testing of
models or theories.
The distinction between information and noise is a clear task only for the
telecommunications engineer: here the problem is only that of fidelity of
a message, after its selection has taken place and only once the decision has
been made to transmit this message, whatever it may be, as information. The
difference between information and noise derives from this decision alone, as it
does from the selection of one scientific model over another. It is this selection
that designates something, anything as far as information. It is only on the basis
of an always already made selection that the telecommunications engineer or
any other statistician can root out any increase in the message’s entropy during
transmission, which can then be discarded as accidental, i.e. as noise. What
remains in the dark, if we consider only the problem of noise in the channel of
communication, is thus the part that the accidental may play in the decision to
select and designate something, in principle anything, as information.
The origin of noise may be an external perturbation (afferent noise) or noise
may generated by the thermal agitation of the transmission itself (as efferent
noise). The engineer is consigned the task of ensuring the safe passage of
information through the channel of communication, warding off afferent noise
and keeping information free from corruption through efferent noise. There
is noise generated by the transmission of information, either produced by the
thermal vibration of atoms in the conductors (Johnson-Nyquist noise), or the
noise already intrinsic to the means of measuring of discrete electric charges, or

of photons in optical devices (shot noise or Poisson process).
In other words, information is always already thrown into the noise of its own
presence, but also that of objects surrounding it (not least the engineer), of earth
and even of celestial bodies and black holes, each emitting a frequency spectrum
characteristic of its own electromagnetic radiation – its black body radiation (so
called not because it is black, but because the eye cannot perceive colour in low
intensities of thermal radiation, unless the temperature is elevated).
All of the above forms of noise participate in a pool of so-called Gaussian
noise, meaning that this ‘general’ noise, is characterized statistically by its
‘normal’ distribution, also represented by a so-called bell curve of the average
of a large number of random Gaussian processes. It is worth dwelling on this
slightly technical point, lest we miss the paradox it implies for our understanding
of noise. Noise (accidental, non-necessary, increasing uncertainty and ‘freedom
of choice’ in probabilistic terms) is also normal and ubiquitous, often subject to
Empirical Noise 99
a highly predictable statistical form, called a normal distribution. So predictable,

in fact, that telecommunications can mimic its uniform distribution across
the frequency band, and use what is called additive white Gaussian noise
(AWGN) as a basic noise model for the underlying behaviour of a system, so
as to later distinguish phenomena like fading, interference, dispersion or other
chaotic or unpredictable (non-linear) events. What at first presented itself as
non-reproducible stochasticity, for the communications engineer noise on the
contrary becomes characteristic for the underlying behaviour of a system. The
least we can say is that the relation between noise, stochasticity, the accidental
and the predictable is far from simple.
In a course entitled ‘Uncertainty in Measurement: Noise and How to Deal
with It’, Peter Scott, Emeritus Associate Professor at the University of Santa Cruz
Physics Department, explains the difference between noise, understood in the
empirical sciences as a statistical term, meaning random error in the sense of
statistical uncertainty, and as a systematic error in measurement, such as it can
arise, for instance, from a mis-calibration of a metre or from some physical effect
not taken into account. The idea of noise in physics applies when a quantity
measured in an experiment involves a large number of random processes that
can only be approximated statistically. The pressure of a gas, for instance, results
from summing the random motions of a very large number of molecules. These
random motions are likely to be distributed normally, according to the Central
Limit Theorem, which states that
if we have a number of random variables, say u, v, w, …, and if we form a new

variable z that is the sum of these (z = u + v + w + · · ·), then as the number of

such variables becomes large, z will be distributed normally, i.e., described by
a normal distribution, regardless of how the individual variables u, v, w, … are
distributed. (http://physics.ucsc.edu/~drip/133/ch2.pdf)
Almost all measurements that are normally distributed around some average
value (over 99 per cent) will lie within three standard deviations of the mean,
giving this normally distributed from of noise a great advantage of predictability.
However, Scot warns that not every data point falls within the classic bell curve
and that ‘many are non-normal – too far away from the mean – A voltage spike
or a vibration caused by a truck passing by, for instance […]’ (http://physics.
ucsc.edu/~drip/133/ch2.pdf).
Such extreme, non-normal values are no doubt closer to our intuitive
understanding of noise in general discourse, derived from the experience
of an unwanted or startling sound. What is clear, however, is that we have a
relatively solid basic conception of noise in physics, tied to the normal statistical
distribution of random events (represented by the characteristic Bell curve),
which does not exclude the definition of other forms of noise that can be defined
as special cases, obeying different mathematical parameters than the normal or
Gaussian distribution.
In short, information is lifted out of an (in principle infinite) pool of
accidental frequencies, of which some are ‘normal’ and others, consequently
anomalous. The fundamental problem in the definition of noise will thus
be, at what point anomalous or extreme values in a statistical distribution
call for a change of perspective (or model) from an approach that privileges
the normal distribution of noise and thus de-emphasizes the importance
of extreme values as rare and insignificant, to an approach that calls into
question a given model, by emphasizing the importance of values that by far
exceed the norm.
The history of science shows how the paradigm of Newtonian physics buckled,
amongst other things, under anomalies in the measurement of Mercury’s orbit,
the smallest and innermost planet of the solar system. It was only when these
anomalies could no longer be contained in a normal distribution of errors
(and hence disregarded) that the possibility of an alternative paradigm could
arise, eventually enabling the emergence of relativity theory. The power of the
dominant Newtonian model to discard the relevance of extreme values in favour
of the normal distribution of the ‘law of errors’ revealed itself to have become
an epistemological obstacle, delaying the realization that a profound change in
theory was called for. Thomas Kuhn later theorized the subsequent change of
perspective as a paradigm shift, which he defined as a revolution of consensus in
the scientific community.
Statistics, as a theoretical and applied discipline, has since created a highly
complex arsenal of mathematical approaches, including the combination of
models and their systematic testing, ranging from Bayesian data analysis to
complex stochastic models and simulation-based model checking. The problem
is that the promiscuity between this highly specialized discipline and the hunger
for statistical ‘evidence’ in general discourse risks a continual return, in general
discourse, to the epistemological obstacle physicists faced when the highly
coherent Newtonian system had to be overhauled. In other words, the public
appetite for the ‘empirical evidence’ accorded to statistics risks falling prey to
an underestimation of the need for testing and revision of models. As a result,
information is deemed acquired, and noise can be more easily discarded as
meaninglessness. As Andrew Gelman and Cosma Rhilla Shalizi put it:
Empirical Noise 101
All models will have errors of approximation. Statistical models, however,

typically assert that their errors of approximation will be unsystematic and
patternless – noise. Testing this can be valuable in revising the model. (Gelman
and Rohilla Shalizi 2013, 20–21)
Gelman and Rohilla Shalizi argue that the central task of statistical analysis,
namely the search for consequential errors by means of ‘severe’ testing, is not
only a problem specific to improving statistics, but a philosophical problem.
The necessary testing and re-calibrating of statistical models leads to a new
understanding, notably of Bayesian statistics, which can no longer be seen as a
merely inductive form of reasoning (where the inflow of new data continuously
alters the distribution of probability), but where data and concepts are in the
constant dialogue of a hypothetico-deductive mode of reasoning: where a
hypothesized model makes certain probabilistic assumptions, from which
other probabilistic implications follow deductively, via a detour of testing and
recalibration. In other words, conceptualization of a model, deductions and their
recurrent critical analysis, are at least as important as the incoming flow of data.
For us, the philosophical relevance of the necessary testing of models is closely
related to the problem noise. Noise, as a result, becomes a highly articulated
concept in the context of statistics, with its own frame of debate. Information,
in turn, remains much more difficult to define. The criterion of intention must
be disencumbered from bias and loosened up, so as to allow for the accidental
as a possible reason to shift the intentional framework. The distinction between
what is accidental (noise) and what is intentional (information, data) thereby
becomes the Gordian knot of a hypothesis and its recalibration: whereby the
information of a poorly calibrated model can degenerate into noise, while what
was a priori discarded as noise may become information. As a consequence,
even the residual or negative definition of information (negentropy), which
implies that information is simply what is left after noise is discarded, becomes
a highly restless and fragile proposition.
III
Ready-Made Information
The distinction between what is accidental and what is intentional, which

still plays a part in Weaver’s definition of noise as ‘spurious uncertainty’ (as
opposed to ‘information entropy’ as desirable uncertainty), thus appears to be
clear-cut, at least in principle, only in a highly formalized empirical setting, and
only provided that the framework of its theoretical conventions and its fitness
for purpose are accepted as a starting point. In turn, the telecommunications
engineer deals with the distinction between ‘information entropy’ and noise
as the consequence of a ready-made decision (the selection of a message to
be transmitted as information). The principal task is to ensure safe passage of
information – whatever its content – through noise: neither the selection of
something to be transmitted as information, nor its evaluation upon receipt as
informative, comes into the equation.
Shannon explicitly defers the problem of the decision, according to which
information is evaluated as informative and hence distinguished form noise. By
separating the probability of its occurrence from the purpose of transmission

and from its subsequent evaluation, Shannon has earned great criticism for his
definition of ‘information entropy’. The first criticisms levied against this purely
quantitative definition of information was, consequently, that it fails to take into
account the heuristic criteria by which we qualify something as informative.
For us, this suspension of the heuristic question is, on the contrary, the
greatest epistemological advantage of Shannon’s definition of information. It is
epistemological in the sense that – taken outside of the technical problem of
noise in the channel of communication – it allows us to think about the shifting
boundary between information and noise in the emergence of knowledge and
its transformation.
While most of what we ordinarily call noise in the statistical or scientific
sense can be categorized as relatively predictable, according to its ‘normal’
statistical distribution, some events, as Scott warned, cannot: extreme or ‘non-
normal’ values, chaotic or non-linear events can become highly informative, if

we consider only the probability of their occurrence, untethering this form of
extreme noise from pre-established criteria of pertinence that would discard it
as irrelevant, because too far from the norm.
By deferring the choice upon which the distinction between information
and noise is based, Shannon renders a hiatus explicit that is tacitly smoothed
over, when information is viewed as already endowed with a signification, as an
already given datum or, in the plural, data.
The artist Marcel Duchamp revealed this implicit conflation between what is
simply given and what is already selected as meaningful, by radically questioning
what is acceptable as a work of art. By presenting an open-entry sculpture
exhibition in 1917 with a urinal signed with the pseudonym R. Mutt., Duchamp
created a situation that rendered the implicit assumption of criteria of pertinence
explicit. His riposte to the removal of the piece from the exhibition and to the
indignation it caused, is here reproduced in his own words, as a brief question
and answer that points to the issue that interests us when thinking about the
difference between information and noise, namely the issue of freedom of choice
in the artistic process, and in its reception:
They say any artist paying six dollars may exhibit.
Mr Richard Mutt sent in a fountain. Without discussion this article
disappeared and never was exhibited.
What were the grounds for refusing Mr Mutt’s fountain:
1 Some contended it was immoral, vulgar.

2 Others, it was plagiarism, a plain piece of plumbing.
Now Mr Mutt’s fountain is not immoral, that is absurd, no more than a bathtub
is immoral. It is a fixture that you see every day in plumbers’ show windows.
Whether Mr Mutt with his own hands made the fountain or not has no
importance. He CHOSE it. He took an ordinary article of life, placed it so that its
useful significance disappeared under the new title and point of view – created
a new thought for that object.
As for plumbing, that is absurd. The only works of art America has given are
her plumbing and her bridges. (Harrison and Wood 1999, 248)
By separating out the act of selection from the attribution of signification,

Duchamp demonstrates the normativity of the artist – insofar as he can set
new norms – and allows these to compete with the normative role of the art
institution – insofar as the latter has a mostly conservative function of enforcing
and reproducing existing norms. Duchamp thus diverts the attention from the
Empirical Noise 105
content and form of the work of art as datum, to the normativity that is in play in
the production/selection, presentation and reception of a work of art.
Duchamp’s ready-made thus enables us to question the problem of normativity
which, albeit in a scientific and technological context that could not seem more
remote from Duchamp’s concerns, also Shannon enables us to single out: what is
transmitted as ‘information entropy’ is transmitted as raw material, isolated from
meaning and value; both the choice that triggers the transmission of something
as information and its evaluation upon reception imply a normative process.
Only once these two normative stages in the information cycle are isolated can
they become apparent as problems in their own right.
Shannon thus illuminates what otherwise remains obscure in the
emergence of something new as significant or otherwise informative: namely
the normative process that must precede and conclude the distinction
between information and noise. Shannon’s definition of ‘information entropy’
as ‘freedom of choice’ thus carries conceptual, theoretical and more generally
cultural relevance well beyond the question of the purely mathematical
evaluation of probability.
What appears to offend common sense in Shannon’s definition of information
is the brutality with which ‘information entropy’ is presented, like Marcel
Duchamp’s urinal, as a brute fact, unprocessed by interpretation, denuded
of signification, and of its greatest value when at its most unpredictable. Yet,
by suspending the question of interpretation and evaluation, Shannon, like
Duchamp, also leaves open the possibility that the constraints of interpretation
may change, that what may be, under current interpretational constraints,
discarded as useless, may on the contrary become highly relevant, if different
rules of interpretation are applied. This normative aspect of the selection and
evaluation of a contingent fact can now be posed as a condition of possibility of
signification.
Shannon’s conceptual audacity was to treat information as a raw fact, in all
its ontological and epistemological nudity. Devoid of Boltzmann’s reference to
physical reality, but also of Wiener’s utilitarian injunctions regarding organization
or purpose, Shannon’s ‘entropic ideas’ help us to rethink information as a pure
event of which we know nothing but the improbability of its occurrence.
What information philosopher Luciano Floridi has called the most profound
epistemic upheaval since the invention of the Gutenberg Press, must therefore
be considered not only in light of the profound impact of new communication
technologies (NTCs) facilitated by Shannon’s contribution, but also in light of its
truly philosophical audacity (Floridi 2014).
In light of this normativity, which characterizes the boundary between

information and noise, it becomes an imperative to rethink the epistemic
value also of noise. As virtual ‘freedom of choice’, understood as a speculative
‘maximum’ of uncertainty, noise is that against which the profile of relative
information is drawn as a ‘relative’ uncertainty. The difference between
information and noise thereby loses the aspect of its Manichean dualism,
characterized by mental images of light and dark, rational and irrational, order
and disorder, life and death.
The relation between information and noise instead becomes a normative
line drawn through the contingency of the empirical world, but also through
the contingency of our mental constructs. If a cut persists, it is thus no longer
pre-established and guaranteed by a moral fault line, but animated by what
philosopher of mathematics, Jean Cavaillès, called ‘successive fractures of
independence’. Whatever imposes itself as the ‘new’ norm for thought thereby
claims for itself whatever imperial profile of the norm preceded it, yet one it
is bound to concede to whatever will come as the next fracture. In a logic not
unfamiliar to the logic of the artistic avant-garde, Cavaillès sees the rhythms of
reason’s unfolding as cadenced according to these
successive fractures of independence, [which] each time detach from what
precedes it the imperial profile of what comes afterwards, necessarily and in
order to surpass it. (Cavaillès 2000, 42)
In light of this conflictual and dynamical rhythm, also the epistemological line
of fracture between information and noise is mobilized, becomes impermanent –
in the empirical sciences even more so than in mathematics. Yet, as we have

seen with regard to Lobachevsky or Cantor, even the history of mathematics
reveals a continuously shifting ground of the certainties of established forms of
knowledge.
Shannon’s very open definition of both information and noise as a pure
measure of uncertainty, thus enables us to think of epistemological normativity,
meaning the emergence and perseverance of new norms of knowledge, as the
act of separating general uncertainty into two kinds: uncertainty we estimate
to be of potential use and uncertainty which we deem to be ‘spurious’, whose
pursuit would be ‘futile’. Shannon thereby enables us to ask, as Duchamp did:
on what grounds do we cut the empirical manifold in two, into what counts as
information (or art) and what can be discounted as noise? While the engineer is
not asked to assess the message s/he transmits, the scientist, the statistician and
even the financial trader must, like the artist, continuously perform a normative
Empirical Noise 107
act in order to draw and redraw the boundary between relevant and irrelevant
empirical contingency, testing and recalibrating running assumptions. Only on
this condition can ‘a new thought be created’, to use Duchamp’s expression, for a
reality that is never entirely divested of uncertainty.
IV
Cosmic Background Radiation1
The background noise of the unfolding universe can perhaps rightly be called
the archetype of noise. As a consequence of its discovery the classical idea
of a cosmos in perpetual equilibrium, alongside the Pythagorean idea of a
harmonious ‘music of the spheres’, no longer provides the stable epistemic space
within which Ancient Greek philosophy hoped to detach itself from myth and
religion. The tranquil firmament for our ideas about the world was irreversibly
lost by the chance encounter of isotropic rays during the 1960s. Arno Penzias
and Robert Wilson’s discovery of isotropic rays, and their identification as
background noise of the universe, roused the classical idea of cosmic order and
equilibrium from its slumber with the ebbing murmur of a catastrophic origin
of the universe.
Cosmic noise henceforth becomes a reminiscence, of Proustian proportions,
of a swarming ‘microgenesis of cosmogenesis’. In its light, or rather in its
soundscape, the classical idea of cosmic equilibrium and our conceptions of
order, in nature and in the structures of our understanding, radically changes.

The perpetuity of cosmic order is henceforth thrown into a hypothetical scenario
of ongoing genesis and perpetual metamorphosis: the stability of the cosmos
itself is bent into the meta-stability of cosmic becoming (Morin 2008, 77).
Its accidental discovery in 1978, initially as a persistent noise in signal
transmission, serves here to exemplify the transition from the empirical problem
of distinguishing ‘information entropy’ form noise, to the epistemological
problem of the shifting boundary between the relatively uncertain and the
absolutely uncertain. For, what at first presented itself to radio astronomers Arno
Penzias and Robert Wilson as unwanted noise, interfering with the reception
of chosen signals, subsequently revealed itself to be one of the most valuable
sources of information in modern astronomy. The two scientists pursued the
source of an irreducible radio noise that eventually paved the way for modern
theories of the evolution of the universe:
The cosmic microwave background radiation, considered a relic of the explosion

at the beginning of the universe some 18 billion years ago, is one of the most
powerful aids in determining these features of the universe. (Wilson 1978, 463)
The information potential of noise was of course not recognized at first and
was, to begin with, treated only as perturbation of transmission, as nothing
but noise. The task therefore was to identify the source of noise in order to
eliminate it from the desired signal transmission. Yet this particular persistent
background noise could not be attributed to any of the known sources of noise.
What subsequently revealed itself to be of utmost importance for astronomy, for
our understanding of the universe and our place in it, was thus at first an element
of the greatest possible uncertainty: not a known source of noise, but pure noise.
Penzias and Wilson’s persistence and openness to noise was rewarded with a
Nobel Prize.
This leads to an understanding of noise of literally cosmic proportions:
everything that is not selected, is in principle considered as noise, the earth’s
atmosphere, no less than the thermal radiation of all things and people around
us, the noise of our activity, of our observation interfering with what is being
observed.
Rather than being generic, noise is thus always already stratified according
to its different identifiable sources. Very rarely do we encounter a noise as pure
noise:
To measure the intensity of an extra-terrestrial radio source with a radio
telescope, one must distinguish the source from local noise sources – noise from
the radiometer, noise from the ground, noise from the earth’s atmosphere, and
noise from the structure of the antenna itself. (Wilson 1978, 466)
In other words, what we consider to be information must be carved out from

noise, which is expected to add nothing but ‘spurious’ uncertainty. But noise,
in the process, must be attributed to a source or at least correlated with a
known variable, and thus dismembered, dissected into to identifiable and non-
identifiable sources of noise: contingent though it may be, noise thereby specifies
itself as a source of information about the empirical world.
The analysis of what at first appeared as one mingled background noise and
its attribution to different sources thus leads to an understanding of noise that
is stratified, separating out the predictable sources of perturbation of signal
transmission, which have already been incorporated into knowledge, from
noise arising from an unknown source. In this case, the signal interference
that could be attributed to no known source, that represented the highest
Empirical Noise 111
possible degree of uncertainty, is what became the object of greatest interest

and the source of the most innovative information about the history of the
universe.
In La Méthode, Edgar Morin underlines the momentous transformation
this implies for the classical conception of order and of the universe as the
cradle of life, for Kepler’s ‘steady state’ and Newton and Laplace’s mechanical
clockwork universe. His fast-paced summary of the profound epistemological
upheaval during the past century and a half contrasts the ‘scenario’ of
cosmogenesis with the classical conception of cosmic order: the hypothesis
at the origin of the universe is that of a photon cloud dilating at a counter-
intuitive temperature of K, before granulating into electrons, neutrons,
protons as it cools: collisions at still formidable temperatures force the
nucleo-synthesis of deuterium, helium and hydrogen. Cosmic expansion
is henceforth wrought with gravitational dynamics that put the expansive
cloud under tension and even fissure it under the pressure of regional self-
amplifying density, from which a ‘schismatic morphogenetic’ process ensues:
the cloud ‘cracks, dislocates’ into proto-galaxies. These proto-galaxies in
turn fissure and break up into gravitational assemblages, accelerating the
localized growth of density to the point where the collision of particles
provoke a nuclear chain reaction: in a titanic explosion, contained only
by the equally titanic force of gravitational pull, a star is lit: a cog in the
gravitational clockwork of the galaxy, industriously producing heavy matter
from atoms that are forged in the kiln of the star, before it finally implodes
or explodes (Morin 2008, 77).

From the first principle of thermodynamics, which postulates the
indestructible nature of energy, i.e. that there can be no loss of energy, to the
second principle of thermodynamics that subjects the conversions of energy
from mechanical to electrical or chemical energy to calorific degradation, the
emphasis shifts from perpetuity of energy to the irreversible loss of the universe’s
potential for transformation. Implied is a dialectic of order and disorder that now
enfolds the classical idea of perpetual cosmic order in the idea of its becoming:
from its chaotic beginning to its inevitable entropic death, based on Clausius’
assumption that the universe can be seen as a closed system vowed to entropic
equalization of energy levels.
The eschatological scenario of entropic death of the cosmos, however, runs
into an aporia: the observable genesis of order. Galaxies form and even life
emerges against the predicted entropic dissipation of energy. The negentropic
capacity of the cosmos to pull itself together at all levels, from gravitational
order to biological organization, defies the idea of the cosmos as a closed system
vowed to entropic death.
The pendulum swings back, from the catastrophic scenario of original chaos
and inevitable entropic exhaustion, to order as the negation of entropy, implying
the marginalization of disorder as background noise. The loss of the classical
ideal of the perpetual cosmic machine is eventually compensated, not only by
life but by our understanding of negentropic processes, and epistemologically,
by the mastery of entropy in statistical mechanics. The maximization of
thermal efficiency in manmade machines reverses the reversal of order that the
understanding of entropy had first brought about, and re-installs the reign of
order, this time as the mastery of entropy. A refined understanding of metastable
systems eventually enables the articulation of entropy and the emergence of
structure. And quantum physics, finally, provide the mathematical formalization
of an uncertainty relation constitutive of matter itself.
Cyberneticians and those working on Shannon’s information theoretical
concepts begin to place entropy and noise at the heart of emerging theories of
complexity, amongst which, to name but a few: the physicist and cybernetician
Heinz von Foerster who discovers the principle of order from noise, based
on the recognition of initial constraints; the mathematician, physicist and
computer scientist John von Neumann, who introduces an understanding of
self-reproducing ‘natural’ automata, functioning with disorder, and of course
Henri Atlan, who incorporates Shannon’s definition of information in light of
Ashby’s law of requisite variety into a theory of biological complexity on the
basis of noise.
Disorder, quantum uncertainty, entropic diffusion of heat are no longer seen
as the mere negation of order, but interweave constraints and determinism with
the indelible singularity of evental conditions. The emergence of structure can
now be thought via the theory of metastable systems such as Prigogine studied
them. The phenomenon of high molecular cooperativity under the effect of
entropy, for instance, comes to explain the emergence of order from entropy in
the form of hexagonal convection cells, observed by Bernard and generalized
by Prigogine in his theory of systems far from equilibrium. René Thom’s
catastrophe theory, in turn, provides a deterministic mathematical framework for
thinking catastrophic bifurcations in dynamical systems, aligning mathematical
rationality with the non-linear, and resulting in the conceptual monstrosity born
from the necessarily unpredictable: deterministic chaos.
Empirical Noise 113
The classical opposition of order and disorder thus enters into a dialogue, an
increasingly intimate cooperation of both notions. There is in fact a dialectical
torsion of order and disorder, which, as Edgar Morin shows in La Méthode (Morin
2008), corresponds to the irreversible transformation and complexification of
these notions, to a refinement of an initially coarse opposition between order
and disorder and to the increasing interpenetration of both.
V
Noise in the Gap between Narratives
Morin’s epic of cosmogenesis and of the parallel complexification of our notions

of order and disorder resembles, by Morin’s own account, an ‘adventure story
with chance, suspense and drama’ (Morin 2008, 76). It is not this narrative of
astronomic proportions, however, that is the item of interest for Morin, but
the transformation of concepts and theories, forced by the breakdown of the
classical notion of order.
In one essential way Morin’s dilemma of drawing the line between science
and fiction is thus analogous to the one faced by the writer Hans Magnus
Enzensberger in The Short Summer of Anarchy (Enzensberger 1977), which
relays Spanish revolutionary Buenaventura Durruti’s last summer. Enzensberger
draws on a myriad of sources and of causal explanations, whose inequalities
and tensions are left to create fissures in the dominant narrative. The peculiarity
of Enzensberger’s account of Durruti’s controversy-surrounded death, is that
he creates a discursive fugue, where historical sources, documents, interviews
and accounts of Durruti’s contemporaries appear to repeat a common theme,

threading together the same events, yet always from different viewpoints,
resonating together to produce a riotous yet confluent discourse.
Enzensberger’s reliance of a great array of sources, however, does not serve
the purpose of consolidating historical facts by synchronizing the sources.
He even derides the anxiety of fellow historians who fear a degeneration of
historical discourse from historical fact to adventure novel. The peculiarity of
Enzensberger’s approach to fiction is analogous to Morin’s Méthode, in that the
seriousness of the historical account lies not where diverse narratives classically
converge into a powerful discourse, but in revealing the friction that bristles
between these narratives:
The contradictoriness of forms only announces the fissures that run through the
material itself. The reconstruction resembles a puzzle, whose pieces do not join
seamlessly. It is on the joints of this picture that one must dwell. Perhaps it is in
these that the truth lays in view of which, unbeknown to the narrators, there is
narration. (Enzensberger 1977, 14)
History as a magisterial discourse, on the contrary, even the history of the

universe, as Morin shows, is a tune that emerges from synchronizing the
heterogeneous sources and discourses. The historian, no less than the scientist
or statistician, is expected to normalize and synchronize the singularity of its
sources. The authoritative tone with which a final narrative is traditionally
presented as objective implies that we must pass a blind eye on the self-interest
of each narrator, who uses, selects and shapes the raw material of empirical
contingency into a streamlined informational flow. To this Enzensberger
objects:
What he finds is not mere ‘material’, unintentionally dumped, in pure objectivity,
untouched by human hands. On the contrary, everything that you see here has
gone through many hands, shows signs of use. (Enzensberger 1977, 16)2
The consolidation of a myriad of narratives into one discursive flow thus masks
the murmur of inconsistency, the loss of precision that, like entropy, makes
every process, also the discursive process, irreversible. Noise is relegated to the
margins of scientific discourse, dispensable and finally cut from the narrative as
mere error or imprecision.
As a consequence, we can say that the form and hence limits of every scientific
narrative or model are drawn against the backdrop of noise, cut out and lifted
from noise with the surgical precision of a theoretical prism. Yet by cutting this
narrative or model out from the empirical manifold, knowledge generates an
excess that is then discarded as noise: a left over discarded on the other side of
this epistemological cut through uncertainty. As mere off-cut, noise becomes
the refuse, generated by the process of information itself – that which falls by the
wayside, and whose negative connotation as epistemological refuse recalls the
Germans call Abfall, and the moral connotation of apostate.3 This implicit moral
connotation is what perhaps explains the anxiety of seeing scientific discourse
contaminated with fiction.
The excess that comes to haunt information as alien, as threat and as
eliminable presence – epitomized by the metaphor of noise as parasite in
the channel of communication – in fact reveals itself to be a function of the
selection itself. If this selection establishes a norm, by drawing a critical line
through empirical contingency, then noise is that which does not conform to
the norm, implying a certain threat of subversion of the norm. To quote Jean
Cavaillès once more:
Empirical Noise 117
[T]he empirical manifests its essential fragility in the unpredictability of its

characters, in the illegality of sorts that it harbours […]. (Cavaillès 2000, 19)
Experience, but also the inferences we draw and the deliberate testing of our
models, compel us to constantly redraw the line that separates information from
noise, to reposition the cut according to which we select useful uncertainty as
information and de-select ‘spurious’ uncertainty as noise. In the process we
may be required to attribute value and accept as legitimate what was previously
devalued and considered illegitimate.
Analysing how the history of the natural sciences has been jolted into action
more than once by phenomena that were previously discarded as marginal
perturbation, how many an experimental perturbation has led to the recasting
of scientific theories, Bachelard concluded that
the very idea of perturbation […] will have to be eliminated eventually. One won’t
speak any more of simple laws that are perturbed, but of complex and organic
laws that are sometimes touched by certain viscosities, certain effacements. The
previous simple law becomes a simple example, a mutilated truth, an unfinished
image, a sketch copied onto a chalk board.
As the distinction between information and noise reveals itself to be subject to

historical change, so the line we draw between information and noise becomes,
like an artist’s sketch, a fine overlay, a vibrant impression transpiring the gist of
the movement of thought itself. As the territories of our knowledge are redrawn
according to new lines that divide information from noise, and the boundaries
between theories shift or disappear, also the rules according to which we
distinguish information from noise are challenged by experience or concept,

and rewritten.
VI
Noise in Finance
Finance provides a setting that demonstrates, like few others, how valuable it
can be to look at the metaphorical latitude involved, when the idea of noise is
converted from a scientific concept to a schema of thought. Here the idea of
noise is generalized in relation to the investment and regulation of financial
transactions – without forgetting its origins in the common understanding of
noise as unwanted sound or disturbance. The financial krach of 2007 and 2008
revealed itself to be tied to a semantic, rather than just financial speculation.
The sophistication of financial products, and of mathematical models of
prediction permitted a new esotericism to arise in financial discourse, deferring
critical analysis of the internal dynamics of the financial sector to the presumed
expertise of those in charge. And yet, the subprime crisis erupted apparently
without warning and even in direct contrast with official prognostics, such as
those famous last words of the International Monetary Fund (IMF), according
to which ‘global growth should remain vigorous in 2007 and 2008’.
The most common definition of noise in finance is that of unexplained

price and volume fluctuations in general equity trading. It can be described by
statistical models that are specific to time scales that vary from monthly or longer
time scales, (often modelled as a diffusion process with a drift), to a ‘tick-by-tick’
resolution, following every up- or downtick in the price or volume of a market or
security price, (unpredictable beyond a few minutes).4 The time scale is essential
to understanding the dynamics of trading. As Eisler, Kertesz and Lillo argue, ‘the
qualitative picture changes dramatically when one moves down to the resolution
of individual transactions’. As a consequence, for example, the notorious high-
frequency trading strategies, remain a challenge for mathematical modelling
despite their notoriety (Eisler, Kertesz and Lillo 2007).
Noise and its conceptual amplification into stock market volatility is in part
blamed on such automatically executed logarithmic program trades, which are
often designed as a hedge, in order to counterbalance the risk of potential losses
due to market weakness. The algorithm triggers the selling or buying of large
quantities of securities, when a predetermined price is reached. As targets are
hit, the liquidation of large volumes precipitates further price falls, which may in
turn trigger other stock liquidations, doing so at the precise moment in which
the same programs also stop buying. This phenomenon was largely blamed for
the 1987 crash, when stocks dropped by 22 per cent in a single day. Due to the
risk associated with algorithmically induced market volatility, exchanges now
limit the time window for program trades.
The second characteristic commonly associated with noise in the stock
market is irrationality. The so-called noise trader is defined variously as an
amateurish investor, the gambling type, whose decisions are based on feelings
rather than either fundamental or technical analysis. In other words, the noise
trader responds to price fluctuations with the gambler’s instinct. Rather than
basing decisions on knowledge of the presumed intrinsic value of a security, he/
she ignores the fundamental macroeconomic factors, such as general economy
and industry-specific conditions, and microeconomic factors, such as company
management and financial soundness. Simply put, noise trader designates
someone who apparently ignores the fundamental data necessary in order to
assess whether a security’s market price is over- or undervalued.
The noise trader is generally believed to also lack the technical knowledge,
which is associated with the purely mathematical forecasting of a security’s stock
market volume and price movements, (often evaluated in and of itself on the
basis of recognizable patterns, without taking fundamentals into account). This
portrait of the noise trader as ignorant and purely impulse driven, however, is in
some need of adjustment, as Alex Preda’s analysis in Noise: Living and Trading
in Electronic Finance reveals: increasingly tech-savvy and mathematically literate
traders are participating in computerized trading (Alex Preda 2017).
In any case, the image that persists of the noise trader is that s/he is incapable
of distinguishing patterns and trends from random fluctuations or noise.
Unlike the institutional investor behind much program trading of large stocks,
it is the large number of individual noise traders that is seen to contribute to
market volatility, by introducing irrationality and accidentally amplifying
random fluctuations into trends. The noise trader is thus not only subject to
the alea of economic factors and random market fluctuations, but also seen
to play a role in its amplification into a trend. As a consequence, the noise
trader stands out, vis-à-vis the technical specialist, or the economist handling
economic fundamentals, when consecutive financial crises are blamed on
reckless speculative behaviour.
Empirical Noise 121
Blaming the noise trader for agitation of the financial markets, in fact,
goes back to the origins of mathematical finance in nineteenth-century moral
philosophy. Jules Regnault’s Calcul des Chances et Philosophie de la Bourse,
published in 1863, first put the question of the financial markets’ morality on
a purely mathematical ground (Regnault 1863; Walter and de Pracontal 2009).
In it, Regnault opposes the natural and just self-regulation of the markets to
anomalous market dynamics, blamed on the irresponsible behaviour of
speculators agitating the markets and perverting the natural course of events.
One could say, to use von Foerster’s expression, that Regnault paints the financial
markets as the spontaneous emergence of order from noise, in which the healthy
and just self-interest of the sound investor, who helps to build the economy, is
opposed to the spurious speculator’s fall into personal and collective ruin, which
is at the same time a moral fall from grace:
The stock market is the temple of modern society: it is here that all the great
interests of an eminently positive and industrial century are destined to converge;
but the stock exchange is also the official sanctuary of gambling, where fortunes
and existences founder. (Regnault 1863, 1)
Regnault goes on to oppose the ‘real’ causes of market variations in offer and
demand to the ‘sterile movements’ of agitation that result from pure speculation,
characterizing this difference by a strongly phrased moral contrast:
There are thus two kinds or varieties of speculators: one […] a real parasite on
genuine speculation […] is based on ignorance, cupidity, satisfaction of brutal
appetites, all passions that engender and characterise gambling; it is shameful
and a disgrace.
The other […] has the talent to create, edify, transform, having as his
only goal the utility for the common good; he corrects the exaggerated
movements, which blind trust or senseless panic produce in the stock
exchange, offers credit and maintains a constant equilibrium between the
diverse values according to the utility of their products […] he cannot be
praised and encouraged enough by all governments, for he is the true source
of public credit. (Regnault 1863, 102)
With these words, Regnault sediments a moral dichotomy in thinking about

finance, opposing on the one hand the legitimate, informed and rational
investor, who is responsible for the normal function of the financial markets
and contributes to financing the economy, to the agitator who acts in pure self-
interest, and whose unsound speculations causes the perturbations responsible
for the abnormal functioning of the markets and ensuing financial crises.
Regnault’s argument is that, rather than chastising speculators for the

immorality of their gambling behaviour, it is far more effective to demonstrate
that the random distribution of wins and losses eventually cancels out any
advantage that the speculator may have obtained over time. In other words,
the chaotic variability of a great number individual speculator’s decisions
eventually convergences around a statistical average, which also represents
the best approximation of ‘true market value’, while more extreme values peter
out in proportion to their distance from the average. Although each one of the
speculators’ decisions can be considered to be independent from one another, like
the toss of a coin, future values always distribute according to the Gaussian curve,
irrespective of the past. What good is it to blame individuals for their reckless
behaviour, if all you have to do is prove them mathematically that they can’t win?
Even though the scope and depth of probability theory and statistics in finance
has changed dramatically since the nineteenth century, Regnault’s confidence
in the normal distribution of wins and losses of pure speculation appears to
have held up remarkably well: despite the fact that private computers have given
millions of amateur traders the possibility of participating directly in trading,
less than 2 per cent stand to make consistent profit (Preda 2017).
Although the blame is still squarely placed on the immorality of the spurious
speculator, Regnault in fact opens an avenue for thinking about market
volatility in structural terms, by proposing a mathematical solution to what he
perceives to be a moral problem. Probability, at this very moment, acquires the
status of a diagnostic and prognostic tool for the financial markets, which the
mathematician Louis Bachelier (Bachelier 1900) will further elaborate in 1900,

in his doctoral thesis Théorie de la spéculation. As Christian Walter and Michel
de Pracontal remark in Le Virus B, crise financière et mathématiques, Bachelier
effectively lays the mathematical foundations for Regnault’s philosophy of
finance, arriving at the mathematical expression of a particular case of the
martingale, where probabilities are independent from each other and their
variance within an interval of time follows a Gaussian curve. The mathematical
properties Bachelier discovers, Walter and de Pracontal note, are equivalent
with those of Brownian motion, which Norbert Wiener will later further define
as a continuous time stochastic process. It is also during the period of this co-
development of theory and technology that mathematical finance becomes a
field in its own right.
After a lukewarm reception in France and a period of neglect, Bachelier’s work
was rediscovered during the 1950s by a group of future Nobel laureates in the
United States. Paul Samuelson, professor at MIT, saw in Bachelier’s work the basis
Empirical Noise 123
for a solution to the problem of estimating the price of options, which obtained
its full mathematical expression in 1973 in the Black-Scholes model. No longer
a mere means of counting and tracking, financial mathematics from this point
onwards becomes a factory of conceptual engines driving the financial markets.
Brownian motion, Walter and de Pracontal argue, will eventually become
the pithy mental image for fluctuations in the stock exchange (Walter and de
Pracontal 2009, 44), and on which the now ubiquitous idea of noise is based,
imagined as a form of static murmur in the regular distribution of variance.
Markowitz 1952 model of portfolio choice, later completed by James Tobin,
and the Black-Scholes option price model, developed in 1973 – all diversify
and further sediment the idea of regular random distribution as a dominant
paradigm (Walter and de Pracontal 2009, 44).
Regnault’s argument that the normal functioning of the markets counter-
balances the agitation of the speculators thus enters the mainstream of
mathematical finance vial Bachelier’s work, and continues to have contemporary
relevance also in relation to recent thinking about financial information and
noise. Paul C. Tetlock, now Professor of Finance at Columbia University, indeed
argued in 2006 that, although noise traders (conceived as agents with hedging
motives or irrational reasons to trade) reduce informational efficiency in a
securities market, their secondary effect is that rational agents counteract this
increased uncertainty by trading ‘more aggressively on their existing information’
and by acquiring better information:
For these reasons, two of the most widely used models in finance, Grossman
and Stiglitz (1980) and Kyle (1985), predict that an increase in noise trading
will not harm informational efficiency. In fact, if one allows informed traders to
acquire costly information, the Kyle (1985) model unambiguously predicts that
an increase in noise trading leads to an improvement in informational efficiency.
(Tetlock 2006)
The moral condemnation of speculation thus persists, obliquely, in the

opposition between information and noise, and in the opposition between
genuine investors and noise traders. But it does so in a relativized form, where
the reality no longer seems to be so clear-cut. Not only does information now
emerge as a reaction to uncertainty, if not noise, but going one step further,
Elena Esposito argues that
[t]here is no difference between ‘objective information’ which changes
expectations about fundamental variables, and pure noise, introduced by specific
agents (noise traders) for purely speculative purposes. (Esposito 2011, 81)
In fact, Esposito goes so far as to say that the very idea of perfect and uniformly
distributed information is unrealistic, and that the explicit recognition of
imperfect information is a better starting point for understanding financial
systems.
Underlying her and Tetlock’s revaluation of information and noise in finance
is the idea of the financial markets as an information system that, like Wiener’s
cybernetic system, has the capacity to self-regulate and compensate noise
through feedback. The point Esposito makes, however, goes one step further:
namely that there is also another factor. Alongside the traditional idea of noise
as accidental perturbation and of information as compensation of noise, this
other factor is the market’s capacity to observe itself. If information efficiency
in finance is to be thought along the lines of a self-regulating system, it is thus
better understood as a self-observing system, in analogy with a Second Order
cybernetic system:
The market provides a framework in which the operators can recognize
themselves and their inclinations. This paradoxically contradicts the hypothesis
of market efficiency because it is instead subject to a ‘“dynamic imbalance” that
is neither efficient nor rational, and can be exploited in a non-random way’.
(Esposito 2011, 66–67)
The persistence of an apparently clear-cut opposition between information

and noise, as indeed between genuine investors and noise traders, thus grows
increasingly out of sync with the insights afforded by Second Order cybernetics
– whose relevance is even more acute in the technical sense, when one considers
the impact of growing availability of new information technologies to individuals

who want to trade.
Alex Preda, interviewed by Mike O’Hara and Ricky Treadwell, in the HFT
Review (High Frequency and Algorithmic Trading) observes that the so-called
day traders, who emerged during the internet boom and were dismissed as
noise traders, have not only gained greater access to the computerized trading of
electronic markets (accounting for about 5 per cent of Forex’s daily value of $4.7
trillion), but are also increasingly educated in the natural sciences, mathematics
or engineering, capable of producing, testing and using their own mathematical
models. In other words, not only do so-called noise traders participate actively
in producing and shaping financial information, they also participate in the
strategic complexity of a self-observing market, whose agents can capitalize
on the discovery of a competitor’s strategies and trading patterns, and by
dissimulating their own:
Empirical Noise 125
I remember speaking with a screen trader not too long ago who was trading
futures spreads, and he said he could always tell when the algos and the bots
really start kicking in, because there were certain identifiable patterns of
activity. He shaped his trading around that, to either step back or to trade
in a certain way to counter what the bots were doing. (Preda, O’Hara, and
Treadwell 2013)
The intricacy of mutual recognition, adaptation and self-dissimulation in

the interaction between humans and algorithms, between individual and
institutional investors, not only complicates the distinction between information
and noise, it also complicates the question who is observing, and who or what is
being observed?
To continue blaming consecutive financial crises throughout the twentieth
century on the ignorance and immorality of the speculator or noise trader,
thus becomes inaccurate from today’s perspective, when hedging, program
trades, institutional investors and individual punters all contribute to a complex
informational eco-system.
It is also blatantly ineffective when moral condemnation regains pre-
eminence in the aftermath of a krach. Although the urge to blame and condemn
was undoubtedly laid in the cradle of mathematical finance by Regnault’s moral
philosophy, it is also he who argued, already in 1863, that moral condemnation
has indeed very little effect on behaviour, in comparison to mathematical proof.
In words that bring home the futility of the moral outcry after the subprime
crisis of 2007, Regnault insisted that
[m]orality […] to date, has never failed attacking the abuse of speculation […]
(yet) it is not by abstract and odious declamations […] (and) empty words that
one can hope to reform bad instincts […]. (Regnault 1863, 1–2)
Regnault clearly has no qualms over the designation of bad instincts, but his
observation of the inefficacy of moral condemnation has lost nothing of its
actuality today. In fact, the public outcry after the financial crisis in 2007 and
2008 led to a major public debate, calling for tighter regulation of the finance
sector, the curbing of immoral bonuses, and an unequivocal condemnation
of unbridled greed and speculation. Yet the moral imperative, so strongly felt
in the aftermath of the crisis, has failed in adequately addressing and hence
extirpating the root cause of exposure to financial crashes. Just this month the
Bank of England was compelled to warn that the financial sector is yet again
precipitating itself on a dangerous slope of easy credit, recalling the crisis of 2007
(Elliott 2017).
In fact, it was already as soon as 2013 that a Financial Times headline warned
of ‘Boom-era credit deals poised for comeback’ (Alloway and Mackenzie 2013).
Despite this warning, Christopher Thompson remarks two years later in a
Financial Times article that ‘Global CDO volumes have totalled $100bn in 2014,
two-thirds more than the year before’, while also ‘Global volumes of synthetic
collateralized debt obligations [CDOs] roughly doubled last year’ (Thompson
2015). Some will recall what role CDOs and swaps played in catalysing the
financial crisis of 2007 and 2008, alongside the bonanza of unsecured debt.5 If
we want to understand the inefficacy of moral indignation, it is worth drawing
a rough sketch of the plasticity with which financial information branches out
from these CDOs into increasingly protracted feedback loops, where financial
information becomes opaque and viscous.
Synthetic CDOs are essentially those structural financial instruments, which
enable banks and investors to reduce their exposure to the risk of unpaid loans
by selling the entitlement to repayment as a security – while at the same time
investing in hedge funds that counterbalance the risk of such credit defaults, by
betting precisely on the defaulting of these loans.
Synthetic CDOs are famously divided into ‘tranches’ with varying degrees of
risk and seniority. In order to promote their most senior tranches, which offer
the lowest returns, banks can make them more attractive by creating a so-called
‘leveraged super senior’ tranche, which allows investors to pay only a fraction of
its total value. Risk is thereby taken off the bank’s books, while in fact continuing
to expose the same bank to the possibility of losses, should the market value of
these products decline, such that contract clauses require investors to provide
‘billions of dollars’ of collateral (assets guaranteeing a counter value to the loan)
or, indeed, ‘walk away’ (Alloway and Mackenzie 2013).
A shadow banking system emerges from this market in synthetic CDOs,
requiring the creation of commercial conduits for the increasingly protracted
chains of debt obligations, so-called special purpose vehicles (SPV). SPVs are
commercial entities set up by banks or other lending institutions, in order to
provide short-term financing for companies or fund investments. Alongside
set-up and running fees, these conduits profit from selling short term by
generating funds with which to buy and then sell financial assets such as MBS
(mortgage backed securities), CMOs (collateralized mortgage obligations) or
CDOs (collateralized debt obligations).6 The gap between short-term borrowing
costs and returns on long-term investments in debt derivatives (‘securities
arbitrage’), in turn, opens up a niche, allowing hedge funds and banks alike to
raise cash from so-called asset-backed commercial papers (ABCP), which can
Empirical Noise 127
be used to buy higher-yielding securities (‘What Are Conduits, SIVs and SIV-
Lites?’ 2017).
As banks are required to hold a capital proportional to the loans and mortgages
they issue, selling their entitlement to mortgage repayment as a mortgage-
backed security (MBS) to such conduits means these debt obligations are taken
off the banks’ own balance sheets, leaving more capital available as leverage to
provide new loans and mortgages to customers. In the meanwhile, Hedge funds
wishing to increase their own returns can then borrow money from banks. As
Lisa Abramowicz puts it: ‘banks are lending money to hedge funds to invest in
derivatives that guarantee losses on loans held by banks’ (Abramowicz 2017).
Not only does the sold-off risk not disappear, in Abramowicz’s words these
individually crafted transactions look ‘a lot like the synthetic collateralized debt
obligations made infamous amid the 2008 financial meltdown’ (Abramowicz
2017).
This multiplying, shuffling and sampling of debt and promise, one could
argue, leads to increasing information opacity and viscosity: it is not only
difficult to see through, but also difficult to dissect the ramifications of liability.
Risk assessment and market analysis is, of course, always already vulnerable to
what one might call the ‘afferent noise’ of unforeseeable circumstances, both
in the form political and macro-economic uncertainty, but also in the form of
historical firsts, so-called ‘black swan’ events that cannot be foreseen on the basis
of past experience. But the obliqueness and viscosity of financial information
becomes a form of efferent noise, meaning a self-generated uncertainty, adding
itself to the basic contingency of pricing on the trading floor (Ayache 1771;
Taleb 2008). Although the mirage created by this vanishing act of exposure to
debt is not what is usually meant by noise in finance, synthetic CDOs, together
with the shadow banking system they generate, nevertheless add a dimension
of uncertainty that is not essentially distinguishable from noise and which
predictive models must strive to take into account.
The unexplained price and volume fluctuations in general equity trading we
commonly associate with the term ‘noise’, thus floats on a deeper uncertainty,
carried by currents of ever more oblique channels of financial information
that pulsate according to the cadence of the up-ticks and down-ticks of trading
positions on the market – an information uncertainty, or to use Shannon’s term
‘information entropy’, whose complexity rivals that of noise, (even without
taking fraud, cybercrime and fiscal crime into account).
Blaming the immorality of speculation not only blatantly fails to bring about
change through moral persuasion. What is worse is that it fails to understand
the way in which these synthetic derivatives are tied to the double bind of tighter
regulation and the demand to make more money available for economies blighted
with austerity. The requirement for banks to increase their capital and reduce
exposure to risk in fact means that regulators are compelled to sign off deals that
enable banks to sell on the risk tied to the loans they issue to hedge funds, which
in turn enables them to have more capital available to lend (Abramowicz 2017).
Far from cowing under the moral pressure to reform the financing culture that
led to the financial crisis, one could argue that regulators are in fact morally
compelled to enable the ever more sophisticated distribution of risk, in order
to generate liquidity and enable banks to fulfil their moral obligation of lending
and investing in the economy.
You may ask: what happens to these loans and derivatives of loans, and
derivatives of derivatives, in a financial world that considers itself enlightened
by the catharsis of the financial krach of 2007 and 2008? The answer is that the
biggest investor base in European synthetic securitizations, after Hedge funds,
are pension funds. This means that banks, regulators and politicians alike face
a nearing ‘silver tsunami’, meaning an ageing population with exponentially
rising care and health care costs (‘The Silver Tsunami | The Economist’ 2017),
with the financial tools that can transform the debt of the young (from student
loans, car loans, credit cards, mortgages, bonds) into liquidity for the old.
Taken as one, the derivatives market is thus a veritable will to the future, both
a will to power in quasi Nietzschean terms, in that it assumes the role of creator of
wealth (worth $553 trillion in over the counter synthetic CDOs in 2015), and a
legally binding pledge not unlike a testament, taking the form of futures, options
and swaps. It is thus both a logos of the future, by means of mathematical models
that underlie the pledged structuration of the future, and a nomos, procuring
entitlement to a future carved up into varying tranches of risk, ranging from
minimal uncertainty with low returns to high risk with potentially great returns.
The immediate future is thereby transformed into a financial legacy that
will be fulfilled after the contractual life cycle of each security. Each security, in
turn, is a ratified, legally binding promise, not unlike a testament in which an
imminent future will be inherited in tranches of risk.
What effectively amounts to a monetarization of the future in fact constitutes
a new paradigm – almost a New Testament – for the financial markets. Although
investment is by definition an investment in the future, the expectation of future
value could until recently still be grounded in its relation with the ‘real economy’.
Real here refers to an economy tethered to an actualized set of relations of
capital and productive forces. The derivatives market, in turn, capitalizes on the
Empirical Noise 129
virtuality of the future, whereby the actual becomes abstract, in the sense that
the link between the virtual and the actual is no longer intuitive in so-called
‘complex’ financial products.
In contradistinction with the New Testament, the fulfilment of this promise
is no longer postponed to the afterlife, but to an always imminent future,
continuously scattering value according specific temporal and legal modalities
(futures, options, swaps, mortgage- or asset-based securities, etc.). What is more,
the owned share of this promised future must mature before, not after the day of
reckoning, or it will be void. The financial crisis of 2007 was such a reckoning,
in which the promise of the future temporarily collapsed, as the noise of stock
market volatility turned into a global krach.
The bible’s New Testament supplants the old paradigm of the law with one of
redemption through Christ’s ultimate sacrifice of himself. Every new financial
crisis, in turn, reveals the stark choice between perdition and redemption in the
form of public sacrifice.
The uncertainty that could previously be blamed on stock market volatility
triggered by so-called noise traders, thus increasingly resembles a flight into the
future, ecstatic with the virtual wealth of futures, options and swaps, such that
the German noun Rausch [ecstasy, intoxication, jouissance], is better suited than
the technical term for noise [the verb Rauschen] (‘Global Economic Prospects
| Data’ 2017).
Regnault’s conviction that moral blame is ineffective in producing lasting
change in behaviour has become inaudible before the colossal, almost sublime
proportions of the risk of financial krachs. Yet what has persisted are the
basic tenets of moral philosophy, which subtend the opposition of normal
and abnormal functioning of the market, and the opposition of financial
information and noise. The apparent paradox is that the moral condemnation
of irrational speculation, inaugurated by Regnault’s moral philosophy, appears
to go hand in hand with the mathematical models used, with their underlying
assumption that the law of normal distribution stabilizes this virtual monument
to the future, and ultimately cancels out the distribution of risk in the securities
market.
There is, in fact, a correlation between this moral dichotomy, opposing
noise and information, and the reliance on the law of normal distribution as
a paradigm for mathematical finance. This is one of the main arguments in
Christian Walter and Michel de Pracontal book Le virus B – crise financière et
mathématiques. The dichotomy of normal versus irrational financial markets is,
the authors argue, precisely what perpetuates the repeated failures of predicting
the financial crisis of 1987, the Asian krach in 1997, the dot-com bubble and
more recently the 2007 subprime crisis, which unraveled in a global financial
meltdown in 2008.
In fact, Walter and de Pracontal argue that the dominant methods used for
mathematical modelling in finance today, are still premised on this distinction
between normal and irrational markets. In an interview with Le Monde, 28
March 2008, they quote Nicole El Karoui, founder of the influential masters in
‘Probability and finance’ at the Pierre et Marie Curie University, as saying that
the current crisis is not a crisis in mathematics, because the probabilistic models
used in finance to assess risk are ‘made to function in ordinary situations’ and
not ‘“in periods of overheating, of bubbles,” during which “behaviour is no
longer rational”’ (Walter and de Pracontal 2009, 16).
And yet, alternative mathematical models exist and risk predictions have been
made, yet could not be heard. Nouriel Roubini’s now famous assessment of the
risk of an impending global crisis was not even included in the IMF report in
April 2007 and his warning was discarded as ‘absurd pessimism’ (Walter and de
Pracontal 2009, 17; ‘Transcript of IMF Seminar – The Risk of a U.S. Hard Landing
and Implications for the Global Economy and Financial Markets’ 2007). In the
words of Nobel laureate Paul Krugman, influent voices, in recognized publications
like the Wall Street Journal, Forbes and National Review, ridiculed the ‘doom-sayers’.
Alternative approaches to the analysis of risk were brushed aside as undue
pessimism at a time of an unprecedented bonanza of debt, fuelled by the growth
of the subprime market, whose value had increased from $35 billion in 1994 to
$600 billion in 2006. The extent of risk was instead considered to be regional
and, in the opinion of Ben Bernanke, then chairman of the Federal Reserve,
limited to the possibility of slowed growth in the United States. What followed
was the krach, which took down three of the five dominating banks on Wall
Street (Lehman Brothers, Bear Stearns and Merrill Lynch) and consigned the
other two (Goldman Sachs, Morgan Stanley) and the insurance company AIG
to the most formidable state intervention in living memory, engulfing the global
financial markets, shaking the economy of many countries to the core and
provoking the worst and most widespread recession since the Great Depression.
A krach is an event of such magnitude that it breaks the cyclical logic of
economic rise and fall, acting instead as a tabula rasa. In other words, krach
amplifies the very idea of noise in finance to the level of an unheard-of crisis.
Charles Kindleberger and Robert Aliber for instance used the term already in
1978 in their book Manias, Panics and Crashes (now in its sixth edition), to
describe the Austrian krach of 1873 (Kindleberger and Aliber 2011, 151).
Empirical Noise 131
Although the term ‘Börsenkrach’ is commonly used in German to translate

financial crash, it is, strictly speaking, only the verb zusammenkrachen that
signifies a crash in the sense of a crashing together of two or more elements,
while the verb einkrachen signifies a collapse. While a crash designates a certain
type of event, such as a collision or collapse (here the collapse of value attributed
to financial products and shares traded in the stock exchange), a Krach in fact
qualifies the subjective dimension of a traumatic noise associated with such
an event. In German to have a Krach also means experiencing a relationship
breakdown, a dispute with high levels of animosity – not a crash then, but a noisy
agitation unravelling chaotically. Krach thus links the evocation of a rousing
sound (or alarming realization) that is in itself potentially traumatic, with the
idea of imminent, highly amplified discord: it would be wise to remember, when
we use the word krach to qualify a major financial crisis, that the origin of the
more common German word for noise, ‘Lärm’, like the English ‘alarm’, comes
from the Italian call for arms: all’arme!
Krach is one of several German words for noise – alongside Lärm (which
designates acoustic nuisance), and Rauschen (which designates the murmur
of common natural phenomena, like rustling leaves in a tree or the regular
rumbling of rolling waves, but also noise in the technical sense of static noise
or entropy). It is also the most emphatic word for noise, an onomatopoeia that
not only designates, but expresses a crashing noise. Closer to the Dutch word
for power [kracht] than either the English noise [etym. nausea] or the German
technical term for noise [Rauschen – rustling or murmur], Krach designates
the traumatic confrontation with a violently unpredictable reality, acoustic or

otherwise, and its catastrophic unravelling. In short: there can be subtle noise,
but not a subtle Krach.
Describing the global repercussions of the so-called ‘subprime crisis’
that rocked the financial markets in 2007 and 2008 as a krach, as many
commentators have done, thus adds one more dimension to our understanding
of noise in finance. It emphasizes not merely a greater, more powerful and more
unpredictable form of statistical variation as chaotic, non-linear, it shifts our
attention to a qualitative dimension of trauma associated with the unpredictable.
A global krach means not just more uncertainty with regard to the probability of
estimating or speculating on stock value (or its collapse), but the experience of a
traumatic implosion of the whole set of assumptions upon which the world-wide
financial system is based.
Rather than considering only the technical definition of noise, we must thus
look slightly askew at the question of the mathematical abstraction of noise:
we must ask ourselves how unburdened the idea of noise is by its sublimation
into pure probability, both of the physical paradigm of entropy (implying ideas
of loss of potential and of capacity to work), and of noise’s first and foremost
socio-aesthetic, moral and political aspects. Rather than sounding out only the
depth of noise’s mathematical formalization, a transdisciplinary approach to the
conceptualization of noise in finance thus faces the task of analysing its multiple
facets, from the anxiolytic (artificially calming) murmur of Brownian motion
to the alarming screech of non-linear, chaotic processes of a crisis, and from
rumour to krach.
VII
Statistics: The Discipline of the Prince
The dictionary defines information as ‘facts provided or learned about something

or someone’ (‘Information’ 2014a). Some dictionaries, like the Merriem
Webster, further consolidate this epistemological aspect of information as ‘the
communication or reception of knowledge’ (‘Information’ 2014b). What these
definitions presuppose, is an equivalence of information and knowledge, which
is relayed also in the Penguin dictionary, where information and knowledge
form a whole, composed of facts, data, signification and even ideas:
Information/[…]/noun 1a knowledge obtained from investigation, study, or
instruction […] b facts that represents data […] esp. regarded as significant […]
2 communication or reception of facts or ideas […]. (Allen 2001)
Given as a contextual reinforcement to this definition is a quotation by

anthropologist Margaret Mead, which adds a moral emphasis to the idea that
information should lead to the reduction of uncertainty by means of accuracy:
I was brought up to believe that the only thing worth doing was to add to the
amount of accurate information in the world – Margaret Mead. (Allen 2001)
Information thus requires trust in the obtainability of accurate facts and data,
indeed information is presented by the Penguin dictionary as ‘B facts or data
[…] 2 the communication or reception of facts that represents data’ (Allen 2001).
The dictionary definition thus builds a moral caveat into the definition of
information, which is that of trusted facts and data, in the form of accuracy. And
this moral requirement that information accurately reflects or indeed is ‘facts
and data’ comes into its own in the juridical aspect of the verb ‘to inform’, which
confers upon it also the power of:
3 a formal accusation presented to a magistrate. (Allen 2001)
Information, conflating accuracy/facts/data/knowledge, thus acquires a moral

and juridical signification, whereby accuracy becomes the basis for legitimacy,
grounding the just exercise of reason and power and safeguarding of law and
order:
Inform/in’fawm/[…] → verb intrans 1 to give information or knowledge. 2
(usu + against/on) to give the police or other authorities information about a
criminal or crime. (Allen 2001)
Facts and data are thus considered to be a given that can be obtained and provided
as accurate knowledge or suspected as false. The question what motivates our
search and our suspicions, and by what means empirical reality becomes a given
thus does not arise in the definition of information. It does not arise, because if
facts and data are given, then their emergence is not a question of becoming, and
the uncertainty preceding the emergence of facts and data does not come into
the epistemological equation of information. Information as facts and data points
to form and its accurate transmission, not to the emergence of form from the
formless and their mutual transformation.
The common definition of information thus has little room for uncertainty
as partaking positively in what we call information. Uncertainty falls on the
side of noise and excess together with inaccuracy, illegitimacy and illegality.
Consequently, it is not surprising that in the dictionary definition of information
also novelty, which necessarily implies a degree of uncertainty, is the least
prominent feature and mentioned last. In fact, a single word for the relation
between information and novelty suffices:
c news. (Allen 2001)
‘News’ is thus, in the context of this definition of information, expected to serve

a purpose, to consolidate knowledge, reinforce order and legitimate the exercise
of the law. If these dictionary definitions accurately reflect the understanding of
information in ordinary language, then information must be understood as facts
and data that serve the consolidation of knowledge and law and order, hence
consolidating established power through established knowledge, by reducing
uncertainty, disorder and subversion.
What follows is that noise is not, in fact, a straightforward opposite of
information in the ordinary sense, since the opposite corresponding to the first
and thus predominant definition of information as ‘knowledge obtained’ would
have to be ignorance or uncertainty, rather than noise, which the dictionary
defines first and foremost as ‘loud, confused, discordant sound, e.g. of shouting;
din’. Noise is only secondarily defined in relation to the concept of information,
and only in the context of the output of a computer, as
Empirical Noise 135
4. Irrelevant or meaningless information occurring with desired information in

the output of a computer. (Allen 2000, 945)
The etymology of noise also prioritizes this socio-juridical rather than

epistemic tension between information and noise:
Middle English via Old French noise strife, quarrel, noise. (Allen 2000, 945)
On the basis of these different levels of definitions of information and

noise, it is worth asking ourselves how it is that noise becomes the opposite of
information, in other words how the ordinary connotations of noise as improper
and immoral behaviour becomes the opposite of knowledge.
To answer this question, it is not enough to simply state that the information/
noise couple is specific to computer science, since we have seen that two very
distinct connotations cohabit in Wiener and Shannon’s definitions of information.
Values thus come into play, even in computer science, communication technology
and cybernetics more generally. The question that the discrepancy between
Shannon and Wiener’s definition of information enables us to ask is: where do
these values come from and how do they arrive at having such a determining
impact on technoscientific concepts?
It is worth looking for the answer to this question in the institutions of
knowledge, where the activity that produces information, i.e. investigation,
study or instruction, is consolidated under the authority to exclude uncertainty,
and to expel noisy individuals and all those unreceptive to the communication
or reception of accurate facts or data.
Only thus can we understand how cultural and scientific conceptualizations

do not constitute two separate spheres. To bring cultural and scientific
conceptualizations of information and noise up to speed with each other may
help us understand the problems inherent in the conceptualization of noise,
and the function of both information and noise more generally in relation to
established knowledge and power.
The opposition between Wiener’s understanding of information as negation
of entropy and Shannon’s definition of information as ‘freedom of choice’ is not
the question of a mere point of view, but a problem deeply rooted in our need for
prediction and control, but also of discovery of novelty. Let us briefly revisit how
these roots intertwine in the history of the calculus of probability and statistics,
from which we derive our contemporary understanding of information and noise.
The historical origin of statistics indeed reveals how the objective of maintaining
stability of power through knowledge becomes a culturally determining factor
for the definition of information and noise.
Statistics provide the data and facts that have become generally synonymous
with information, for instance when we obtain news about the rate of
unemployment or price indexes. It is also the discipline that has enabled
Boltzmann’s non-classical definition of entropy and, consequently oriented
Wiener and Shannon’s definitions of information and noise. This discipline
comes, as Alain Desrosières reminds us (Desrosieres 2006), from the German
Statistik. In its historical origin, the relation between knowledge and power
is thus not only explicit, but fundamental and moreover foundational, since
statistics first designated the ‘science of the state’, and was institutionalized as
such during the seventeenth century by Hermann Conrig (1606–1681) who
called it the: ‘nomenclature of knowledges necessary to the Prince’ (Desrosieres
2006).
As such statistics were first of all the sovereign’s privative domain of
knowledge. It is only from the 1830s that the domain of statistics was opened up
to be accessed by ‘enlightened men’ (Desrosieres 2006, 1019).
Before being reduced to the numeric data we now associate with statistics,
the discipline of the prince was broad ranging, comprising history, law, political
sciences, economy and geography, all classified according to Aristotle’s logic
into material, formal, final and of efficient causes. It is when John Gaunt (1620–
1674) developed Conrig’s system of taxonomy into a new method of ‘political
arithmetic’ that the mathematical foundations were laid for what we now
call statistics. Gaunt’s technique of transforming parish registers of baptisms,
weddings and deaths into facts and data relevant to the exercise of power, was
subsequently systematized by William Petty (1623–1687) as the beginning of

demographics. The knowledge statistics are now able to produce thus arises,
first of all, as a by-product of state administration, detailing not only the number
of births, deaths and marriages, but also of the number of crimes, epidemics,
commerce, schooling, prison and hospital admissions.
What we call statistics today thus emerges from this history of state
administration, from the correlation it establishes between the knowledge and
power. Yet this perspective would be one-sided, if one therefore overlooked that
contemporary statistics also emerge from the merging of two very different,
if not opposed forms of knowledge, by combining the ‘political arithmetic’ of
statistics with the calculus of probability. Desrosières shows how these two very
different disciplines intertwine historically and how great an epistemological
obstacle had to be overcome in order to arrive at today’s hybrid of statistics and
probability, which articulates the need to know with the mathematical measure
of doubt (Desrosieres 2006). It is this tension that persists in the apparently
Empirical Noise 137
paradoxical definitions Shannon and Wiener give of information. And it is in

the space of doubt that is characterized by this tension that we can find reasons
for the new conceptualizations of noise as a culture of doubt.
The historical origin of the calculus of probability in games of chance dates
back to the sixteenth century with Gerolamo Cardano. It receives its mathematical
foundation in the seventeenth century with the work of Fermat and Pascal, and
Bernoulli and Laplace in the eighteenth and nineteenth centuries7 (Desrosieres
2006; Snell 2012, ii). Bernoulli’s 1713 law of great numbers demonstrates that
a priori probability will be confirmed a posteriori, when events are repeated a
sufficient number of times in experience. The ‘normal distribution’ of Bernoulli’s
law of large numbers is also called the ‘law of possibilities’. Bernoulli thereby
provides the mathematical basis for the ‘hypothetico-deductive’ method of
probability and statistical frequency, laying the ground for the ‘experimental-
inductive’ method of statistics. The ‘law of possibilities’ is subsequently
interpreted by Laplace as an adequate representation of ‘errors in measurement
in astronomy’ and re-baptized the ‘law of errors’.
Amongst statisticians of the period, however, this new alliance between the
purely theoretical calculus of probability and empirical, inductive evidence is by
far not accepted as self-evident. The calculus of probability, as a purely formal,
abstract theoretical method of a priori determination of a problem first appears
too ‘subjective’ to statisticians, for whom the ‘frequentist’ observation of statistical
frequency of an event provides the needed empirical a posteriori grounding
required for reliable information. This distrust on the part of the ‘frequentists’
persists until the end of the nineteenth century, generating a material, frequentist
‘avalanche of statistics’ well into the 1930s (Desrosieres 2006, 1019). For two
centuries the emerging discipline of modern statistics is thus taught between on
the one hand the ‘subjectivism’ of the calculus of probability, which provides an
a priori ‘measure of uncertainty’, and on the other hand the certainty provided a
posteriori by the statistical frequency of events.
Twentieth-century statistics thus result not only historically, but also
epistemologically from the hybridization of the discipline of doubt and
the discipline of the prince, combining the known uncertainty of a priori
probability with the need to know, satisfied by the a posteriori frequency of
events and the taxonomy and enumeration that formed the ‘nomenclature of
knowledge necessary to the Prince’. This crossroads, where the measure of doubt
meets the need to know, consolidating the law of large numbers with William
Petty’s ‘political arithmetic’ (1623–1687), becomes the birthplace of modern
demographics.
VIII
The Man without Qualities
The Belgian astronomer and mathematician Adolphe Quetelet (1796–1874)

marked nineteenth-century statistics not only with institutional reforms,
but also with his moral interpretation of the ‘law of errors’. If the average of
imperfect measurements of a star give us a true idea of the real star, he argued,
then the average of the empirical variation of the height of military conscripts,
for instance, would give us a true idea of the ‘normal man’. Just as beyond the
distribution of observed positions of a star there was a real star, so the Gaussian
distribution of the sizes of military conscripts was the indication of a truth
comparable to the reality of the star. For Quetelet the ‘normal’ distribution of
heights of conscripts was proof of a constant cause, comparable to the real star
that transpired through the average of astronomic measurement. The statistical
average thus drew the portrait of man as a Gaussian ideal: the statistical average
henceforth becomes the ideal form, the true idea (Desrosieres 2006, 1019).
Not only the height of conscripts can now be measured against an ideal average
or norm, but any contingent event. The average number of crimes or suicides for
certain populations, can henceforth point to a real ‘moral propensity’, just as the
average of fluctuations in the stock exchange becomes indicative of true market
value in Jules Regnault’s philosophy of finance.
The singularities we now associate with the idea of noise as ‘unexplained
variation’ fail to constitute an ‘event’ in the eye of the nineteenth-century
statistician, as they are subsumed and diffused into the necessary unfolding of
the statistical tide of large numbers. So too the singularity of free will, blends itself
into the propensities of the masses. What emerges is the ‘man without qualities’,
whose singularity is washed away by the flow of great numbers, subjecting the
individuality of human agency to the same laws as statistical physics:
‘Statistical physics’ were constructed upon the idea that erratic movements
of microscopic particles, defined in a probabilistic manner, could result in
macroscopic regularities, just as the average man was relatively stable and
predictable, while individuals were volatile and unpredictable. A macroscopic

determinism was compatible with radical uncertainty at the level of elementary
particles. (Desrosieres 2006, 1020)
Francis Galton’s (1822–1911) concept of distribution eventually supplants

Quetelet’s notion of statistical causality with that of partial causality, sought
in the correlation between variables, in the co-occurrence and simultaneity
of traits8 (Desrosieres 2006, 1021). Karl Pearson further formalizes this ‘table
of contingencies’ during the 1880s and eventually renounces the concept of
causality altogether: causality is no longer deduced even from strong correlation,
nor from regression towards a normal average (‘regression toward the mean’),
as one variable cannot be said to be oriented towards another. Pearson will thus
reject causality as a purely nominal concept: reality henceforth resides in the
subjective correlation, or as Hume would say, in a habit of perception. Even
scientific laws, in this light, become summaries of such habits, codified routines
of perception.
Twentieth-century statistics inherits Pearson’s inferential use of statistics,
which it increasingly uses together with probabilistic models to prove or test
a hypothesis. The calculus of probability is thereby made to operate alongside
statistical inference, articulating the conditions of a priori uncertainty with a
posteriori certainty, bringing together the need to doubt and the need to know.
The difference between the calculus of probability and the analysis of statistical
frequency, in Desrosieres’ words, is now that between the ‘indeterminacy of the
world itself ’ and the ‘protocols of observation’. The merging of these two distinct
methodological aspects has also been called ‘the taming of chance’ (Desrosieres
2006, 1018; Hacking 1990).
Yet despite inheriting Pearson’s epistemic humility, statistics have become the
bedrock of public persuasion. The rhetoric of contemporary news media and
government reports alike, and even the dictionary definition of information, call
upon an almost infantile trust in facts and data – in other words in data as facts.
These are represented not only as indicative of real causes, but often also imply
Quetelet’s ideal of the statistical norm as indicative of truth and moral inclination.
The epistemic humility of a priori uncertainty, imposed by the calculus of
probability, no less than by Pearson’s inferential pragmatism, is thereby effaced
in the common perception of statistical results as ready-made facts and data,
which are taken to be indicative of causal determination. In the process, the
function of a priori doubt and uncertainty is supplanted by a quasi-religious
trust in the accuracy of the facts and data we derive from large numbers. The
more data, the more truth, hence big data.
Empirical Noise 141
And yet, we have seen from Eisler, Kertesz and Lillo’s analysis, how important
for instance time scales are for the statistical analysis of stock market variations,
that unpredictability and dynamics are highly dependent on the degree of
temporal resolution, and that, as a consequence, ‘the qualitative picture changes
dramatically when one moves down to the resolution of individual transactions’
(Eisler, Kertesz, and Lillo 2007). This being just one of many parameters that
make up the conceptual framework of statistical analysis, it should be evident
how much knowledge is required before statistical data, and the appreciation of
its accuracy, is no longer just given, but acquired.
Although Quetelet’s interpretation of the statistical norm as revealing
an ‘ideal cause’ eventually faded from statistical theory, it nevertheless left
a mark on what we now perceive as normal or abnormal. Today’s dominant
idea of standardized beautify is a sorry reminder of the soft power of Quetelet’s
statistical idealism. But also the implicit moral connotations that still linger in
our concepts of information and noise, as what is conform or divergent from
expectation, as what is true in essence as opposed to corrupted and distorted by
experience, still pay tribute to Quetelet’s numerical interpretation of the Platonic
ideal.
As Desrosières points out, the literature on the history of statistics is divided
between distinct approaches that cover either the sociological analysis of its
concepts and methods (Porter 1986), the history of institutions (Anderson
1988; Coll. INSEE 1987), the history of its mathematical formalisms, or the
philosophical implications of this new discipline of knowledge (Benzecri 1982;
Stigler 1986). This division of labour in the attempt to understand the full
breath of the significance of statistics and its history gives an appreciation of the
numerous fields of research that ought to be taken into account, if we wanted
to understand the now dominant conflation of information with both data and
knowledge.
Our trusting reliance on statistical of data, consolidated in the dictionary
definition of information, appears to owe a non-negligible debt to the conflation
of these fields. The epistemic, legal and moral implications of statistics,
acknowledged by the division of labour in its historical, sociological, formal and
philosophical analysis, remain implicit in the notion of information and must be
read between the lines of today’s dictionaries.
Information is conflated with data, data is conflated with facts, both together
become the medium for informing on or against. As data and facts become the
bricks and mortar of the reinforcement of law and order, information is affected
by a feeling of an ought to, in other words of a moral imperative, implicit in
the act of informing. It is thus necessary to remain vigilant of the conflation of

information and data especially in light of today’s culture of socially networked
personal confessions, paired with the means for statistical data mining and hyper
surveillance, which become all the more sinister when information is treated as
a given, when data are treated as facts, and when information effectively eclipses
uncertainty. We have the carelessness of rhetorical persuasion to thank for, if the
era of ‘post-truth’ can fall back on the brandishing of statistics.
It is worth remembering that what is a given at the origin of the modern
discipline of statistics, from which we derive our notions of data and information,
is not information, but power. What is given as the origin of Statistik is indeed
the God given power of the Sovereign, consolidated by a form of knowledge and
administration that becomes the discipline of the prince. This originally political
power of information is distributed, anonymized and generalized throughout
the rise of state administration during the nineteenth century. Statistical data
is opened up for the first time to public administration and the enlightened
public. The foundational role of power, as a consequence, is pulverized into a
disinterested scientific discipline: in other words, the power of information acts
incognito.
Conversely when we now speak of noise in statistics, of error in signal
transmission or in the logical chain of reason, we must beware of connotations
that are as much cultural and political as they are technoscientific. The
connotations implicit in the words used in scientific discourse act as a murmur
of moral and political values. The value judgement levied against noise, when
defined as ‘parasitic’ on information, thus expresses not only a measure of

uncertainty, but also an implicit threat of insubordination to the norm.
IX
Noise Abatement: The Dawn of Noise
Of course, we no longer mean acoustic noise, when we speak about noise in

statistical terms, be it in the stock exchange or in molecular biology. And yet,
despite the complexity of thinking about noise in mathematical terms, the term
noise, with its imagined inverted commas, has become increasingly current
because it benefits also from its intuitive appeal. The common experience
of acoustic noise remains strongly suggestive, even if we know that what is
meant is not, for instance in finance, the din of the trading floor. The ‘special’
understanding of noise may no longer be reducible to the ordinary sense of noise
as acoustic nuisance, but its empirical relevance nevertheless remains associated
with the intuitive experience of acoustic noise.
Noise appears to be the acoustic waste generated by industrial development,
a side effect of the mastery of thermal noise in machines. The latter entailed the
sudden presence of acoustic noise from machines, automobiles and, increasingly,
the growth of entertainment and communication technology for an increasingly
stimulus hungry modernity. While the general notion of acoustic noise may be at
the origin of the noise metaphor in information theory, cybernetics, but also in
finance and the empirical sciences more generally, ‘noise pollution’, in turn now
appears as a side effect of technological innovation and industrialization.
The two noise discourses, statistical and acoustic, henceforth develop side by
side, yet surprisingly without much dialogue: while the humanities take stock
of the effects of acoustic and visual noise on society, Science and Technology
studies discover, alongside the visual design of new technologies, also the
economic incentive for designing the sound new technology makes. Engines
purr, domestic appliances bring a panoply of signalling sounds into our homes,
while the ideal operation of an extractor hood or vacuum cleaner is subject
to decibel ratings. Transport, industry, machines, but also the presence of
technology in our homes makes itself heard, producing a cacophony without
orchestration.
An author whose analysis of the condensation of the scientific, technological,

aesthetic, moral and political dimensions of noise is, in my view, exemplary is
Karen Bijsterveld. Although noise has an impressive bibliography at its heals in
historical, cultural and even sound studies, Bijsterveld’s articles ‘The Diabolical
Symphony of the Mechanical Age: Technology and Symbolism of Sound in
European and North American Noise Abatement Campaigns 1900–1940’ and
‘City of Din: Decibels, Noise and Neighbours in the Netherlands, 1910–1980’
provide a remarkably rich entry point to the theoretical polymorphism of
noise, which will help us to contextualize the contemporary problem of ‘noise
pollution’ and of its quantification.
Technology and science are not only what enables us to discern the formal
acoustic properties of the sounds we call noise, but it is the very progress of
science and technology that has irreversibly altered the soundscape of modern
society. All the more surprising, Karin Bijsterveld notes, that there is what one
could call radio silence on the topic of noise between the cultural theory of noise
on the one hand, and Science and Technology studies on the other (Bijsterveld
2001).
It might be surprising that such a ‘radio silence’ could long persist between
academic disciplines that are implicated, respectively, in assessing the sounds
of technological innovation and the sociocultural transformation of our
soundscape. While cultural theory takes into account the symbolic values and
sociopolitical significance of noise or silence, Science and Technology studies
focus on functionality and design of new technologies, until recently privileging
visual characteristics over acoustic design and more generally privileging issues
of sound engineering over an understanding of the wider acoustic resonance
of modern technologies. Yet the absence of dialogue between Science and
Technology Studies and the cultural analysis of the affective and symbolic
dimensions of noise can only be detrimental to both. It is
a silence that should be broken, since the sound of technology not only tunes our
sonic environment, but has also been a highly controversial aspect of technology
loaded with symbolic significance. (Bijsterveld 2001, 37)
The affective and symbolic charge, associated with the noise of modern
technology, co-determines desirability or rejection of innovations. Road and
air traffic noise; car alarms and ambulances; the ubiquity of mobile phones
ringing; fragments of private conversations dissipating into public space; call
waiting lines with digitalized Mozart tunes; shopping malls resounding with
a cacophony of ‘ambient’ music: if the noise of industrialization could arouse
Empirical Noise 145
futuristic excitation at the beginning of the twentieth century, the post-

modern soundscape has become a delirious imposition. Noise becomes an
interdisciplinary paradox: while the noises of modern life require orchestration,
so does the knowledge about noise coming from the humanities on the one hand
and science and technology on the other.
As the enjoyment and rejection of the sounds of new technologies divides
opinion, the apparently straightforward problem of noise pollution becomes a
sociopolitical problem of power: the power to judge what is noise and what is
not and the power to regulate noise.
The divisiveness of value judgements regarding the noises of modern
technologies accompanies the history of noise abatement right from its
beginning. Technological novelties such as the telephone, radio and gramophone
are greeted with enthusiasm about modernity, but also raise concern about the
deafening presence of new technologies in social life:
[People] wanted to listen to their radio, but some nearby electric machinery
interfered with proper reception. They wanted to talk to each other, but people
shouting into their telephones prevented them from doing so. […] radio
was third on the list of the most reported sources of annoyance [while] the
gramophone came up for discussion in Dutch local politics as early as 1913.
(Bijsterveld 2003, 175–78)
In response to the perceived intrusion of the noise of modern technologies,

arises the need to identify, assess and quantify the perceived rise in noise
nuisance in the urban space. Audiometers (previously developed for hearing
tests), are converted into noise meters and put to use by public law enforcement
in London, Chicago and New York as early as 1926. Bijsterveld’s historical case
study of the Netherlands brings us the delectable story of the collaboration
between a scientist and an Amsterdam police chief, in developing the first
portable noise measuring device, the ‘silent witness’ or ‘Silenta’. The Silenta will
become the first legally valid means for the measurement of noise, recognized
by the Dutch Supreme Court in 1939 (Bijsterveld 2003, 185). A ‘Silence Brigade’
is formed in the police force, and ‘four policemen equipped with a motorcycle
with sidecar’ are attributed the task of ‘the hunt for decibels’ in the pursuit of
traffic noise offenders.
Almost like a comical staging of the scientific problem of measuring noise, the
attempt to measure noise is immediately confronted with the noise arising from
the method of measurement itself: the roaring of the police officers’ motorbikes
hunting down narrow streets of Amsterdam and the silenta’s airstream sensitivity
condemn the exercise to a first wave of public scepticism, as initial enthusiasm

in the press quickly yields to disappointment (Bijsterveld 2003, 186). The tide
of public opinion sways against the upholders of civil and acoustic order, whose
scientifically objective noise assessment is ridiculed in newspaper articles, such
as the following extract of an article in De Telegraaf’s of 7 July 1937:
You snatch the phone and dial our silence dictator, Police Chief Bakker, and
explain the case to him […] Hardly three minutes later, you can hear the silence
brigade’s motorcycle rush into your street […] With them, they have professor
Zwikker’s sound meter and indeed, true enough – this is too much needless
noise. Yet whereas the noise little John produces exceeds the limit by 0.12, little
Peter, his twin brother, exceeds the maximum limit by no less than 8.65cB. What
is the brigade to do in such a case? Arrest little Peter and take him with them
[…]? (Bijsterveld 2003, 187)
In the process, it is not only the image of the Silenta’s scientific accuracy that
is tarnished. More importantly, it is the moral high-ground associated with
purported scientific objectivity that is now subject to ridicule, as the Silenta comes
to represent not progressive technology but a reactionary attitude to modernity:
a symbol not of technological innovation, but of conservative reticence.
As the scientific challenge of measuring noise in view of its regulation
increasingly meets the rise of leisure occupations, associated with loud sound,
so the technological problem becomes a moral and aesthetic debate over the
attempt to curb noise – which soon takes on the political dimension of self-
determination. The right to protect one’s acoustic space from the noise of others
is confronted with the claim to the right of one’s own acoustic presence.
The debate dividing Dutch opinion on noise in fact dates back to the
beginning of the century, when new technologies such as the radio and the
gramophone are blamed across Europe and the United Sates for the increase
in noise. Mass media stand accused and yet are called upon to instil acoustic
manners by teaching a ‘noise etiquette’ through public education – requiring the
radio for instance to advise listeners to turn down the volume at a certain time
of the evening. Combatting noise becomes a problem of redressing the moral
decadence associated with modernity. As Bijsterveld notes,
[M]aking noise was thought of as barbarian, uncivilized, anti-intellectual and
disruptive behaviour – in short, as a lack of self-control. (Bijsterveld 2003, 175)
In a backlash against the apparent moral high-ground of the assault on noise,

political representatives of the working classes will in turn assert the right of
workers to listen to popular music or to their favourite radio shows on their
Empirical Noise 147
new gramophones and radios, notably at night, because the long working hours
in factories prevent daytime leisure and justify night-time recreation. The low
cost of the gramophone makes it the ‘“musical instrument” of the lower classes’,
whereby not only the decibels it emitts, but also the popular music played on
it and the behaviour associated with it are seen as objectionable by the middle
class:
One did not hear gramophones, a critic claimed, in the city’s upper-class
districts. (Bijsterveld 2003, 178)
A political rift thus arises at the heart of the noise dispute, which takes noise from
the problem of quantification, via a detour of aesthetic and moral judgement,
to that of political arbitration. Leftist council members rebuke the conservative
accusation of acoustic ‘debauchery’ as being elitist, pointing out that that the
workers have a right to a ‘sound culture’ of their own, and that ‘one could be
equally bothered by lady singers, trombonists […] “maltreated” pianos [and the]
“miserable lamentations” of concertinas’ (Bijsterveld 2003, 179).
Interestingly, a first victory is won by the conservatives, on the basis of the
difference between mechanical sources of music and acoustical music practice,
offering the perhaps surprising argument of the latter’s moderation due to human
fatigue:
[A]fter an hour’s practice flutists, oboists and so on […] feel a need to do
something else, like smoking a cigar, talking to someone or devoting themselves
to some other study. They get tired and stop playing and therefore they cause
les hindrance, nuisance, harassment and irritation. (Bijsterveld 2003; ‘MA
Rotterdam, Proceedings City Council Rotterdam, Session 26 February 1914,

64–65’, n.d.)
The argument succeeds in passing an ordinance in Rotterdam with a vote won

by 24 to 13, followed by an amendment giving the police the right to penetrate
the private property of the noise offender against his will, when accompanied
by a high official. A decade later this ordinance is adopted by the councils of
The Hague, Leiden, Breda, Utrecht, Amsterdam, Maastricht and Groningen,
countering the invasive nature of the noises and new technologies with the
invasive power of policing and regulation. The conservatives ban the ‘stupid,
machinelike’ playing of records and radios for ‘hours and hours without really
listening’ (Bijsterveld 2003, 180). Yet the conservative reaction to noise appears
increasingly reactionary in the eye of progressive members of the public and
‘the grounds on which the loud playing of gramophones and radios could be
legally punished became narrower’. The continuing resistance of left liberals,
social democrats and communists eventually leads to the obsolescence of

neighbourhood noise regulations in general (Bijsterveld 2003, 180). Noise in
the process becomes more than acoustic excess, it becomes the soundtrack of an
emancipatory confrontation with dominant norms.
Today’s porous regulation regarding neighbourhood noise, noted by the
WHO report on noise pollution (‘WHO Guidelines for Community Noise’
1999), is a direct result of this cultural and political divisiveness of the definition
of noise.
X
Noise Pollution
The historical shifting of the boundary between what is considered a noise offence
and what is considered a legitimate acoustic presence and civil right generates
not only new social and political norms, but also creates a new culpability of
transgression. Noise becomes the excess generated by the acceptance of a new
norm. What is at stake here is not only the changing soundscape of modern
society, but the divisive and decisive power of norms. While neighbourhood
related noise fails to generate consensus, growing awareness of the toxicity of
‘noise pollution’ means that the quantitative measure of noise remains an urgent
topic.
Acoustic noise, of course, is not a novel concern, tied only to modern
technology, but one of the most ancient recorded objects of legal dispute.
The Romans already imposed legislation regarding noise of ironed wheels
hitting stone paved roads at night. It is with the advent of modern transport,
however, that noise from lorries, diesel engines, aircraft, trains and industrial
means of production becomes an indelible and increasingly toxic feature of

industrialization and urbanization, recognized by the World Health Organization
(WHO) as a serious cause of concern for public health:
The extent of the noise problem is large. In the European Union countries about
40% of the population are exposed to road traffic noise with an equivalent
sound pressure level exceeding 55dB(A) daytime and 20% are exposed to levels
exceeding 65 dB(A). Taking all exposure to transportation noise together about
half of the European Union citizens are estimated to live in zones which do not
ensure acoustical comfort to residents. More than 30% are exposed at night to
equivalent sound pressure levels exceeding 55 dB(A) which are disturbing to
sleep. The noise pollution problem is also severe in cities of developing countries
and caused mainly by traffic. (Berglund and Lindvall 1995, iii)
A side effect of economic development, noise thus presents itself as a luxury

problem of developed countries: the noise of economic success. This view, however,
risks underestimating the impact of noise on public health, an underestimation

reflected by the lack of decisive action at the level of local, national and international
policy. Noise, the report argues, must be recognized as an issue of unsustainable
development. Defined in terms of its toxicity as a pollutant, ‘noise pollution’ is
comparable in its toxicity with chemical substances, subject to cumulative adverse
effects on health, not only in the present, but on future generations.
In order to shift the lethargy in the perception of the noise problem amongst
policy makes, the WHO report pitched the public health risk of noise in the
only terms it thinks will rouse attention: namely in terms of the economic
impact of noise pollution. Intensive urbanization of large cities in particular
raises the problem of noise as a threat to productivity. The reduction of hearing
range and even loss of hearing, due to noise, is an obvious example. But also
other psychophysiological variables of the city dweller are affected by noise,
threatening notably the performance of urban white collar workers:
A major step forward in raising the awareness of both the public and of decision
makers is the recommendation to concentrate more research and development
on variables which have monetary consequences. This means that research
should consider not only dose-response relationships between sound levels,
but also politically relevant variables, such as noise-induced social handicap;
reduced productivity; decreased performance in learning; workplace and school
absenteeism; increased drug use; and accidents. (Berglund and Lindvall 1995, xviii)
Noise impairs the learning process of school-children, especially in the phase

of speech acquisition, burdens communication in communal spaces with
poor architectural design of reverberation and even the increase of aggressive

behaviour in ‘predisposed’ individuals is considered a correlate of noise
pollution. Also aesthetic considerations come into play, concerning for instance
the report’s recommendation to preserve the acoustic serenity of conservation
areas (Berglund and Lindvall 1995, vii).
The report thus places the urgency of a quantitative measure of noise within
a complex framework of its adverse effects, from the medical, as in hearing loss,
to the sociopolitical, touching on issues of sociability, all of which become more
relevant, it seems, in economic terms of productivity rather than wellbeing.
The question is, how can such a complex set of variables be pinned down to a
quantitative measure of noise? Is a quantitative measure of the ‘complex pattern
of sound waves’ that makes up the urban soundscape adequate as a basis for
policy making and accountability? Critics have been quick to warn that a purely
quantitative measure of noise fails to address the complexity of the problem of
Empirical Noise 151
noise, unless it can take into account the competing interests in making and
curtailing noise (Hydaralli 2012).
While this sociopolitical aspect remains a controversial dimension of the
problem of noise, as we have seen with Bijsterveld, the WHO report also stands
out with a highly stratified methodology, taking into account not only loudness,
but frequency, time variables, context and hearing range, differentiated enough
to encompass a wide scope of environmental noise, comprising noise from road
and rail traffic, air traffic, industries, construction, but also neighbourhood
noise arising from the catering sector, from live or recorded music, from sports
events, playgrounds, car parks and even domestic animals, such as barking
dogs. Not only outdoor, but also indoor noise is listed amongst the sources of
noise pollution; loud, but also anodyne noises are taken into consideration, such
as the noise of ventilation systems, of office machines and home appliances,
loudspeakers and headphones and architectural reverberation properties of
buildings.
Frequencies are measured alongside sound pressure and also the ‘signal-to-
noise’ ratio is factored in, in order to assess impact on communication, basic
tasks and performance, including sleep. Context- and task-related specificity
also means taking into account the progression of noise over time: does it have a
sudden onset with startling effect, or does it weigh continuously on daily activity,
like traffic noise? Is its intensity stable or variable? A very quiet or low decibel
noise and certain types of frequencies that barely register in the measurement of
traffic noise, become highly significant in the signal-to-noise ratio, when they
impede a conversation, disturb sleep or impair cognitive effort. Noise affects

not only individuals but groups, for instance when poor architectural design
results in reverberation in public buildings, impairing learning in schools
or convalescence and even medical staff ’s attention to signals coming from
monitoring devices in hospitals.
Logarithms for ‘frequency weighting’ adapt the ‘combined sound energy’ over
a period of time, according to different parameters. The logarithm LAeq, T, for
instance, measures the energy average of continuous noise, such as road traffic,
so-called A-weighted sound, over a specific period of time, weighting lower
frequencies less than mid- and high frequencies, which are more noticeable in
the human hearing. The type of noise onset and its intensity are measured against
maximum noise level (LAmax) and sound exposure level (SEL). Frequencies, i.e.
vibrations per second, measured in Hertz (Hz), as well as the loudness of sound
pressure levels, measured in Decibels, are measured against the range of human
hearing, situated between 20 and 20,000 Hz for unimpaired younger listeners.
This logarithmic scaling takes into account not only context as a time variable,
like night-time or daytime sounds, but also accounts for signal-to-noise ratio
and degree of habituation in order to assess its impact on activity, be it sleep,
communication or cognitive work load (Berglund and Lindvall 1995, vii).
The quantification of noise is thus already armed with a highly sophisticated
methodology, measuring additive sound events against a maximum tolerable
to human hearing, and stratifying noise according to context and occupation.
In other words, this method provides quantitative pointers for the qualitative
tipping point at which our normal experience of the acoustic environment
becomes toxic and is suffered rather than experienced. The quantification
of noise is thus not a simple metric of loudness, but a tailoring of method
to context as well as task specificities. While limitations of these measuring
methods and their articulation are acknowledged in the report, both the
economy and practical advantages of a standardized approach are argued to
outweigh these.
The difficulty that arises for us from the quantitative measure of noise is
thus not the lack of complexity of this measure. It is clear that a methodology
underlies this approach that can take many variables into account. As a result,
however, also the definition of noise scatters over multiple contexts and acoustic
characteristics. Noise is not necessarily loud, but undesirable across a large array
of factors, involving acoustic properties, time, disposition and activity. This
leads to a definition of noise as audible frequencies and sound pressure, whose
toxicity must be classified according to both intrinsic and extrinsic conditions. As
a result, the measure of ‘noise pollution’ leads to a taxonomy which, rather than
giving a unified definition of the concept of noise, on the contrary pulverizes
its object: there is not noise, but an open ended series of definitions of noise
corresponding to an open ended series of contexts and tasks.
The starkest contrast is to be found between the definition of noise in a
purely statistical form of acoustic analysis on the one hand, and the analysis
and taxonomy of ‘noise pollution’ on the other. In purely acoustic terms,
noise has a very simple definition:
Noise is an aleatory or irregular wave. In acoustics, therefore, noise is not

an unpleasant or un-aesthetic sound, but a continuous signal in which no
particular frequency can be distinguished. […] What we call white noise is a
noise containing all the frequencies at the same intensity. (Volcler and Volk
2013, 10)
Empirical Noise 153
The concept of noise pollution, far from being simple, thus becomes
polymorphous, moulding itself to different criteria and types of experience.
As a consequence, as the method of measuring noise pollution increases in
mathematical specificity, the concept of noise loses the simplicity of both its
original intuitive appeal and of its purely statistical acoustic definition. Yet was
it not either this intuitive appeal, or the trust in its mathematical formalization,
which we presumed to underlie the ease with which the idea of noise has
invested theoretical and experimental fields? What had appeared to cast the
great conceptual arch of noise, from unwanted sound to statistical variation and
even stock market volatility, was the simple idea of the undesirable, articulated in
so many analogies of acoustic perturbation.
What does the quantitative measure of ‘noise pollution’ have in common
with the quantitative definition of noise in information theory and cybernetics?
The obvious point in common is that both propose a quantitative measure of
noise using statistical analysis. But ‘noise pollution’ and noise in the channel of
communication are not just two different objects of scientific investigation. ‘Noise
pollution’ measures and classifies noise as an object of experience (acoustic noise
in the audible range of human hearing), according to its different logarithmic
weightings. Information theory, as we have seen in Part One, measures both
noise and information as a relation of probability. The two quantitative measures
of ‘noise pollution’ and of ‘noise in the channel of communication’ thus share a
statistical base, but they differ, apart from their domains of application, also in
their epistemological priority.
The definition of ‘noise pollution’ gives epistemological priority to the object

(the phenomenon of noise with its various properties), which constitutes one term
in a relation of perception, (the other term being that of individual or collective
perception). The definition of noise in information theory, on the other hand,
gives epistemological priority to the relation of probability between a signal
and the set of all equally possible signals given certain constraints. Information
and noise are constituted by this relation of probability. ‘Noise pollution’, too,
pays attention to relation, by emphasizing context and task specificity – but
it still measures the amplitude, frequency and duration of something that
constitutes, and is categorized and weighted as, an object of perception: noise.
The very term, ‘noise pollution’, designates a substance, a pollutant. The quantity
designated as entropy in information theory, on the other hand, measures
the relation of probability of an event, not its ontological characteristics. This
degree of probability can also be expressed in terms of frequencies, amplitude or
duration, but it is the relation with a set of virtual events of equal probability that
characterizes both noise and information in information theory. The difference

between ‘noise pollution’ and noise in information theory is thus a subtle shift
of emphasis from object to relation, easily overlooked in favour of objective
‘similarities’ and the analogy with perturbation.
The difference is thus not merely that both approaches deal with a different
set of phenomena. Information theory does not define the object it measures,
because it is precisely the ontologically arbitrary definition of a relation of
probability from which Shannon’s definition of information and noise derives
its wide applicability. This precisely is the problem posed to the definition of
noise by the difference between the quantitative measure of noise (as sound) and
the quantitative measure of information and noise (as a relation of probability).
One measures noise as object of perception, refracting the definition of this
object over an in principle infinite taxonomy of types of acoustic properties,
with different weightings according to context and task. The other measures
neither object nor content, but the relation of probability between our power
to predict the actual occurrence of an event on the basis of what came before
and the number of virtual possible alternatives occurring with equal probability
within equivalent constraints.
The difference between ‘noise pollution’ and noise in the channel of
communication thus comes down to the difference between a theory of objects
of perception, which constitute the terms of a relation between individual and
environment, and a theory of relation that constitutes its terms as probable or
improbable. This difference may seem irrelevant at this point, but it acquires all
its importance if we recall Bijsterveld’s historical analysis, where noise becomes

the function of a conflictual relation that crystallizes the sociopolitical terms of
the relation, both in the form of opposing political groups and in the form of
their respective soundscapes.
XI
Toxic, Viral, Parasitic
The prime concern for the WHO’s definition of noise’s toxicity as a pollutant
is its impact on health. The most obvious toxic effect of noise is, of course, the
impact of high decibels on the hearing apparatus. What surprises are the figures,
which reveal the gravity and the commonality of the problem:
Worldwide, noise-induced hearing impairment is the most prevalent irreversible
occupational hazard and it is estimated that 120 million people worldwide have
disabling hearing difficulties. (Berglund and Lindvall 1995, viii)
Hearing impairment is defined here as the increase in threshold at which sound

becomes audible. Also tinnitus, a permanent ringing sound, may be triggered by
high frequencies between 3,000 and 6,000 Hz, but even by frequencies as low as
2,000 Hz when exposure takes place over a prolonged period of time (Laeq, 8h).
The consequences of hearing impairment are not only medical, however, but
also and eminently, social. Impairment in the ability to understand speech, even
at a reduction of hearing as small as 10 bB, becomes a ‘severe social handicap’

affecting communication and learning processes that involve language
acquisition or recognition of speech. Not only exposure to extreme levels of
sound, affecting the hearing apparatus, but even subtle noises that interfere
with speech understanding contribute towards social handicap induced by
hearing deficit, delaying language acquisition and impairing communication.
Consequently, it is not only deafening noises, but even the slightest reverberation
in poorly planned architecture of public spaces, especially schools and hospitals,
where signal to noise ratios matter most, that have toxic noise effects, and
where affected speech discrimination increases cognitive load, making ‘speech
perception difficult and straining’ (Berglund and Lindvall 1995, ix).
Guidelines concerning the control of noise pollution therefore concern
not only exposure to decibels above 140 dB, to prevent damage to the hearing
apparatus, but also the ‘signal-to-noise’ ratio in buildings with reverberation.
The difference between speech level and surrounding noise should be 15 dB(A),
but there is also reverberation time, which is desirable to be below 0.6 second,
even in a quiet environment, and especially where vulnerable subgroups are
concerned, such as children in the process of language and reading acquisition
or those not yet familiar with the language spoken. Also hospitals, where
the perception of signals from monitoring devices is vital and where delayed
recovery of patients is correlated with disturbed rest, present an area where
the signal-to-noise ratio is of the greatest importance (‘WHO Guidelines for
Community Noise’ 1999, 11).
What the impact of noise on learning highlights, is the effect noise has,
beyond damage to the hearing apparatus and beyond the mere annoyance or
disturbance that it ordinarily evokes, on the process of cognition and the wider
impact on communication and participation in the social fabric. This aspect
of toxicity de-qualifies any relativistic approach, that would have the idea of
unwanted sound become a personal point of view. Nor can noise be reduced
to a merely quantitative variable, correlated with impaired performance of
school children and cognitive load of white collar workers exposed to multiple
sources of noise – comparable perhaps to having to carry a surplus weight
while working. Noise instead acquires an eminently psychosocial importance
that pertains to hearing, but also to the way in which the social fabric can tear
at the slightest lowering of threshold of sensitivity to sound.
Although findings on the relation between environmental noise and mental
health effects were as yet inconclusive at the time of the cited report, the WHO
guidelines nevertheless stress the need to further investigate the scientific basis for
understanding the relation between noise and mental health. The insufficiency
of scientific evidence for the impact of noise on mental health is criticized as
demonstrating lack of interest in the problem at the level of government, and
this despite the availability of data concerning the measurable use of drugs such
as tranquilizers, sleeping pills and hospital admission rates. The lack of scientific
dedication to the matter is all the more disappointing, according the report,
as statistical correlation between occupational noise and the development of
mental disorders, such as neurosis, but also acceleration and intensification of
latent mental illness, already exists (Berglund and Lindvall 1995, x).
There is, of course, an established field of research into the psychology of
perception of acoustic noise, which the WHO report does not mention for
obvious reasons. However, this field has been fruitful less in cementing scientific
proof of the correlation between ‘noise pollution’ and mental health problems,
than in developing technological means aimed at the deliberate use of noise as
deterrent, non-lethal and even lethal weapon.
Empirical Noise 157
By presenting noise as a toxic, but accidental by-product of industrial

development and urban density, the WHO report in fact casts a blind eye on the
development of technologies for the deliberate use of noise. These technologies
aim at exercising a new form of control and at producing a more acceptable show
of power than the now unpopular visibility of violent militarization. Research
into the psychology of perception and the physiological and psychological
effects of noise has indeed become a valuable resource for the development of
military weapons to be deployed in situations where blood-shed is not politically
acceptable, and of commercially available deterrent devices, aimed at situations
where the visibility of such measures would be detrimental to the perception of a
brand. What is striking that the scientific research and technological advances in
the field of noise weapons appears to have no productive overlap with available
scientific data on the correlation between noise and mental health. It is, on the
contrary, characterized by the absence of public information and lack of access
to independent experts. In Jürgen Altmann’s words:
Acoustic weapons are under research and development in a few countries.
Advertised as one type of non-lethal weapon, they are said to immediately
incapacitate opponents while avoiding permanent physical damage. Reliable
information on specifications or effects is scarce, however. (Altmann 2001)
Such technologies range from the extremely high power sound pressure,
deployed with intent to injure, disorient, incapacitate and even kill, to the barely
noticeable, nauseating effect of low and almost inaudible frequencies, used as
a deterrent of loitering behaviour. The Long Range Acoustic Device [LRAD]
(LRAD 2015), for instance, can be used to transmit warning messages at long
distances, but can also be used with the intent of transmitting pain-inducing
and harmful sounds in a 30° beam at 2.5 kHz. Its applications range from the
deterrence of wildlife from industrial facilities to counter-piracy maritime law
enforcement and crowd dispersion at public demonstrations and events. The
LRAD is now widely used by governments as an effective means of control, even
within their own territory and against their own population (Thomas 2012). In
the commercial sector noise is not only used as alarm, but as a subtle deterrent
device based on the diffusion of high frequencies audible only to young people,
and aimed at dispersing loitering, anti-social behaviour and vandalism near
shops (BBC 2008; Campbell 2008).
Juliette Volcler’s Extremely Loud, Sound as Weapon (Volcler and Volk 2013)
gives a thoroughly researched account of the military and commercial history
of sonic weapons, spanning a period starting with the use of loudspeakers on
the battlefields of the Second World War, used with the intent of deception
and psychological abuse, through to the development of deafening infrasonic
grenades during the 1960s and 1970s, and including the most recent developments
emphasizing ‘non-lethal’ technologies as a means of bypassing public opinion
on torture in the ‘war on terror’. ‘Sound cannons’ have seen deployment as crowd
control and dispersal devices, in both Wall Street and Gaza, and were on standby
also during 2012 London Olympics.
Military deployment of sound, has long been made use of in fittingly called
theatres of operation; notoriously in Iraq during 2003 and 2004, where trucks
with loudspeakers broadcast ‘harassment operations’ that consisted in playing
hard rock, heavy metal and rap for several days and nights on maximum
volume. The dramatic siege of Fallujah in 2004 (during which the United States
admitted also to using white phosphorous), the so-called ‘clash of cultures’ also
took the form of an acoustic battle: US loudspeakers battled for dominance of
the urban soundscape by broadcasting high volume AC/DC and Guns N’Roses
titles, in response to which the mullah’s broadcast chants of Al-lahu Akbar and
Arabic music. US military spokesman, Ben Abel compared ‘these harassment
missions’ in urban settings with the disorienting and confusing effect of a
‘smoke bomb’ (Volcler and Volk 2013, 104).
Also CIA interrogation techniques are known to have long relied on sound
in so called ‘no-touch torture’, relying on ‘the capacity of sound and music
to destroy subjectivity’ (Volcler and Volk 2013, 104; Cusick 2006; ‘White
Phosphorus: Weapon on the Edge’ 2005). But as Volcler points out, using Axel
Schafer’s expression: ‘hearing is touching form a distance’. Sound is a mechanical

vibration, whereby a pulse is transmitted thanks to the oscillation of molecules
or atoms, whose intensity is measured in watts per square meter, whose pressure
is measured in pascals and whose amplitude is measured in decibels (dB),
reaching the human pain threshold at about 140 dB, while respiratory problems
become severe at 150 cB and frequencies between 50 and 100 Hz and inaudible,
infrasonic, sounds (below 20 Hz) are deemed to be potentially fatal from 174 dB
(Volcler and Volk 2013, 8, 14, 28).
This, certainly, is where the analogy between acoustic and statistical noise
breaks down, if the latter relies, as Weaver does, on the distinction between
the intentional (information) and the accidental (noise). Wherever extremely
loud or uncomfortable audible noise is used as a threat or even weapon, used
with intent to injure or kill, the message is certainly clear: noise becomes a
perverse form of information, in the sense of an imperative signal, while its in-
capacitating properties retain the characteristics of noise, not information.
Empirical Noise 159
As we have seen, Volcler specifies that sound, even when used with the intent
to disturb or harm, is not equivalent with the acoustic definition of noise, which
is confined to an aleatory wave with equal distribution of frequencies. In the
specific context of defining the toxicity of noise, however, it still makes sense to
speak of noise rather than using the generic term sound. Consequently, reference
to Volcler’s insights into the military and commercial use of sound with intent to
harm will feed directly into our problem of conceptualizing noise.
The dominant idea of noise as audible disturbance, in the meantime, presents
us with a serious drawback, if it neglects the non-audible range of acoustic events.
It thereby fails to answer the challenge posed by the sophisticated use of noise
in the defense industry and commercial security, but also by the experimental
use made of noise in contemporary film, music and art. Although the difference
between audible and non-audible noise may appear to be little more than a
difference of degree or intensity, the consequences for the conceptualization of
noise are worth considering. For what is lost, when the core conceptualization of
noise is limited to the auditory range of acoustic events, is the full breadth of the
physical phenomenon, comprising sound pressure levels and frequencies above
and below hearing range, whose impact on health and cognitive performance
is well known, relating sub-base frequencies that fall in the barely perceptible
range of audible frequencies, as well as infrasound below 20 Hz, with states of
anxiety and nausea, if not physical harm.
Laura Wilson, for instance, analyses the strategic use of low frequencies
in avant-garde cinema, with the intention of causing physical and emotional
unease, provoking in the viewer what Wilson calls a ‘physical spectatorship’.

The conscious processes of perception and cognition and their cultural
coordinates are intentionally subverted by inaudible noise, imposing an
involuntary physiological response that can be disconnected from to the
visual spectatorship of the witnessed scene. Wilson gives the example of a
rape scene in the film Irreversible by Gaspar Noé, 2002, where noise in the
range of barely audible frequencies constitutes an onslaught on the cultural
dominance of vision and hence, implicitly, on the role of the voyeur (P. Wilson
2012).
The profound effect on the viewer of barely audible noise is here used
deliberately as a critical and subversive technique, in order to high-jack the
cultural dominance of vision, and of the male gaze in particular – meaning here
not the male gaze as a form of self-perception of dominance in male spectators, or
their potential identification with the rapist, but more generally the dominance
of vision associated with a sense of power and control, which is culturally
codified as a tacit assumption that this power is one of male dominance over the
object of his gaze.
The strategic use of non-audible noise is thus a way of extending the viewing
experience beyond the predominance of seeing, but also of subverting conscious
and culturally formatted perception processes. This subversion of conscious
processes of perception in the service of creative practices, gives us a wider
angle on the problem of noise than that associated typically with the perception
of unwanted audible sound, extending the latter to the full spectrum of sound-
pressure levels and frequencies and their effects on us. Also the critical limit
between conscious perception of phenomena and the pre-conscious substrate of
perception is called into question and rendered evident by its strategic artistic
manipulation.
However, the anxiogenic effect of barely audible and inaudible noise features
not only in artistic practices – as a means of critique of our faculties and cultural
codes of perception – but also in audio-visual media developed for mass
consumption more generally. Artists, the defence industry and commercial
interests in mass media alike, compete in the use of non- or barely audible noise,
with awareness of its potential to override the rational and culturally codified
criteria of perception. By targeting the preconscious substrate of perception,
and manipulating the physiological and affective dispositions of those exposed,
audible and inaudible noise becomes effective in bypassing the cognitive
functions not only of individuals, but of groups and potentially of entire
populations.
Steve Goodman’s Sonic Warfare – Sound, Affect and the Ecology of Fear in
fact argues for the idea of a ‘sonic ecology’, less in terms of noise pollution, than
with respect to what he sees as a wholesale assault on perception, by means
of technological mechanisms of fear production. In addition to the already
mentioned use of noise as weapon or threat, and even artistic subversion of
traditional codes of perception, Goodman draws our attention to the urgent
need for a critical understanding of the way in which acoustic ambience in
general is being manipulated – ranging from branding experiences to the
induction of a general sense of unease, and even of fear or dread, notably by
mobilizing the periphery of auditory perception or what he calls the ‘unsound’ of
vibrational environments. Interestingly, he speaks of a transduction to describe
the propagation of affective tonalities that modulate collective dispositions of
fear and anxiety, and thereby potentially ready the ground for the reception of
ideologies (Goodman 2012, xx):
Empirical Noise 161
As film sound designers know only too well, certain frequencies can produce
an affective tonality of fear in which the body is left poised in anticipation,
expectant of incoming events: every pore listens for the future. Just think of the
uneasy listening of atonal or discordant sound, or the sense of dread induced
by low- frequency drones. […] Unlike an emotional state, affective tonality […]
envelops a subject […] short- circuiting […] attention and consciousness […]
(Goodman 2012, xx, 34).
Goodman sees in the interplay of analogue and digital technologies the

articulation of vibrational substance and information, a ‘sensual mathematics’,
evolving into an ‘ecology of code and vibration’ (Goodman 2012, xix). At stake
in the notion of a sound ecology is thus not the question of ‘noise pollution’
as the unregulated production of sounds in an industrial and post-industrial
world, but rather the purposeful manipulation of audible and inaudible acoustic
properties, the deliberate creation or distortion of ambience, with the intent
of overriding rational cogency and conscious perception. Dean Lockwood
comments on Goodman’s notion of sound ecology by describing it as
‘an affective capture of the materiality of the body [via a] sonic ecology of fear in
which we are collectively controlled, kept on edge, by the cultivation of a chronic
affect of dread’. (Lockwood 2012, 74)
We may or may not agree with Goodman and Lockwood on the shift from the
evidence of a weaponized noise industry to their much more general hypothesis
that appears to include a conspiracy to orchestrate large-scale affect-modulation
through the targeted manipulation of our soundscape. However, what emerges
clearly, is that noise is no longer only a question of reason, of ratio, in the sense
of calculation, (as the statistical measure of noise suggest), but also points to
a reality that engages pre-conscious levels of perception, which in turn may
become the object of targeted manipulation of perception’s affective disposition.
The well-worn metaphor of noise as parasite in communication technology
here finds its almost physiological counterpoint. Goodman indeed insists on the
epidemiological metaphor to describe the contagion of affect as a viral process
that affects groups or populations. He describes the use, for instance, of jingling
noises in advertising and its evolution into corporate sonic branding, notably
through earworms, as an ‘affectively contagious radiation of sonic events
through the networks of cybernetic capitalism’ (Goodman 2012, xix).
It is no longer clear whether this manipulation of ambience and perceptive
disposition counts as sound, noise or even information, (be it in the form of
veiled imperatives to buy and consume, like earworms). What this stretching
of the objective parameters of noise beyond the audible means for us, is that
whatever noise is, it can no longer be understood merely as something added to
the field of perception – be it intentionally or not. It is perception itself, rather
than that which is perceived, that is subject to contamination.
Although no clear definition of noise emerges from these new parameters,
thinking about audible and non-audible sound in the context of intentional
manipulation of the soundscape does have one important consequence. It means
that our focus must change from what is perceived, to the act of perception. This
act of perception is what is vulnerable – either to mechanical damage of the
hearing apparatus or to the short-circuiting of conscious processes of selection
and of pre-conscious physiological and affective dispositions.
The conceptualization of noise is thus no longer limited to the classical
philosophical problem of determining what we can understand of the reality
of noise ‘in itself ’ or even ‘for us’. It is irreversibly contaminated by a political
problem, which is the possibility of deliberate distortion of our critical faculties
through noise.
Goodman’s proposition to think in terms of a ‘sound ecology’, his recognition
that sound and what he calls ‘unsound’ is correlated with individual and
collective affective disposition, has implications that cannot be contained in the
mere quantification of acoustic phenomena as audible or even inaudible noise.
Noise is no longer merely a question of defining tolerable ranges of audible
frequencies and sound pressure levels, of finding optimum signal-to-noise ratios
to measure the effect of noise on communication pathways. It becomes an issue
of vital self-determination, be it at the psychological level of affect, or, as we have

seen with Bijsterveld, at the level of sociopolitical cohesion.
Notes
1 A part of this chapter is drawn from a previously published article (Malaspina

2014).
2 ‘[…] was er vorfindet, ist kein bloßes “Material”, absichtslos vor ihn hingekippt, in
reiner Objektivität, untouched by human hands. Im Gegenteil. Alles was hier steht,
ist durch viele Hände gegangen, zeigt Spuren des Gebrauchs.’
3 apostate, mid-14c., from apostenai ‘to defect’, literally ‘to stand off ’, from apo- ‘away
from’ (see apo-) + stenai ‘to stand’, (ww.dictionary.reference.com/browse/apostate).
4 A security is a tradeable financial asset that entitles either to a share or stock
ownership in a publicly traded company, or the share of a debt repayment, (which
Empirical Noise 163
can take the form of a bond that entitles the owner to interest and repayment of a
mid- to long-term loan), or a derivative such as a future, which commits to buy or
sell an asset at a certain price at a certain date in the future (and thus rises or falls in
value as real prices of the commodity fluctuate), or an option which entitles to the
same right as a future, but without obligation to buy or sell.
5 CDOs are derivatives of securities, so-called, because they derive their value from
an underlying asset, such as entitlement to the future repayment on auto, credit card
or mortgage loans (MBS, mortgage-backed security) or corporate and business debt
(ABCP asset-backed commercial paper). Another type of derivative is a swap, where
one asset (i.e. a debt, currency or interest rate) can be swapped for another, for
instance to insure against the default of asset- or mortgage-backed securities (such
as MBS or ABCP).
6 Mortgage-backed securities (MBS) are a guarantee of repayment of a loan that can
be sold as a financial asset, and from which collateralized mortgage obligations
(CMOs) can be derived, which essentially bundle a pool of mortgage-backed
securities (organized by date of maturity and level of risk, different principal
balances, interest rates, maturity dates and rise of repayment defaults). As borrowers
repay the mortgages that act as collateral on these securities, principal and interest
payments are paid to investors based on CMO terms. The value of these financial
assets fluctuates with interest rate changes, refinancing and foreclosure rates as
well as house prices. CMOs are a subcategory of more general collateralized debt
obligations (CDO) including mortgages, bonds and loans, but also CDOs derived
from CDOs called CDO2.
7 ‘Probability theory, Encyclopaedia Britannica’. Britannica.com.
8 The height of parents, for instance, henceforth explains but does not determine the
height of children: the average height of sons of a same father is a growing linear
function of the father’s height, but the dispersion around this average is independent
of the father’s height, whereby dispersion of heights of all sons is equal to that of all
fathers (Desrosieres 2006).
Part Three
The ‘Mental State of Noise’

I
The Crossroads: Mathematical, Technical,

Empirical and Subjective Noise
The problem of noise now reaches a theoretical crossroads, characterized by

a dynamic correlation of scientific formalization, technological regulation,
psychosocial convention and, not least, the act of perception (understood as
an act of self-determination of criteria of pertinence, vulnerable to affective
dispositions). This conceptual crossroads is relevant for anyone who wants to
understand the transdisciplinary appeal of noise. It is what accounts for the
ease with which the concept of noise has conquered general and specialized
discourse in a great array of theoretical and experimental domains. It requires
us to conjugate artificial and natural variables of noise, its digital and analogue
dimensions, its rational and affective parameters, its conscious and pre-conscious
perception, no less than individual and collective dispositions.
The relation between cognition and the real it attempts to cognize constitutes
the speculative dimension not only of the experience of noise, but of any
scientific concept of noise. The problem raised by Goodman’s notion of ‘sonic

ecology’ is that of recognizing within the process of cognition also the possibility
of inclinations, of affective dispositions and hence also of the possibility that
cognition is ‘contaminated’ by a pre-cognitive ground of experience. When
noise is thought in epidemiological terms and presented as parasitic upon the
conscious processes of perception of its host or, importantly, host population,
then what we are potentially dealing with is a bio-politics of noise in the
Foucauldian sense. The critical problem is thus not to determine noise ‘in itself ’
or even ‘for us’, but emancipation: because it concerns the power of judgement
and the power of control over its pre-cognitive ground.
We have seen that failing to expand the conceptualization of noise beyond the
audible phenomenon means that the concept of noise is unable to encompass
the full breath of the experience of noise, including its strategic use where noise
below hearing range is employed as artistic device or, in the case of commercial
and military applications, as a deterrent or weapon. What the use of noise below
the audible range emphasizes, is the continuing relevance of the etymological
twists and turns that have led to the contemporary notion of noise, which leads
us via the Latin term nausea, back to the Greek ν α υ σ ί α [nausia].
It is this aspect that is most relevant for us to pursue, because it leads to the
core of the conceptualization of noise. If noise can be argued to affect cognitive
and pre-cognitive processes, then the conceptualization of noise touches on
more than the quantitative measure of sound volume or frequency, no matter
how sophisticated, on more than statistical analysis, and certainly on more than
mere aesthetic appreciation or personal taste. Noise becomes a philosophical
problem, when it has to be factored into the conditions of possibility of cognition
itself. For it is hard to see how the conditions of possibility of rational thought
can be engaged with, without enquiring also into the distortion of cognition. It
is thus not only the impact of noise on cognition, as extraneous factor, but the
role of noise within the process of cognition that is at stake. It requires, in other
words, that we think of noise not as object of thought but as a variable within the
process of thought.
This correlation between noise and cognition, between noise as distortion
of information and noise as a factor of the distortion of cognition, emerges
as an important aspect of the conceptualization of noise. Any philosophical
enquiry into rationality, human agency and collective self-determination
must therefore arrive at an understanding also of the state of indecision and
confusion associated with noise – a state to which information and knowledge
are temporary and always fragile solutions. Any epistemological enquiry into
the nature of knowledge, finally, must contend with the role of noise as lived
ambiguity, indecision and error.
II
Internal Chaos, Terror and Confusion
Flectere si nequeo superos, acheronta movebo.

(If I cannot bend the higher powers, I will move the infernal regions.)
Aeneid, Virgil
In 1986 two colleagues at the Massachusetts Mental Health Center wrote an

article entitled ‘The Concept of Noise’, pressing ahead where the WHO will
years later still deplore lack of scientific investigation into the relation between
noise and mental health (Sands and Ratey 1986). In it, then clinical instructor in
psychology Steven Sands and Assistant Professor in psychiatry John Ratey, both
in the Harvard Medical School’s Department of Psychiatry, set out to add a new
concept to clinical assessment: that of the ‘mental state of noise’. The authors
hoped to show that the concept of noise is pertinent not only in diagnostic and
therapeutic terms, but as an epistemological category for research into mental
health.
In practical terms the article sets out to articulate a psychiatric concept of

noise. It provides a starting point and incentive for research into the concept
of noise in view of therapeutic measures of psychiatric conditions, and more
generally as a conceptual contribution to medical epistemology. The idea of noise
is thereby brought to the foreground of attention and theorized as a hitherto
unrecognized common denominator of psychiatric illnesses. The authors
suggest that nosology, or classification of psychiatric illnesses according to sets
of symptoms, may change profoundly once the role of noise is conceptualized
and taken into account as transversal to most mental illnesses.
The article begins with the author’s proposition of a general definition
of noise. Today, thirty years after the article was written, many a white-collar
worker, with high cognitive load and at risk of ‘burn out’ (Schaufeli, Leiter and
Maslach 2009), may recognize some aspects of the ‘mental state of noise’, whose
definition is here still reserved for very ill psychiatric patients:
Noise is a term we are using to describe a complex and distressing aspect of the
bodily and cognitive experience of many very ill psychiatric patients. By ‘noise’
we mean an internally experienced state of crowding and confusion created by
a variety of stimuli, the quantity, intensity and unpredictability of which make
it difficult for individuals so afflicted to tolerate and organize their experience.
Attempts to do so may only add to confusion and psychotic phenomena. (Sands
and Ratey 1986, 290)
In order to arrive at a concept of noise relevant to the domain of psychiatry,

the authors draw concentric circles around the initial definition quoted above,
situating the concept of noise within ever wider limits to the theoretical context.
The issue of the psychiatric aspects of noise is thereby raised in the contexts
of developmental and evolutionary psychology, at the level of learning and
memory, as well as affect and social relations. The authors’ references span as
wide as urban noise studies, psycho-analytic considerations and even art theory.
Towards the end of the article we come to understand the extent to which
the authors’ interest in noise is guided by research into to the pharmacological
containment of the effects of the ‘mental state of noise’, when placed alongside
other therapeutic measures. This is the perspective that, in practical terms, orients
their need for a more rigorous understanding of noise. The pharmacological
control of noise is what finally articulates the problem of noise in light of the
amplification of the response to ‘the mental state of noise’, via feedback in the
nervous system.
Referring to the nervous system as a self-regulating system with feedback
here appears to place the idea of the nervous system’s homeostasis, (i.e. of self-
regulation as the ‘internal milieu’ of a living organism), into the logical framework
of the cybernetic theory. Cybernetics theorizes machines with self-regulation
through feedback and extends findings from control theory in mechanics
beyond the machine paradigm to the analysis of living systems with homeostasis.
Yet despite the apparent closeness, noise is never defined by the authors in
cybernetic or information theoretical terms. While the understanding of noise
is clearly extended beyond the common understanding of acoustic noise, and is
couched explicitly in more general terms of systems, there is only one mention
of Norbert Wiener, who is at once acknowledged and dismissed by emphasizing
the ‘different’ role noise plays in psychiatry. This difference, however, is stated,
but not rendered explicit.
Sands and Ratey’s perspective on the pharmacological containment of the ‘mental
state of noise’ nevertheless testifies to an implicit cybernetic inclination towards the
problem of noise in psychiatry, as noise ultimately becomes a problem of control of
The ‘Mental State of Noise’ 171
physiological feedback and amplification of arousal in the nervous system. What we

can glean from this fleeting, if uneasy, reference to cybernetics is the malaise towards
a reductive mechanistic paradigm that has, thirty years after the publication of this
article, become common currency in cognitive neurosciences (Wiese and Metzinger
2017).1 Cybernetic theory, before effectively gaining the allure of a paradigm that
explicitly structures research, has thus gone through a phase of latency, where its
claims are not yet accepted as authoritative, but where the theoretical field is readied
through a semantic expansion, a subtle change in discourse.
Sands and Ratey thus chart multiple dimensions of the conceptualization of
noise, but stop short of an explicit reliance on cybernetic theory. Their malaise
towards cybernetic reduction is perhaps symptomatic of the tension that arises
from this multiplicity. What could at first give the impression of eclecticism,
accidentally reveals the real philosophical stakes of the problem of noise: the
non-reductive articulation of biological, psychosocial and pharmacological,
but also technocultural perspectives. In the process this article becomes an
opportunity for us to hinge the philosophical understanding concept of noise,
on problematizing this theoretical field of tension.
It is the difficulties the authors face in conceptualizing noise, difficulties
arising from the plurality of perspectives they engage, that will inform us as
much as what they achieve in this paper. These difficulties become symptomatic
for all other attempts to conceptualize noise, insofar as they are situated at a
critical junction. The conceptualization of noise here touches upon the core
of its epistemological relevance. As it oscillates symptomatically between the
perspectives of object and subject, it presents itself alternately as object of

cognition and as a factor in the process of cognition. The difficulty of stabilizing
the perspective on noise reveals something essential about the subject of the
theory of knowledge.
The experience of ‘self ’ and of control over one’s own experience is seen in
a new light only when it is in crisis. It is when the sense of self and of control
breaks down that some fundamental cognitive and pre-cognitive factors step
into the foreground as a problem, while they operate silently in the background
of cognition during ‘normal’ functioning. If René Leriche could famously say
that ‘health is the silence of the organs’, Sands and Ratey draw our attention to
the philosophical implications of the fact that mental health is the silence of our
embodied cognitive faculties. As a consequence, their article helps contribute a
new aspect to our understanding of the preconditions for rational discourse –
and a fortiori for the possibility of rational discourse without error or ambiguity,
in other words ‘without noise’.
It is thus important to situate the relevance of Sand and Ratey’s article in view
of this epistemological question of noise. Even if the new understanding they
propose of noise as a mental state dramatizes the embodiment of perception,
it is not a phenomenology of the acoustic experience of noise that we seek to
enrich here with recourse the psychiatric notion of noise. Nor is it in view of
an existential philosophy of noise that we call upon Sands and Ratey’s article,
such that the individual becomes the core reference of the concept of noise or
even of ‘epistemological noise’. However, this crisis will help us to think about
normativity more generally, including in its collective dimension, at the edge of
reason.
The relevance of Sand and Ratey’s proposition for thinking about noise
as a mental state is that of highlighting a blind spot in the modern theory
of knowledge, which existentialism was not alone to address: our modern
assumptions about rationality are built on the Cartesian presupposition of a
coherent self. Kant helped us to specify this presupposition, by showing that
rationality rests on the universal structure of apperception, preformatted by a set
of unchanging transcendental a priori.
Philosophy’s critical and self-critical method has long called the subject
of rationality into question. It was notably in response to Hume’s scepticism
that Kant sought to stabilize the foundations of rational thought. Marx, Freud
and Nietzsche have each taken a brick out of the monolithic edifice of the
classical subject of reason, followed by the structuralist analysis of linguistics,
of anthropology and of Lacanian psychoanalysis, which, together with post-
colonial theory, revealed the subject of modern reason to be a constituted

subject, before being a subject constitutive of reason and of the world it thinks.
These traditions of critique have arrived at an anti-humanist understanding of
the subject – anti-humanist not because against the human, but against classical
humanism, whose hybris loomed large against the background of the traumas
of the world wars, of Stalinism and colonialization (Alliez 2017). Analytical
philosophers like Wittgenstein and Feyerabend have rattled at the scaffolding of
rationality no less effectively – only to find that the presuppositions of rational
discourse and the legitimacy of consensus are far from set in stone.
Our reading of Sands and Ratey’s article thus serves the purpose of revealing,
through the tensions and contradictions that arise within it, some of the fault
lines and paradoxes that the concept of noise reveals within the constitution of
the rational subject. Far from dealing with noise as mere object of perception, we
will see that noise is also what un-conditions the capacity to discern and evaluate
the object of perception. Not just a state of confusion and indecision, noise is
also amplified by the panicked attempt to redraw the boundaries of the sense
of self. To regain the sense of self as first object of cognition thereby becomes
the precondition to reasserting its relation with other objects of cognition. At
stake, in other words, are not the noises we perceive, but the noise of cognition
constituting itself, against the always looming crisis of its dissolution.
The mental state of noise is first defined by Sands and Ratey as ‘internal
chaos’, even as ‘inner confusion and terror’. The feeling of being overwhelmed is
correlated by the authors with psychiatric symptoms that range from a lowered
level of adaptation, to various forms of psychopathology: perceptual distortions,
impulsive actions, impaired functioning and increased physiological stress
are only some of the coordinates of the mental state of noise. Sands and Ratey
begin their analysis by listing classic studies in evolutionary and developmental
psychology that position noise as a fundamental correlate of avoidance
behaviour and threat. A series of empirical studies in evolutionary theory of
perception indeed correlates the effect of bright lights and startling, loud noises,
with avoidance mechanism observed even in unicellular organisms.
The conceptualization of noise is thus placed on a par with basic evolutionary
concepts, in the sense that all organisms, including unicellular ones, are said
to display withdrawal behaviour in response to intense stimulation by light or
sound (Sands and Ratey 1986, 290; Schneirla 1959). The avoidance of noise is
thus posited as a basic, even evolutionary aspect of perception (Schneirla), which
is then linked by the authors with theories of perception and developmental
psychology of human infants (Watson and Rayner 1920). The negative reaction
to noise is similarly posited as a basic developmental disposition. Sands and

Ratey mention J. B. Watson and R. Raynor’s observation that human infants
respond with fear to ‘loud noise, unpredictable shock, removal of familiar forms
of support’ (referring to Watson’s controversial experiments with ‘little Albert’
(Harris 1979)).
Starting with the rejection of startling loud sounds as a basic feature of the
experience of noise, Sands and Ratey move on to refer also to D. C. Glass and
J. E. Singer’s theory of stimulus-overload from the environment. Unpredictable
or uncontrollable stimuli are identified in the context of urban noise as a social
stressor, associated with the experience of ‘internal chaos’ and correlated with
consequences ranging from personal distortions, impulsive actions, impaired
functioning and increased physiological stress (Glass and Singer 1972; Sands
and Ratey 1986).
It is, however, not only external stimuli coming from the ‘outside world’ that
come into the equation of ‘the mental state of noise’. The crux appears to be
that the ‘mental state of noise’ arises as a correlation between environmental

stimuli, which act as extrinsic stress factors, and the individual’s own, intrinsic
capacity to deal with the complexity of cognitive tasks. Sands and Ratey here cite
Norman’s 1968 study on the ability to ‘filter’ information for pertinence: while
criteria of pertinence vary in relation to momentary interest (one’s own name,
for instance, tends to remain pertinent at all times), individuals vulnerable to the
‘mental state of noise’ appear to have in common that the critical threshold of
attention, which sets the criteria of perception, is inoperative or impaired.
This critical threshold of perception is what cuts through the continuum of
sense data. Where it appears to be lacking or defective in some individuals, they
become vulnerable to mental disorders. It is, in Sands and Ratey’s words,
as if patients vulnerable to the chaos of overloading that we are calling noise are
always wide open. (Sands and Ratey 1986)
The image that is evoked here is one of a lacking boundary that would otherwise
protect the subject from the excess of external stimuli. As if the vulnerable
subject was a fortress whose drawbridge is always ‘wide open’. To a certain extent
we could infer, from the use Sands and Ratey make of this notion of threshold,
that such a filter of perception is cognition’s own pre-critical boundary, filtering
stimuli according to pre-conscious a priori. These pre-conscious a priori act as
a perceptive firewall, separating out not only superfluous stimuli, but thereby
ensuring the very condition of perception: a stable sense of self. The excessive
openness, defined as vulnerability to noise, is subsequently associated by Sands
and Ratey also with the individual’s diminished confidence in the ability to
organize his or her experience adequately. Sands and Ratey refer again to Glass
and Singer’s studies, in which the stressful effect of external stimuli is found to
be proportional to self-perception as ‘helpless’.
The ‘mental state of noise’ is thus posited, in the first instance, not only as the
result of the ‘crowding’ of sensations due a failed critical threshold of perception,
but as noise in the sense that it is amplified by the fear of losing this critical
boundary, in other words of losing control. The fear of disintegration of the
sense of self is thus also a consequence of the inability to impose a critical limit.
Loss of confidence furthermore implies the threat of losing a reliable sense of
self. Loss of boundary, loss of a defined sense of self, loss of control, thereby
emerge as a chain reaction amplified by the loss of confidence that leads to the
heightened sense of vulnerability that characterizes the ‘mental state of noise’.
However, it is not only the correlation of external stimuli and internal
disposition, which becomes relevant to understanding the ‘mental state of
noise’, but the changes of external stimuli over time and the changes of internal
disposition over time conjointly modulate the experience of ‘the mental state of
noise’, potentially progressing from confusion and anxiety to what Sands and
Ratey will identify as the ‘catastrophic reaction’.
The range of behaviours Sands and Ratey associate with this ‘catastrophic
reaction’ to the ‘mental state of noise’ may take the form of a wide variety of
behaviours and psychodynamic processes, spanning from ‘boisterousness to
fainting and passive weakness’, from ‘internal and social withdrawal to catatonia’
and ‘stereotypies’. Significantly, also ‘excessive orderliness’ is listed alongside
other behaviours, as a form of behavioural and cognitive withdrawal from noise.
In defining these behaviours as a ‘catastrophic reaction’, Sands and Ratey refer to
early-twentieth-century neuropsychiatrist Kurt Goldstein.
Observing soldiers returning from the First World War, having suffered
brain lesions or shell shock, Goldstein noted that even ordinary situations were
experienced as ‘catastrophic situations’. Brain damage and other types of physical
and psychological trauma were seen as increasing the vulnerability, not only to
noises commonly perceived as excessive acoustic stimulus, but even to ordinary
situations. Goldstein paid attention also to the patient’s feeling of inadequacy,
resulting from the radically decreased ability to deal with normal and ordinary
experiences. Together with the feeling of being defenceless against the stimuli
coming into the central nervous system, this feeling of inadequacy appeared to
contribute to the anxiety of impending catastrophe:
As soon as an excitation is felt that emanates from an objectively dangerous
situation, a catastrophic reaction occurs immediately, all other adequate

utilization of excitation is excluded and the ill individual appears completely
closed to the world. (Goldstein 1983, 37)
Goldstein’s clinical observations led him to correlate this perceived threat of

disintegration, not only of familiar patterns of experience, but of the individual’s
sense of self, with the individual’s adoption of a ‘rigid attitude’, intended to stave
off the state of mental confusion. This rigidity, in turn, was seen to aggravate
the catastrophic experience, provoking ‘disorderly, disharmonious, defective
performances’ (Goldstein 1948; Sands and Ratey 1986, 291).
Sands and Ratey’s conceptualization of the ‘mental state of noise’ is thus
referring to a mental state more complex than the idea of confusion and
overstimulation appears to suggest, more multifaceted also than today’s
attention to ‘information overload’. One of the factors of this complexity is the
correlation of extrinsic and intrinsic noise factors. This correlation leads to a
concept of noise that must encompass the nexus between both. This nexus,
moreover, cannot be reduced to that between two states, but must encompass
the correlation between two forms of duration: between the change over time
of extrinsic noise factors (for instance, the difference between sudden onset,
progressive increase or intermittence) and of intrinsic noise factors (progressive
openness, disintegration of critical pre-cognitive thresholds of attention, erosion
of boundary, of sense of self, progressive loss of control and loss of confidence).
What makes this correlation more complex is that also a tipping point must
be taken into consideration, where the correlation between stimuli and internal
disposition abruptly lead to a catastrophic reaction, which rather than ending
or even alleviating the ‘mental state of noise’ in fact amplifies it like feedback
between a microphone and a speaker.
What is required, therefore, if we understand the correlation between these
factors as a correlation of processes with a duration in time, is to see the ‘mental
state of noise’ as subject to a variable of variables, making the concept of noise
subject to a function of functions, rather than simply a form of excess of stimuli,
whose modulation is either extrinsic or intrinsic or a simple correlation of these
two factors. Its complexity can thus be reduced neither to the unpredictable or
intense nature of stimuli, or to an individual’s internal disposition, nor to a mere
summation of both. The experience of noise must instead be understood as a
stochastic effect of correlation between extrinsic and intrinsic variables, as a
complex function of functions that conditions or deconditions our experience,
rather than being reducible simply to an anxiogenic state of over-excitation,
overstimulation or confusion.
We could question in passing to what extent this subject, the subject of noise,
is based tacitly on sociopolitical assumptions about male presuppositions of
‘self ’. Let us pause, for argument’s sake, on a sociopolitical aspect that is often
neglected in scientific discourse. Is not, at least in evolutionary terms, the
extremely low threshold of pertinence of a mother’s perception of stimuli the
very condition of survival of an infant, soliciting response to the infant’s every
minute face expression, to every change in the infant’s behaviour even during
the mother’s sleep? Is not the perpetual attention of a mother, presumed always
mentally available for her offspring, taken by psychologists and psycho-analysts
to be the ground for the latter’s satisfactory development and insertion into
society? A society that, in many parts of the world, politically structures childcare
so that mothers must refrain from full insertion into the labour market and stay
continuously available for their offspring, while men can afford a more selective
attention threshold? Is not the very definition of a woman’s ‘self ’ premised on
being always ‘wide open’ to her children’s needs, to her partner and parent’s
needs, to her community’s needs?
It would seem then, that the mental state of noise is perhaps not a neutral
criterion for the assessment of mental health in general, but that it is an implicitly
gendered one, whose sociopolitical aspects are ignored as a matter of discursive
convenience: the mental state of noise is perhaps a normal state of mind for
many women who assume tasks with a high cognitive load, while assuming
also the traditional role associated with child rearing. It becomes pathological
or abnormal, when it impedes the traditionally male privilege of focus. The
loss of confidence and of a stable sense of self may indeed be a more common
experience for many women raising a family in modern patriarchal societies
than is often aknowledged – indeed perhaps more so as career expectations
increase the tension between the requirement for openness and the requirement
for focus.
Openness is thus more likely to be correlated with pathological aspects of the
‘mental state of noise’ when the prerequisite for survival in a capitalist economy
based on cognitive labour is threatened: namely male focus and confidence.
However, it is perhaps not only a question of gender, but also more generally
one of care. Men and women alike can be subject to a requirement of ‘openness’
experienced as excessive in an economy with high cognitive load, notably in the
medical profession. The latter appears to put also men on a spectrum of required
‘openness’ to a multitude of solicitations and others’ needs, comparable to what
is traditionally the case for women. The prevalence of mental health issues
associated with the medical profession, unsurprisingly, appears to fit some of the
criteria for the ‘mental state of noise’, as indicated by a recent survey published
in the British Medical Journal: ‘A recent Medical Protection survey of over
600 members revealed that 85% have experienced mental health issues, with
stress, anxiety, low self-esteem, and depression being among the most common
complaints’ (‘BMJ Careers – Doctors’ Own Mental Health Issues’ 2017).2
III
The Vicious Whir of Sensations
Sands and Ratey go on to compare the ‘mental state of noise’ in the adult to the
‘vicious whir’ of sensations supposed to be experienced by the infant, due to
an initial lack of cognitive differentiation. Following Piaget, Sands and Ratey
draw attention to the infant’s lack of cognitive differentiation, as a result of
which neither stable representation, nor continuous memory can as yet serve
as grounds for a stable sense of self. The low degree of cognitive differentiation
and hence low capacity for integration of experience’s ‘component parts’ are
deemed to expose the infant to an experience that can be compared, in Sands
and Ratey’s view, to the excessive ‘openness’ of those who suffer from the ‘mental
state of noise’. Both the infant and the mentally ill are thus compared in their
vulnerability to a feeling of being overwhelmed and powerless in the face of
unpredictable and uncontrollable stimuli.
Sands and Ratey cite Phyllis Greenacre’s 1952 study on trauma, growth
and personality in the earliest period of life, in which she defines the infant’s
helplessness in coping with what she calls the ‘vicious whir’ of sensations, as the
result of the immaturity of the sensorial and cognitive systems. Greenacre calls this
basic form of bodily and psychic distress in the infant ‘pre-anxiety tensions’, which
are compensated only by a ‘holding environment’ that facilitates the integration
of early experience. This ‘holding environment’ in turn would act as a foundation
for future organizational capacity (Greenacre 1952; Sands and Ratey 1986, 292).
Greenacre and Piaget are thus called upon by Sands and Ratey, in order to
compare the adult, in whom this differentiation is impaired or lost in the ‘mental
state of noise’, to the infant’s initial deficit in cognitive differentiation. Unstable
representation and the defaulting sense of self are thereby posited as affecting the
infant and the mentally ill in analogous fashion, as both are seen as ‘trapped in
experiential noise’, because the criteria for continuous representation, for stable
memory and hence for a consolidated sense of self are inoperative (Sands and
Ratey 1986, 295).
Sands and Ratey cite, alongside Piaget and Phyllis, also an author
contemporary to their work, whose observations suggested that schizophrenia
can be associated with a process of ‘de-differentiation’ of discourse (Frosch 1983;
Sands and Ratey 1986, 292). Now, how this research into the de-differentiation
of discourse holds up to more recent understandings of schizophrenia is less
important for us, than the fact that Frosch’s hypothesis serves to consolidate
Sand and Ratey’s hypothesis, namely that the ‘mental state of noise’ can be
associated with a form of regression, not only on the level of cognitive and
sensorial development, but also on the level of discourse’s regression the infant’s
undifferentiated stage of linguistic development. Sands and Ratey thereby place
the entire logic of the ‘mental state of noise’ under the authority of Freud’s
theory of regression: the return to the ‘vicious whir’ of sensations of the infant,
to the vulnerability to chaos and tensions characteristic of an immature stage of
neuropsychological development, and finally in terms of the de-differentiation
of discourse.
Suggesting that psychosis is related to the dedifferentiation of discourse, they
also see in this process a possible explanation for Schachter and Arieti’s clinical
observation that psychosis is associated with heightened ‘evaluative needs’. The
idea is that, in response to the dissolution of the threshold of perception, of the
sense of self and its discourse, the psychotic reacts with a ‘defensive searching for
a “name” or label’ for their experience. For Schachter and Arieti it is the ‘basic
need to understand one’s experience’, precipitated and amplified by anxiety, that
provokes psychotic delirium as a ‘premature flight into meaning’:
Schachter (1959) suggested that when the body is in a state of hyper-arousal,

there arises in the patient […] evaluative needs – in other words, a defensive
searching for a ‘name’ or cognitive label to help him explain and understand
his bodily feelings [leading to the] premature flight into meaning [as] a source
for psychotic explanations for sensations and associated affects that cannot be
recognized and integrated. (Lemaine 1960; Sands and Ratey 1986, 292)
The idea of closure of discourse, through a premature precipitation of meaning,

also recalls Goldstein’s observation of the catastrophic reaction, in the form of
a rigid ‘methodical character’ and obsession with order, the attempt to create
a focus and thereby stabilize the cognitive process. Yet despite attempting
to stave off the ‘mental state of noise’ through the closure of discourse and
methodical rigidity, the ‘catastrophic’ reaction fails to alleviate the tension and
the catastrophic attempt to organize experience ‘may only add to confusion and
psychotic phenomena’ (Sands and Ratey 1986, 290).
IV
Keat’s Negative Capability
Sands and Ratey go on to contrast the flexibility that, according to Piaget, is

essential to the child’s development with the ‘catastrophic reaction’. Indeed,
they will emphasize openness as an ‘essential human trait’, by referring to
art historian Meyer Shapiro. Shapiro was known to have taken great interest
in psychoanalysis and for having interpreted the conservative rejection of
modern art as indicative of a loss of this ‘essential human trait’ of openness.
He interpreted the conservative rejection of abstract modern art a classicist
‘reliance on details, for lack of a grasp of the gestalt’, which he compared to some
pathologies resulting from brain lesions. Shapiro saw in the conservative ‘need
for sameness’ the sign of an ‘impoverishment of an essential human trait’ (Sands
and Ratey 1986, 292).
Sands and Ratey thus make a more general statement about the catastrophic
reaction to the ‘mental state of noise’ than the initial clinical context for the
contextualization of noise appeared to suggest. Extending the psychiatric
conceptualization of noise in some way to a conservative cultural disposition,

Sands and Ratey go on to compare the child’s healthy attitude to what the poet
John Keats called the ‘negative capability’, by which Keats meant the ability to
approach new,
strange or confusing material without prematurely resorting to an armour of
pre-set attitudes or behaviours. (Keats 1958; Sands and Ratey 1986, 292)
This definition of a ‘negative capability’ as the opposite of both the conservative

reaction to abstract modern art and the catastrophic reaction, however, also raises
a problem in Sands and Ratey’s line of reasoning. Is there not an ambivalence
between this emphasis on openness as an ‘essential human trait’ and the negative
aspect of excessive openness, which first characterized the internal disposition
and vulnerability to the ‘mental state of noise’?
This ambivalence, in our view, is of the same nature as the one we encountered in
part one of this thesis, when we found that ‘information entropy’ can be evaluated
positively as ‘freedom of choice’ that augments the quantity of information while
also augmenting uncertainty, or negatively, when information is on the contrary
defined as negation of entropy, as negentropy, by pitting information as reduction
of uncertainty against both ‘information entropy’ and noise.
Much is at stake in this comparison because negentropy is also that which,
according to Brillouin, characterizes the degree of organization of living beings.
If one were to speak of a negative capability in the sense of negentropy, it
would be the capability of negation of entropy, the capability in other words, of
imposing constraints and critical limits and thereby negating openness as the
threat of organization’s liquefaction. Did Sands and Ratey not at first suggest that
the organization of discourse relies on the capability of imposing a threshold of
attention, in order to ensure stable structures of representation, of memory and,
as a consequence, of the sense of self?
Keats ‘negative capability’ on the contrary implies not this form of negation,
but the negation of this negation. It is worth going back to Keats’ letters in order
to reveal just how radical the consequences of this ‘negative capability’ would
be, in the context of Sands and Ratey’s article, if fully taken on board. The
‘negative capability’ is not only the ‘absence of an armour of pre-set attitudes and
behaviours’, but very much the negation of a sense of ‘self ’. It is the capability of
‘being in uncertainties, mysteries, doubts’, of accepting ‘half-knowledge’ serenely
because one ‘trusts in the heart’s perceptions’, negating only the closure and
concreteness of an ‘irritable’ attitude that ‘reaches after fact and reason’ (Keats
1958, 193–94). In other words, Keats’ negative capability essentially negates the
flight into meaning and the closure of the sense of self.
If we give Sands and Ratey’s reference to Keats’ ‘negative capability’ its full
weight in the definition of an ‘essential human trait’, then we must ask ourselves:
what does it imply if the essential trait is essentially negative? The ‘negative
capability’, for Keats, is far more radical than a tame liberal motto of refraining
from preconceptions. Keats goes one essential step further than being merely
‘open minded’, he takes one step further into the abyss of reason – the ‘negative
capability’ is essentially the courage of allowing the representative structures
of one’s own ‘self ’ to dissolve. In his 1818 letter to Woodhouse, Keats opposes
the poet to the virtuous philosopher, affirming that ‘[w]hat shocks the virtuous
philosopher, delights the chameleon Poet’. The poetical character, according to
Keats, relishes that he ‘has no self ’, ‘has no identity’ not even the ‘the egotistical
sublime; which is a thing per se and stands alone’.
If the reference to Keats’ ‘negative capability’ is to have any weight, then

the analogy between the ‘negative capability’, and the openness that Piaget
sees as essential in the child’s development, needs a critical proviso: for, in
contradistinction to the insouciance of a child at play, the ‘negative capability’
of the poet implies knowingly taking the risk of losing one’s mind, the voluntary
regression of the sense of self to a dedifferentiated state:
When I am in a room with People if I ever am free from speculating on creations
of my own brain, then not myself goes home to myself: but the identity of
everyone in the room begins so to press upon me that I am in a very little time
annihilated […] I will assay to reach to as high a summit in Poetry as the nerve
bestowed upon me will suffer.
There is thus a knowing risk involved in the negative capability which, when
we read Keats, is more radical than that of merely overcoming preconceptions,
because it makes strategic use of a process of de-differentiation of identity, which
alone makes poetry possible according to Keats. While the child may be open
by default, it happens upon an already structured environment (the ‘holding
environment’ mentioned earlier in the article as foundation for the integration
of experience). The poet, on the contrary, seeks out the absence of structure and
pre-conception, where society is already structured and closed.
Not only the dissolution of a sense of self is at stake in the negative capability,
‘not one word I ever utter can be taken for granted as an opinion growing out of my
identical nature’, but also the capability of others to tolerate the poet’s withdrawal
from identity (Keats 1818). It is not only the poet’s own anxiety, but that of his
addressee, which is the object of Keats’s letter on the ‘negative capability’:

I feel your anxiety, good opinion and friendliness in the highest degree. (Keats
1818)
The negation implied in Keats’ ‘negative capability’ in the service of artistic

creation is thus not negation of contingency, as that implied in the notion of
negentropy, but negation of the negation of contingency. It is negation not only
of the individual’s own negentropic needs for organization and stability, but
negation also of the negentropic needs of others for stable identities and thus
stable intersubjective relations.
If Keats’ negative capability truly represents the ‘essential’ human trait for
Sands and Ratey, then we must acknowledge that it stands in contradiction with
the individual’s ‘pertinence filter’ which Sands and Ratey cite in the beginning of
their article as requisite protection of the individual from the excessive openness
of the ‘mental state of noise’. The partitioning, regulation and judgement of
incoming stimuli can only occur on the basis of pre-conscious criteria of

pertinence, without which the individual would be flooded with stimuli, unable
to discern and judge what constitutes information or noise.
If we interpret this pre-conscious filter of pertinence as a ‘fire-wall’ of
perception, protecting healthy individuals from stimulus overload and
experiential ‘chaos’, then defining Keats’ ‘negative capability’ as an ‘essential
human trait’ means that it is an essential human trait to risk one’s mental sanity.
While Keats clearly thought that this opposes the poet to the philosopher, we
cannot help but recall here George Canguilhem’s well-known vindication of the
freedom of thought:
The norm in matters of the human psyche is the vindication and use of freedom,
as power of revision and of institution of norms, an assumption that implies,
normally, the risk of madness. (Canguilhem 2000, 217)
The paradox at the heart of Sands and Ratey’s article on the ‘mental state
of noise’ is that it refers at once to the negative function of a pre-conscious
selective threshold of attention and to the negation of any such pre-established
armour of perception, encapsulated in Keats’ ‘negative capability’ as an
‘essential human trait’. The ‘essential human trait’ thus appears to run counter
to the evolutionary and developmental avoidance of noise. It implies, on the
contrary, a voluntary vulnerability to the ‘mental state of noise’. By referring to
art history and poetry, Sands and Ratey appear to get more than they bargained
for, making the ‘mental state of noise’ a problem of culture that exceeds the
evaluative and therapeutic objectives of the clinical context with which they
set out.
We are, at this point, in the paradoxical situation that openness, flexibility
and abstaining from the critical and pre-critical faculty of negating contingency
constitutes an ‘essential human trait’, that this openness is both requisite for
cognitive assimilation during the normal development of the child and a fortiori
for creativity in the adult, in the form or the ‘negative capability’. Yet on the
other hand this openness implies also the absence of a critical faculty of negation
(which we could call a negentropic function in the sense of organization, certainty
and information), which is no less essential to mental health.
What qualified as a description of the mentally ill, ‘always wide open;
anything seems to satisfy their pertinence filter and they are thus prone for
flooding’, thus becomes the prerequisite for the child’s normal development
according to Piaget, for cultural progressiveness according to Shapiro, and
for poetic creation according to Keats – and even for philosophical thought,
if the inclusion of Canguilhem is granted (Sands and Ratey 1986, 291). If one
were to take this paradox to its natural conclusion, then one would have to
say, with Canguilhem, that in order to maintain one’s health one has to risk
one’s health.
V
Closure to Noise and the Paradox

of the Declining Life
What remains to be answered, then, is the difference between a declining life and
a thriving life. We have seen that openness of the as yet not fully differentiated
nervous system in the infant presumes an external ‘holding structure’ that
facilitates integration of experience’s component parts in the infant, alleviating the
‘pre-anxiety’ tensions caused by the ‘vicious whir of sensations’. A certain degree
of closure thus appears to be a precondition the infant’s cognitive development.
It is also what, in the form of a pre-conscious threshold of attention, keeps at
bay a ‘catastrophic reaction’ to the ‘mental state of noise’ in the adult. There is
a tension between the requirement for both openness and closure, and also a
risk inherent to both openness, which may provoke the ‘catastrophic reaction’
and closure, which risks impeding the openness required by development and
culture.
This problem complicates Sands and Ratey’s suggestion that the ‘mental state
of noise’ can be thought in Freudian terms as a regression. It is not stated clearly

whether this regression is meant to explain the similarity between the excessive
openness of the child and of the psychotic experiencing the ‘mental state of
noise’, or the similarity between the behaviours that are implied to characterize
the immaturity of the child (such as shouting, or withdrawal behaviour) and the
behaviour that characterizes the ‘catastrophic reaction’.
Yet what is clear, is that the infant’s ‘vicious whir of sensations’ is different
from the excessive openness in the ‘mental state of noise’, first of all because
the child is thriving on this openness, while for the mentally ill it is associated
with a decline in health. There is thus a flaw in the comparison between the
mentally ill and the ‘pre-anxiety tensions’ of the infant and, as a consequence, in
the idea of regression: even if the anxiety provoked by the ‘mental state of noise’
in the mentally ill is comparable to the ‘pre-anxiety tensions’ Phyllis observed
in the infant, the mentally ill cannot be said to regress to the infant’s normal
disposition to thrive and develop.
The tension between openness and closure, between progression and
regression, between extroversion (i.e. excessive openness) and introversion (i.e.
withdrawal) thus constitutes a fundamental difficulty that Sands and Ratey face
in the definition of the ‘mental state of noise’. This tension is dramatized by the
proneness to switch catastrophically from excessive openness to excessive closure.
The authors’ acknowledge this difficulty to a certain extent. After comparing
the mental state of noise to the infant’s ‘vicious whir of sensations’, Sands and
Ratey refer once more to Piaget, in order to highlight the difference between the
‘catastrophic reaction’ and the ‘flexibility essential to adaptive functioning’ in the
normal development of the child (Piaget 1999). The ambiguity appears to lie in
the status of openness in the infant and child, and in the adult experiencing a
‘mental state of noise’.
According to Piaget reflex and sporadic imitation serve the purpose of
initiating a learning process in the infant, by acting as a structural basis for
the consolidation of experience, upon which the child will be able to build
the capacity for symbolization and representation: after the first year of life
sensorimotor schemas begin, according to Piaget, to structure memory, which in
turn enables the transition to the formation of conceptual schemas, enabling the
assimilation of new stimuli and the accommodation of new experiences and new
behaviours. Piaget’s schema of systematic imitation, formation of sensorimotor
schemas, consolidation of memory and the formation of conceptual schemas,
thus provides several strata of what we could call structural redundancy.

If we rephrase Piaget’s notion of imitation, by interpreting it as a function
of redundancy informing the child’s development (redundancy being achieved
in the channel of communication by repeating a message or signal and used to
counter-act noise in the channel of communication), then the child’s openness
to imitate serves essentially a negentropic function. Imitation indeed appears
to constitute an information process that, like negentropy, reduces ‘freedom of
choice’. By laying the necessary structural conditions for sensory-motor schemas,
for the consolidation of memory and the formation of conceptual schemas,
imitation, like redundancy in the channel of communication, would appear to
make experience more predictable, in other words, it enables the child to learn.
In order to overcome the ‘vicious whir’ of sensations and create a basis for the
capacity for symbolization and representation, the infant could be said to reduce
the ‘information entropy’ of its initial sensorial openness through the repetitive
structure of imitation – decreasing uncertainty, by increasing redundancy.
What we could call the maximum entropy of the ‘vicious whir’ of the infant’s
undifferentiated sensations is thus to be reduced, in order to generate a stable
basis for a growing interaction with the environment. Imitation can thus be
understood as introducing a structural redundancy that transforms the pure
contingency, or ‘maximum entropy’ of the ‘vicious whir’ of sensations into the
‘relative entropy’ of what we can now call potential information (Sands and
Ratey 1986, 294).
Yet, if such is the open disposition of the child, open in other words to
imitate and acquire structural redundancy in the process of learning, then
the poet could not be more different to the child, at least insofar as Sands and
Ratey refer to Piaget and Keats. The ‘negative capability’ that characterizes the
poet’s openness is not a ‘default’ openness (it is not predisposed to growing
structural redundancy and organizational constraint of the initial ‘vicious whir’
of sensations), but on the contrary constitutes a knowing and willed undoing of
the certainties that have arisen on the basis, at least initially, of learning through
imitation and repetition.
Implied in the reference to the Freudian notion of regression was the idea
that the experience of the ‘mental state of noise’ affects the arrow of time, that it
bends the progression, the natural development of the subject, backwards: the
experience of ‘the mental state of noise’ would be like a time boomerang returning
the adult to the ‘vicious whir’ experienced by the infant. The catastrophic
reaction, in turn, would be a regression to coping mechanisms comparable with
those of a child. Logically, the opposite of a catastrophic reaction would then
be the idea of progress in healthy mental development: yet, if regression is the

return to a maximally un-differentiated state of openness, then progress would
logically consist in progressively closure.
Sands and Ratey’s reliance on the Freudian concept of regression thus raises
a problem analogous to the one we encountered when comparing Shannon’s
definition of information as ‘information entropy’ and Wieners negation of
the latter as negentropy: ‘information entropy’ becomes indistinguishable in
principle from noise, while negentropy ultimately means redundancy. It is
not clear, whether Sands and Ratey saw the contradiction in their concepts of
openness, that arises between the openness of infant, the openness that serves as
a basis to explain the vulnerability to the ‘mental state of noise’ in the adult, the
openness of the child according to Piaget, and the wilful openness of the poet.
VI
The Catastophic Reaction to Noise
The importance of Goldstein’s concept of the catastrophic reaction is evident in

the central position it is given in Sands and Ratey’s article. Beyond its importance
in psychiatry, however, Goldstein’s concept of the catastrophic reaction also has a
role to play in the philosophical conceptualization of noise which we undertake
here, because it is relevant not only for Sands and Ratey’s interpretation of noise,
but also for a particularly important moment in the history of twentieth-century
philosophy. It is well known how important Goldstein’s influence was on a book
that has become a corner stone of French epistemology: George Canguilhem’s
The Normal and the Pathological (Canguilhem 2009).
Canguilhem’s turn towards the philosophy of medicine has played a significant
role in the unfolding of contemporary French philosophy when, during the
1960s, it intersected with two other trajectories of contemporary thought, the
formal engagement with systems and their structure and an emancipatory
philosophy of politics. Canguilhem rendered evident the intersection between,
on the one hand, the value judgement inherent in the concepts of health and
disease, and on the other hand the disinterested search for scientific truth in
the concepts of the normal and the pathological. His analysis resonated with
the philosophical tension between the need to act and the pursuit of a formal
engagement with a structural understanding of systems that would erupt in
1968, amidst of a number of highly original philosophical projects.
In what follows we will develop some of the implications of Goldstein’s notion
of the ‘catastrophic reaction’. This serves the double purpose of deepening our
understanding of its role in Sands and Ratey’s concept of the ‘mental state of
noise’, but also of drawing out some of the philosophical consequences for a more
general view on the epistemological relevance of noise. Goldstein’s ‘catastrophic
reaction’ will serve as a basis notably to approach Canguilhem’s concept of
normativity, whose meaning is here not restricted to the power of the law, but
extended to the question of the source of our norms of thinking and living. In
the context of the philosophy of medicine, normativity, for Canguilhem, is first

of all the individual’s capacity to organize and, more importantly, reorganize
experience after a traumatic event or lesion. Rehabilitation therefore is not
merely a question of therapeutic or pharmacological reduction of noise, but of
facilitating the reassertion of subjectivity via the normative power to shape one’s
own experience.
At stake in Canguilhem’s concept of normativity is therefore not merely the
power of existing norms, and hence not the individual’s return to a previously
normal state, but normativity understood as the individual’s reassertion of his or
her power to act, judge and decide, in other words, the power to generate new
norms in answer to life’s contingent events. Grounded not in existing norms
but in the individual or even the organism’s capacity to invent new norms in
response to contingency, this normativity presents itself not as a function of the
understanding or reason alone, but as a gamble, of which Canguilhem famously
said:
[L]ife gambles against growing entropy. (Canguilhem 1991)
The concept of normativity can thus be understood as a covert way to pose the
problem of what makes an individual a subject, capable of judging, deciding
according to norms invented by him/herself – covert, because Canguilhem
guarded himself well from producing a theory of the ‘subject’ in a time when
existentialism and phenomenology occupied this philosophical terrain. Another
reason why there is no abstract theory of the subject in Canguilhem’s work is
that while the concept of a subject is linked to this normativity, the notion of
the subject itself is not reducible, for Canguilhem, to that of a conscious self, as
it is for existentialism and phenomenology. Nor can the subject be generalized
beyond the conscious self to an abstract notion, such as Bergson’s élan vital.
The subject instead remains a difficult and open question in Canguilhem’s work
(Badiou 1993).
The philosophical consequences we will draw from Goldstein’s concept of
the ‘catastrophic reaction’ will address Canguilhem’s concept of normativity,
which in turn will enable us to broaden the philosophical approach to noise, by
taking into account once more Shannon and Weaver’s definition of ‘information
entropy’ as freedom of choice, but also Sands and Ratey’s attempt to conceptualize
noise as a mental state. It is in this broader constellation that Goldstein’s concept
of the ‘catastrophic reaction’ and Canguilhem’s concept of normativity become
important, because they allow us to ask anew the question, in extremis, at the
edge of reason: how do we draw the line between information and noise?
Although the topic of Sands and Ratey’s article and indeed of Goldstein’s
concept of the ‘catastrophic reaction’ is clinical, focussing in particular on
mental health issues that deserve to be studied with due care for their singularity
and without rash generalizations, it is difficult to overlook the resonance with
current political developments. The return of isolationist politics on the global
stage, accompanied by a wave of xenophobia, and the concurrent politicization
of extremist, conservative interpretations of religion, cannot help but resonate
darkly with Goldstein’s concept of the ‘catastrophic reaction’.
While every care must be taken not to generalize where careful attention to
the specificity of a problem is required, it would also be wrong to completely
ignore that George Canguilhem’s thesis on the Normal and the Pathological, was
written while he was an active member of the resistance against fascism; that the
profound innovation Goldstein brought to the rehabilitation of the mentally ill,
came in response to a generation of young men returning shell-shocked from the
First World War; and that his Logic of the Organism was written in 1934, when
Goldstein was forced to give up his position as clinical director of psychiatry in
Königsberg and flee Germany, after being arrested and imprisoned for being a
Jew. Not only their implicit rejection of the conservative impulse, at the level of
both concept analysis and methodology, but also the extreme circumstances in
which Goldstein and Canguilhem wrote these important works, testify to the
ethical and political relevance of their thinking about normativity in the context
of pathology.
VII
Anxiety
In defining the ‘mental state of noise’ in terms of a ‘catastrophic reaction’ Sands

and Ratey place Kurt Goldstein at the core of their definition of the ‘mental state
of noise’. Here, we will go back briefly to Goldstein’s Logic of the Organism in order
to complement Sands and Ratey’s interpretation of the catastrophic reaction, in
particular by zeroing in on the opposition between noise and order (Goldstein
1948; Sands and Ratey 1986, 290). As Sands and Ratey note, the ‘catastrophic
reaction’ consists in an attempt to establish an atmosphere of ‘sameness’, in order
to compensate for the difficulty of controlling confusing internal affective and
cognitive states, frequently achieved via ‘excessive orderliness’, social withdrawal,
or recourse to ‘stereotypes’ (Sands and Ratey 1986, 291).
Goldstein indeed observed in the context of mental illness that a ‘methodical
character’ tied to the ‘fanatic need for order’ (Goldstein 1983, 39) serves the
purpose for the affected individual, ‘to maintain himself in a situation he is able to
master’ (Goldstein 1983, 37). He found the excessively methodical enforcement
of order to be symptomatic of a conservative attitude that is, in his now famous

expression, the phenomenon of a ‘declining life’:
The instinct for [self-]preservation can appear like an essential trait of the
organism, even though in reality, the tendency towards preservation is a
phenomenon of disease, of a life ‘in decline’. (Goldstein 1983, 355. Emphasis in
the original.)3
Goldstein thus saw in the catastrophic reaction a desperate attempt by the

individual to re-organize experience, by drawing attention around a radically
reduced focus of attention. He interpreted the ‘catastrophic’ reaction as a retreat
to a more constricted form of organization, whose sole purpose becomes the
preservation of self:
In order to escape catastrophic situations, the sick who suffer cerebral lesions
have a particularly characteristic means: it is their methodical character. These
individuals have a truly fanatic need of order. The wardrobes of cerebrally

traumatized I have observed during many years were exemplary models of a
certain type of determined order. […] If one places a messy heap of objects
before such an individual one soon observes that he, if he notices them, places
them immediately methodically one next to the other, sometimes even making
little heaps of those which, for him belong to a same category. (Goldstein 1983,
39. Emphasis in the original.)
The avoidance of stressful situations, by narrowing the focus of attention around

a single event or situation, however, invariably fails to have the desired effect.
The ‘catastrophic reaction’ in fact lowers the capability of adaptation, rather than
enabling the individual to adapt to the new situation. Even if a release of tension
is afforded by the ‘catastrophic reaction’, it ultimately augments the feeling of
impending catastrophe. Rather than reducing it, the ‘catastrophic reaction’ thus
amplifies the perception of a catastrophic situation.
Goldstein took his assessment of the rigidity of the conservative impulse a
significant step further, by comparing the spontaneous closure that characterizes
the ‘methodical character’, the isolation from the exterior world and obsessive
orderliness, with a state of death-like paralysis, which he compared with the
‘simulated death’ of certain animals, characterized by paralysis and fixation of
the source of danger:
This isolation from the exterior world […] must probably be considered as
analogous to what we call simulated death of animals. (Goldstein 1983, 39)
The animal’s rigid fixation on the source of threat of imminent death, in

the human situation, is a fixation on that which is at once frightening and

inconceivable. In order to fully characterize the ‘catastrophic reaction’ Goldstein
therefore insists on a subtle nuance: the ‘catastrophic reaction’ is reducible to
neither fear nor instinct, it is of the order of anxiety. The ‘catastrophic reaction’ –
and this will be essential to our more general epistemological conceptualization
of noise – does not arise from fear of something, but from anxiety: it is the fear
of something inconceivable.
What characterizes the ‘catastrophic reaction’ according to Goldstein is
thus not the fear of disorder, strictly speaking. He sees compulsive order as the
response to an anxiety aroused by an experience that has become inconceivable:
[What] causes him to cling obstinately to the order which is adequate to him
and which appears to us, normal beings, as an abnormally primitive order,
abnormally rigid and forced, [is not] the fear of objective disorder, which the ill
person naturally doesn’t feel objectively as such, but [of that] which he feels as
anxiety. (Goldstein 1983, 40–41)
When introducing the distinction between fear and anxiety, as between fear of
something and fear of the inconceivable, Goldstein puts the catastrophic reaction
into an explicitly philosophical context: ‘I am thinking of Pascal, Kierkegaard,
Heidegger’ (Goldstein 1983, 250). An existentialist orientation in Goldstein’s
thought could be further pursued here, by referring notably to Sartre’s La nausée.
But what we seek in Goldstein’s concept of the ‘catastrophic reaction’, in the
broader context of our argument about the epistemological aspect of noise, is
his attention to anxiety as the fear of the inconceivable. It is less an existentialist
état d’âme, than his definition of disorder that is of the highest relevance to our
considerations about noise:
What does disorder mean in this case? It goes without saying that objective
disorder does not exist any more than objective order. Disorder means an
arrangement such that it imposes no single, determined perspective, nor a
unique mode of utilization, but allows several or even many. Total disorder
(if it were possible) would, however, impose nothing, but freedom of choice.
(Goldstein 1983, 39)
The catastrophic reaction is thus characterized by the rigid negation of all

contingency. It is what we could call a catastrophically negentropic attitude. Its
stability is in fact rigidity, a state of control petrified in an inflexible structure.
The only certainty it avails is the avoidance of all ambiguity and complexity.
Disorder for Goldstein means, as entropy did for Shannon and Weaver, ‘freedom
of choice’.
Such ‘freedom of choice’ becomes inconceivable to the individual struck by
the devastating effect of contingency, traumatized or injured, because openness
requires of the individual the association of contingency with the freedom of
choosing amongst more than one possible perspective, of acting and reacting
in more than one possible way. This pluri-valence of an open situation results
precisely in what ‘the catastrophic reaction’ seeks to avoid, because of the
uncertainty it entails.
Goldstein’s insightful notion of anxiety as fear of the inconceivable, as
inconceivable multiplicity of choice, now illuminates for us Shannon’s conceptual
feat in defining information as ‘information entropy’, as that which occurs with
the lowest probability and whose prediction is thus characterized by the greatest
uncertainty, but also by the greatest ‘freedom of choice’ in probabilistic terms.
Goldstein’s concept of the ‘catastrophic reaction’ resonates so compellingly
beyond the realm of the individual’s psychopathology to the level of collective

attitudes, that it enables us to fully appreciate the weight of conservative
preconceptions that persist against Shannon’s audacious interpretation of
‘information entropy’ as ‘freedom of choice’. This definition of information was
so vehemently rejected, said to have no relevance beyond the sphere of electronic
signal transmission, because it implied disorder, uncertainty and thus stoked
anxiety about loss of control. Shannon and Weaver’s definition of information
as ‘information entropy’ appeared to sacrifice too much of what we ordinarily
associate with information (certainty, knowledge, facts and data) to an intimate
relation of information with contingency (Capurro and Hjorland 2003; Janich
2006).
The difference between information and noise in Shannon and Weaver’s
definition of ‘information entropy’ is not, as we have seen in the first part, a
difference between order and disorder, but the finest of lines separating the
contingency of both ‘information entropy’ and noise, a line drawn by the
intention alone with which a certain ‘entropy of information’ is selected and
transmitted as a message, against the backdrop of an accidental entropy that is
discarded as noise. By analogy we can infer the fundamental difference between
the intentional openness and ‘freedom of choice’, embraced by the poet and the
philosopher, and the accidental and in this sense excessive openness suffered as
the ‘mental state of noise’.
Novelty and uncertainty, which for Shannon go hand in hand with a high
content of information, require of the individual the capacity to assert uncertainty
as a choice rather than be swamped by it in the mental state of noise as confusion

and indecision. Noise is thus not an object of perception, but that which swamps
it. In this respect we could adopt Goldstein’s perspective on anxiety and say that
noise is, like disorder, an inconceivable freedom of choice.
VIII
Order
Entropy is often defined as molecular disorder; noise in molecular biology

implies the idea of variation from the norm, or of disorderly DNA transcription
or disorderly cell proliferation. Noise in communication technology evokes
the disorderly transmission of signals, and noise litigation in the social context
evokes most emphatically the idea of disorder: here noise signals most explicitly
behaviour that disrupts the social order and defies orders given by the authorities.
Intuitive as it may be, the notion of order, which is frequently opposed to noise
as disorder or chaos, is not all that obvious. It is therefore worth dwelling on the
interdisciplinary circulation of the concept of noise and its cortege of concepts,
order and disorder, even if these considerations appear to take us too far away
from the problem of the mental state of noise.
Although we were here dealing with the excessive orderliness of the
individual’s catastrophic reaction to noise, and therefore with a behaviour
defined as pathological, the relevance of the concepts of order and disorder
appears to extend beyond the psychiatric definition of the ‘catastrophic reaction’,

to epistemological and methodological considerations at the level of scientific
and philosophical discourse. Recall, for instance, the assessment the philosopher
of biology, Marjory Greene, gave of the rigidity of Logical Positivism, of its
incapacity to deal with the imprecision of the empirical world. This rigidity led, in
her view, to the ‘catatonic, vegetative state’ of Logical Positivism in contemporary
philosophy. The question is not how accurate a description Greene’s is of Logical
Positivism, which is not the object of study here. What is striking rather, is
that the passions run high, even in scientific and philosophical discourse,
when the requirement to engage with the uncertainty and imprecision of novel
perspectives clashes with an authoritative call to order that is experienced as
intellectually repressive.
The first thing to point out is that the concepts of order and disorder
are, scientifically speaking, no less ambiguous than the concept of noise. In
mathematical terms it makes no sense to equate even chaos with disorder, since
mathematical chaos, for instance as treated it in catastrophe theory by René Thom,
is deterministic – determined and thus orderly in its necessary unfolding, even
if its graphic representation is baffling and unpredictable. For this reason alone
it would be incorrect to equate entropy with disorder or with molecular chaos:
because the idea of disorder wrongly implies that physical reality favours one state
over another. It would be wrong also because entropy is a concept belonging to the
field of probability and is thus inherently not deterministic and thus cannot be said
to contravene a deterministic order – and how can order be anything other than
deterministic?
The difference between probability and determinism is, in fact, a line of
fracture that runs through the history of the sciences, between classical and non-
classical thermodynamics, classical mechanics and quantum mechanics. This
fracture is exemplified by the dispute between René Thom and Ilya Prigogine,
recounted here by Rainer Zimmermann in ‘Order and disorder – on the recent
dispute about determinism between Thom and Prigogine’:
Besides Prigogine, Thom accuses other authors such as Monod, Morin, Atlan
and Serres of attributing an inordinate and thus inappropriate significance to
notions such as contingency, noise, oscillations: ‘[…] all make contingency
responsible for either the organization of the world or the emergence of life
and thought […]’. Thom sees this as an ‘anti-scientific attitude par excellence’, a
‘mental state of confusion’, which one may forgive authors in the human sciences,
but never in the natural sciences. (Zimmermann 1988)
Boltzmann’s statistical formalization of entropy thus signals a radical departure

from classical physics and mechanical reversibility, which plunges physics into
what we may opportunistically call a mental state of noise, and which Heisenberg
established as an uncertainty relation in quantum physics.
Even in physics it is therefore difficult to give an intrinsic reason why one
should qualify one state of a system as more orderly than another – according to
which intrinsic physical criteria? How does the value of a physical state as order
or its disqualification as disorder insert itself into the indifference of physical
states, which simply are what they are and which unfold, by definition, in perfect
accordance with the laws of physics? Unless one asserts with Hume and Pearson
that there are no laws of nature, only habits of perception – but in this case
also the notion of disorder no longer has any relevance, because disorder simply
becomes that to which we are not habituated.
It is not that the power of prediction has lessened with the introduction
of the statistical method in physics, on the contrary, physical processes with
entropy have only been mastered on account of probability and statistics. What
has happened is that the idea of indeterminacy has gained prominence with
the introduction of probability and statistics: the impossibility of an observer’s
perfect knowledge of initial conditions of a deterministic chaotic system, calls for
the probabilistic method, which in turn introduces the sliver of in-determinacy
at the microscopic level of obersvation into an overall highly performing method
of prediction. Even if mass phenomena are statistically mastered and allow for an
impressive power of prediction, this sliver of indeterminacy in turn introduces
an irreducible unpredictability and with it the irreversibility of any probabilistic
process with increasing entropy. This sliver of irreducible indeterminacy in the
probabilistic process thus severs, irreversibly, probability from determinism.
Yet if nothing, at least in physical reality, can be said to be fully determined
by reason, on what grounds can an idea of order be erected against which we
can measure disorder? In biology the idea of molecular disorder, associated
with the notion of noise and entropy, preserves an implicit determinism that
lingers with genetic theory, despite the softening of the concept of teleology,
which Jacques Monod, in his Chance and Necessity, already replaced with the
term teleonomy (Monod 1973). The now common, if still controversial term,
teleonomy, implies the rule-bound unfolding of molecular mechanisms, such as
that of the structural modification of proteins in relation to their environment,
or the role of nucleic acids in regulating the formation of proteins (Moulin 2006,
1073). Rather than reducing the observable regularity of biological mechanisms

to a purposive logic, teleonomy reduces teleology to the acknowledgement of
apparent norms in biological mechanisms, without inferring a telos or purpose.
The neologism teleonomy was introduced to replace the metaphysical notion of
teleology, with the assumption that biological organisms act in a rule-bound way
and submit to norms, even though these may be the chance result of evolution.
As the philosopher of biology, Michel Morange, notes, the idea of an
implacable order in the form of a ‘genetic program’ was proposed independently
by Ernst Mayr, Francois Jacob and Jacques Monod, inaugurating the golden age
of genetic theory, where
[l]ife was considered to be the possession of genetic information and of a genetic
code allowing this information to be translated. (Morange 2005, 432)
This reduction of life to a predetermined order, even if the latter is the child of
evolution’s contingency, provoked the biologist Stanley Shostak to declare that
the reduction of life to the sum of the properties of the macromolecules present
in organisms, was nothing other than ‘the death of life’. Yet if, like Monod, we
accept the premise that the order imposed by the genetic ‘program’ is itself the
result of evolutionary contingency, or to say it with more emphasis, that the very
concept of order in biology is grounded in contingency, then noise, as maximum
information entropy or maximum in terms of evolution’s ‘freedom of choice’,
acquires the status an altogether fundamental role for the very concept of order in
biology, as that which both precedes and exceeds order, as that from which order
arises and into which it founders when pathology, eco-systemic transformations
or geological catastrophes reassert contingency over predictability.
There still lingers in the idea of noise in molecular biology the idea that
it is a source of disorder or deviance from the orderly and regular unfolding
of biological processes, which in turn points, in the concept of order, to an
implicit function of purpose. Even if teleology is ruled out as a metaphysical
concept, order and purpose appear to remain close cousins. They are related, in
conceptual terms, to the ideas of use-value and work, which we had previously
seen in relation to information (as opposed to the ‘spurious uncertainty’ of
noise), and of physical entropy, which according to Carnot’s classical definition,
is energy that is no longer available to perform work.
To oppose order to noise thus continues to imply that disorder, uncertainty
and error are the negation of a rule, of purpose, use-value or work. In this light
the entire enterprise of rational and scientific discourse, in so far as it relies
on the idea of order both in discourse and in the empirical world, must be
understood as the assertion of the power to impose order and purpose where
confusion, error and uncertainty loom large. This power, perhaps, is nothing
other than the more or less successful attempt to exercise control, temporarily,
in the midst of the fundamentally contingent unfolding of events that accounts
for time’s historical irreversibility and for the future’s irreducibility.
IX
Control
Noise, beyond the reference to unwanted sound, thus reveals itself to be

conceptually polymorphous because it has never been about types, classes or
measures of phenomena that qualify noise as a particular type of disturbance,
but about the relation between contingency and control. Contingency and
control, especially loss of control, means that in various domains of theoretical
investigation and practical application, very different types of phenomena are
at stake.
Sands and Ratey explicitly distance their use of the term ‘noise’ from Wiener’s
‘different’ understanding of noise in cybernetics. Without wanting to diminishing
the reasons for insisting on this difference we are nevertheless obliged to address
the link between their own approach to the ‘mental state of noise’ and the idea
of control in cybernetics, when Sands and Ratey emphasize their ‘theoretical’
formulation of the ‘catastrophic reaction’, by phrasing it as a ‘closing down of the
system’s circuitry’ (Sands and Ratey 1986, 296). Their decision to formulate noise
as a trans-nosological concept appears to expresses, indirectly, a generalized

diffusion of the cybernetic notion of ‘control’ to scientific discourse, even where
the cybernetic paradigm is not adopted explicitly.
It is not only the wording in this key moment of the article, but also the
objective of Sands and Ratey’s conceptualization of the ‘mental state of noise’
that links back to the cybernetic notion of control: for it is, ultimately, in view
of a pharmacological approach to feedback and amplification mechanisms
in the nervous system, that the notion of noise is developed here. Sands and
Ratey’s objective is ultimately to elucidate the potential benefit of certain drugs
in alleviating the ‘mental state of noise’, in combination with psychological
training of awareness of bodily sensations associated with it. This becomes
clear in the conclusion of their article, where they refer to the effects of cardiac
drugs on autonomic reactions. The attempt to conceptualize noise in the
context of psychiatry is thus finally put in the service of illuminating the role
of the involuntary nervous system in the ‘mental state of noise’, acting below the
level of consciousness and controlling visceral functions including heart rate,
respiratory rate, perspiration and arousal of the nervous system. It is at this pre-
conscious level that the authors propose to intervene with the use of cardiac
drugs such as beta-blockers and clonidine, in view of reducing what the authors
call the ‘reverberating circuit’ of emotional agitation and autonomic reactions.
The objective is thus to control the feedback mechanisms that occur between
the cognitive aspects of confusion, anxiety and agitation and the involuntary
physiological response to stress, which together characterize the ‘mental state of
noise’ and the ‘catastrophic reaction’.
This approach, Sands and Ratey argue, is relevant for the treatment of all
psychiatric illnesses, as wide ranging as brain-damage, mania, mental retardation,
schizophrenia and autism, but also in treating the effects of antipsychotic
neuroleptic drugs, such as restlessness (Sands and Ratey 1986, 293). Drugs
such as Beta-blockers and clonidine are discussed as acting on the ‘mental
state of noise’ by regulating feedback mechanisms between the central and the
peripheral nervous systems. So what are these drugs and what role do they play
in saving the ‘system’s circuitry’ from breaking down? Clonidine notably acts
on the ‘central down-regulation’ of the locus ceruleous, in other words, on the
part of the brainstem involved in the synthesis of the hormone noradrenaline.
Noradrenaline is a neurotransmitter involved not only in physiological
responses to stress and panic, but also in the state of vigilant concentration. By
lowering sympathetic outflow, this drug lowers the levels of stress hormones in
the autonomic nervous system, whose function is both to maintain homeostasis,

but also to mobilize the body in a ‘fight or flight’ response.
Beta-blockers, in turn, are drugs that act on the peripheral nervous system, by
blocking the reception of the stress hormone noradrenaline in the sympathetic
nervous system. Both drugs modulate the physiological response to stress. They
are found, according to Sand and Ratey, to ally anxiety, impulsivity and psychic
and bodily disorganization. The soothing effect of beta-blockers in relaxing striate
musculature (comprising both voluntary muscles and cardiac muscles), reduces
anxiety response in a way the authors compare to the ‘holding’ environment that
alleviates the ‘vicious whir’ of sensations in the infant, described by Greenacre
(Greenacre 1952).
These drugs, proposed to combat the ‘mental state of noise’, can be said
without too much exaggeration to act in an analogous way to the control of
noise in cybernetics: both approaches have in common that information and
noise are not dealt with at the level of signification or conscious processes, but
by controlling the amplification of undesired perturbations through feedback

mechanisms. This is indeed how Sands and Ratey argue for the use of beneficial
effects of these drugs, which ‘interrupt the reverberating circuit of emotional
agitation, cognitive disarray and peripheral arousal’. What underlies the
conceptualization of noise is thus ultimately a cybernetic logic of control, driving
the complex dynamic functions involved in the experience of the ‘mental state of
noise’ and its alleviation.
At stake is thus the positive amplification of noise in the catastrophic reaction,
and its control through negative feedback. In other words, what is at stake is
the barely acknowledged yet thereby even more crucial, because subterraneous
relation with cybernetics. The authors do acknowledge the cybernetic
conceptualization of noise in passing (‘The concept of noise, a term fist applied
by Wiener […]’), but what is more, they describe the ‘catastrophic reaction’ in
terms that leave no doubt about the technical paradigm (‘concreteness and closed
circuitry of the system’), even while they distance their own use of the term
noise, by specifying that the term is used by Wiener ‘in a somewhat different way
to the discussion of mental processes […]’.
The theory of cybernetics is, of course, itself from the start closely related
to the physiological concept of homeostasis, the self-regulation of an internal
milieu, which cybernetics analyses in analogy with man-made systems, i.e.
machines with self-regulation through feedback mechanisms. The ‘circuitry’
of man-made systems, of machines, thereby acquires a paradigmatic status.
Our understanding of machines henceforth encompasses our understanding
of natural and biological systems, because of the greater certainty and the
possibilities they offer for accurate mathematical analysis, in comparison to a
living organism, whose internal regulations and relations with the environment
are still too complex to provide an equally stable theoretical framework.
This is not to say that the convergence of physiology and cybernetics around
the term ‘noise’ expresses an explicit adherence to either cybernetics, second-
order cybernetics, or even Shannon’s conceptual framework for thinking about
information and noise. Indeed, neither the mathematics of Wiener’s cybernetics
nor those of Shannon’s information theoretical algorithms are, strictly speaking,
taken on board or even mentioned as an explicit theoretical framework when the
term ‘noise’ is used. It is rather the oblique reference to theories of contingency
or control that gives this and many other conceptions of noise their quotations
marks, allowing noise to become a transdisciplinary passe-partout.
Sands and Ratey’s ambivalent acknowledgement of the technical paradigm in
relation to their approach to the ‘mental state of noise’ is understandable, insofar
as the psychiatry and the neurosciences of the 1980s were perhaps still closer
to the humanities and psycho-dynamic approaches than today’s wholehearted
cognitivist mechanization of the human mind and its metaphorical reduction
of the brain to the model of a super computer. The theoretical influence of
cybernetics on Sands and Ratey’s approach to the ‘mental state of noise’ is not
intended here as an argument for the idea that noise is reducible to ‘crossed
wires’ in the brain. The computer, no matter how sophisticated and even
superior to the brain in its calculating power, memory and speed, does not
suffer the noise that interferes with its optimal functioning, it makes no value
judgements, and it does not fear the loss of control. The computer applies norms
of optimal operation, it does not invent these norms – optimal conditions are
‘optimal’ for the user and completely indifferent to the computer. In this sense
one could perhaps say that, ironically, the computer is superior not only because
of its superior combinatorial capacity, but simply because of its indifference to
man’s needs.
Even though the computer submits to rules and norms that may indeed be
designed to allow room for the ‘learning’ of new decisions, the computer does not
make judgements that ground these decisions. It can adjust and diversify, but it
cannot know the difference between adjustment and a just act according to self-
generated norms. And yet, something is lost, if this already pregnant cybernetic
link remains obscured, something that can illuminate differently not only the
idea of a ‘mental state of noise’, but also the concept of control. Sand and Ratey’s
highly original attempt to define the ‘mental state of noise’ allows us to think
about control as a fine-tuning and adjustment of complex dynamical variables,

rather than the simple imposition of a norm through pharmacological control.
Although this is hardly news to the engineer or mathematician, it introduces a
nuance that is lacking wherever cybernetics become the excuse for reductive
mechanization in general discourse, and wherever control assumes a dark
connotation, as in the context of political and commercial abuse of civil liberties
through new vulnerabilities that come with the ubiquity of new communication
technologies.
X
The Helmsman Metaphor: Kybernetes
A nuance to the reductive analogy between the computer and the brain can
be found, perhaps surprisingly, if we go back to cybernetics’ origins in control
theory. Here we find not only the metaphorical link between control and power
in the political sense of governing, but also its intrinsic limits as a mathematical
paradigm. Contemporary modelizations of complex systems owe much to
the field of control theory that emerged during the 1950s. Since then the idea
has taken hold that such modelizations offer a sufficient explanatory basis for
complex systems, such as the brain, when combined with statistical data form
empirical observations. Yet while there can be no doubt of the technical utility
of control theory for mechanical engineering, and indeed for our growing
understanding of complex systems, including the neurological basis for
cognition (‘Bluebrain | EPFL’ 2013), the question remains: what are the limits of
control theory as a conceptual framework for our understanding of the control
of powers, be they mechanical, psychosocial or downright political? Where does
the legitimacy of the mathematical model end, and where does the ideologically
motivated capturing and manipulation of processes of cognition, and of flows of
information and wealth begin? This requires us to ask what we mean, when we
speak of control in the context of cybernetics and what conceptual conversions
are required for the concept of control to be exported to non-mathematical
contexts.
Although archaeological evidence suggests that the invention of control
mechanisms dates back as far as irrigation systems in Mesopotamia, it is not
until the 1950s that control theory became a theoretical field in its own right.
It now extends from human or automatic control of mechanical devices, for
instance through measuring devices such as thermostats turning refrigerators
on or off, to remote controls or servomechanisms and motor governors. It deals
with individual systems or with the coordination of devices at a large scale, such
as a power plant or traffic control. More recent developments see the importance
of control theory in bio-technology, for instance in the development of artificial

organs or nerve-controlled prosthetics and artificial intelligence through pattern
recognition and speech recognition (‘Control Theory | Mathematics’ 2017a).
In his 1961 Cybernetics, Control and Communication in the Animal and the
Machine, Wiener refers to the origins of cybernetics in the formal mathematical
analysis of control mechanisms, by referring to James Clerk Maxwell’s 1868 On
Governors (Maxwell 1868):
We have decided to call the entire field of control and communication theory,
whether in the machine or in the animal, by the name of Cybernetics, which we
form from the Greek χυβερνήτης [kybernetes] or steersman. In choosing this
term we wish to recognize that the first significant paper on feedback mechanisms
is an article on governors, which was published by Clerk Maxwell in 1863,4 and
that governor is derived from a Latin corruption of χυβερνήτης. We also wish to
refer to the fact that the steering of engines of a ship is indeed one of the earliest
and best-developed forms of feed-back mechanisms. (Wiener 1961, 11–12)
A governor is a device that measures and regulates the speed of a machine or

engine. It is the basis for the invention of the servo mechanism, initially devised
for the steering of big ships. The governor makes the steering engine of a ship
independent of its load, by using readings from the steering wheel to regulate steam
pressure valves offset from the tiller (the lever used to turn the rudder of the ship);
the tiller’s movement to one side progressively closes some pressure valves, while
opening others, so as to increase the force that produces the rotation (torque). Any
force opposing the motion of the tiller keeps up the admission of steam, thereby
increasing the push of the tiller to the desired position. However, if feedback is too
brusque the rudder overshoots, leading to feedback in the other direction, which
in turn amplifies into a ‘wild oscillation or hunting’ in the steering mechanism.
Although Maxwell’s work predates by almost a century the emergence of
control theory as a field of research in its own right, it lays the foundations
for a fundamental problem tackled by control theory and cybernetics. And by
providing a mathematical theory of ‘self-oscillation’ it paves the way for the
understanding of phenomena of feedback, amplification and overcompensation
in increasingly complex dynamical processes.
Control thus means, first of all, the calibration of powers that vastly exceed
the steering capacity itself. This means that the governor not only informs a
greater power, but more importantly, that it calibrates the control action through
continual feedback, so as to avoid wild oscillation with chaotic consequences.
The idea of the governor thus represents more than a technical concept. It stands
for a paradigm of self-regulation that can be extended to natural and human
forms of organization, and can even be said to imply an ethics of self-regulation

for human forms of organization or governing.
However, despite its vast possibilities of application, it must not be forgotten
that control theory is first of all a consolidation of two classical branches of
mathematics, the calculus of variations and the theory of differential equations.
Differential equations help correlate a system’s measurable changes and rates
of change. By providing a function that expresses dependences amongst these
variables, it allows prediction of the behaviour of the system within given
constraints. The calculus of variations in turn helps understand phenomena like
elasticity, vibrations or electromagnetic theory. Finding the least surface area for
a given volume so that it encounters minimum resistance during its trajectory
is a typical aerodynamic problem solved through the calculus of variations
(‘Calculus of Variations | Mathematics’ 2013).
Control theory lends itself to any empirical domain, on condition that
its parameters are understood comprehensively enough to be subject to
exhaustive mathematical analysis. This means that whatever the system
under consideration, it must afford precise mathematical description both of
its internal dynamical state and of external influences affecting any control
intervention, in every possible circumstance. It is not only the precise
mathematical description of the system itself that is required, however, but also
the mathematical definition of purpose, in terms of control criteria. Together
with the environmental conditions and disturbances, these criteria constitute
the margin of ‘optimization’ for control. The future behaviour of the system can
then be deduced from knowledge of both the present state and future inputs
on the basis of the so-called ‘control law’. The latter is the rule that defines the
relation between variables of the state, in other words, it is the function that
determines the control action to be taken. This control action may indeed take
the form of feedback or feedforward methods, but the ‘control law’ is in itself
a more general concept about the functioning of a system in its environment,
than feedback (‘Control Theory | Mathematics’ 2017b).
Since not every component of the dynamical state of a system (or state
vector) can be measured simultaneously with instantaneous exactness, statistical
prediction and filtering theory step in to determine the control law with the
obtained estimate state vector. Cybernetics, and in its wake complex systems
theory, thus emerged from the historical coincidence of new methods of
statistical calculation and new means of computerized data compilation and
powers of calculation. Although it was by then possible to accurately represent
the state-vector equations of those physical systems in the natural world which
experience only small deviations from the steady state behaviour, the question
arose whether control theory could be applied also to non-linear models, in other
words, to systems in which small changes in input can result in large deviations.
Depending on the mathematical method, these systems are called stochastic or
chaotic. The crucial point, however, is that whatever knowledge control theory
affords us regarding empirical systems with greater or lesser complexity, its most
critical conceptual contribution is the precise mathematical definition of the
limits of controllability.
This inherently critical understanding of the conditions and limits of control
theory in the mathematical sense is forgotten at great cost. For this reason control
theory undoubtedly also lends itself to spontaneous ideological distortion.
The historical origins of statistics in the ‘discipline of the state’ indelibly links
statistics with the question of power: the power to collect information and the
power to predict and thereby influence the course of events. The word ‘control’
is itself so highly charged with political connotations that it cannot help but
blur the boundaries between mathematics and common language, when the
model of control theory is reintroduced into common discourse – distorting
the critical limit between abstract concepts and concrete interests, between the
need to know and the need to act. Without a clear understanding of its inherent
critical limit, control theory thus becomes prey to its ideological appropriation
for a control society without noise.
Nevertheless, if we are to think freely about the conceptual potential inherent
in this inventive convergence of diverse branches of mathematics in control
theory, without either succumbing to, or ignoring, the sinister link between
mechanical and political control, we may wish to recall George Cuvier’s ‘Report
on the Progress of the Sciences and Mathematics’, presented to the French
government in 1808. In this plea for parliament’s support of the sciences, Cuvier
addresses the government by drawing the cautionary tale of ‘ordinary men’
who see in science only their immediate advantage. Supporting the sciences
only when they can grasp their practical applications, they fail to understand
the principles of science, like the ‘vulgar’ who fails to appreciate a work of art.
And yet every proposition in science, Cuvier insists, is the germ of ‘a thousand
common inventions’ that, in turn, affect the very bases upon which not only the
state, but also the political relations between nations rest:
[F]eudal anarchy would perhaps still subsist, if canon powder hadn’t changed
the art of war; the two worlds would still be separate without the compass; and
no one can predict what today’s relations would become, if one could supply
colonial food-stocks with indigenous plants. (‘cpa9.17.cuvier.pdf ’ 1968)
What would the post-Second World War world look like, without the
development of computer science, and what would the Cold War have been,
without the contributions made by cybernetics to the development of self-
directing missiles? We may well ask ourselves what scientific principle could
more profoundly affect government, the relation between nations, and indeed
the relation between mankind and nature, than the mathematical harnessing
of uncertainty, the control of noise? But to fail to understand the sprawling,
practical fecundity of a single valid mathematical principle, Cuvier warns, is
also to fail to understand that a mathematical principle’s power lies within the
intrinsic understanding of its limitations:
In the mathematical sciences […] a single well stated and precisely measured fact
becomes a principle, a starting point; the rest is the labour of calculation: but the
limits of calculation are also those of the sciences. The theory of moral concerns
and of their powers stops here and more abruptly still before the continual and
incomprehensible mobility of the heart that ceaselessly defies every rule and
every prediction […]. (Cahiers Pour l’analyse N° 9/Genealogie Des Sciences 1968;
‘Cpa9.17.Cuvier.Pdf ’ 1968; Cuvier 1837)
We would do well, then, to heed Cuvier’s warning when overextending the powers
we attribute to control theory, by imagining a society controlled by data, if not
by artificial intelligence, in which the human becomes superfluous, replaced not
only as tool-bearer, but as critical intelligence. For the mathematical specification
of uncertainty in control theory also implies, by definition and by design, that the
limits of controllability are clear and that uncertainty, in the form of variation,
feedback and perturbation, is stratified: carefully distinguishing between

uncertainty that can be harnessed and predicted mathematically, uncertainty
that can be proven mathematically to be unpredictable, and uncertainty on
which we have no scientific handle. This knowledge of limits and stratification of
uncertainty provides a conceptual dexterity that is lost, when general discourse is
all too eager to soak up the ideas of control and prediction, hurried to promote
the idea that extremely complex and poorly defined systems, like economics
or the brain, are ‘controllable’ in the sense of causal explanations and control
mechanisms. We need to remember that control theory itself – and by extension
the forms of reasoning based on its mathematical paradigm, from cybernetics to
complex systems theory – imposes a critical check on the political appropriation
of scientific discourse for its own means, provided it can be articulated and heard
in public discourse.
XI
The Helmsman in Plato’s Alcibiades Dialogue
Nevertheless, it is also clear that the power of prediction, afforded by control

theory, is a hybrid of political power and scientific inventiveness – of the need to
know and the need to act. When Wiener adopts the governor as the emblem of
cybernetics, linking it to the image of the helmsman (Gr. χυβερνήτης, kybernetes),
he in fact rekindles a classical philosophical analogy between the art of navigation
and the political art of governing. The cybernetic metaphor of the ‘helmsman’
thereby cements a classical link between the art of navigation and the art of
governing in the contemporary technical paradigm, dating back to the Platonic
dialogues. Plato’s Alcibiades dialogue is crucial in this regard, not only because
of the importance given to the helmsman as metaphor for the art of governing,
but also because of the injunction it makes against ‘the most sickening’ aspect of
poor governance, which is the ignorance of one’s own ignorance. (How could we
fail to mention here that the etymology of noise leads back to the nautical field,
i.e. nausea, or sea-sickness, completing the metaphorical analogy between noise
and uncertainty.)
In this dialogue Socrates is the confidante and adviser to the inexperienced
Alcibiades, who expects his imminent ascension to political power on the
grounds of his noble heritage.5 Socrates’ advice, however, concerns not how
best to govern and control the populace. Instead, he insists that governance
requires self-control of a particular kind: only the knowledge of one’s ignorance
of unintended consequences can act as guarantor for good governance (Plato
1997, 557–95). Through the dialectical method of question and answer, Socrates
leads Alcibiades to demonstrate his own ignorance to himself. At the risk of
displeasing the future ruler, Socrates indeed forces Alcibiades to acknowledge
that his is the worst of all forms of ignorance: ignorance of his own ignorance
and, without holding back, condemns this as the most sickening of all forms
of ignorance. Carefully calibrating the risk of too brusque a ‘feedback’ in the
dialogue with Alcibiades’, Socrates thus treads a tightrope between the need
to achieve an effect powerful enough to change the direction of Alcibiades’

thinking, and the risk of a catastrophic unravelling of the situation for Socrates.
In this powerfully phrased condemnation of privilege as an excuse for ignorance
in political governance, Socrates speaks truth to power, in the sense that Foucault
would celebrate in his later lectures at the College de France. Here Foucault defines
the role of the teacher or master, by distinguishing the modalities of speaking the
truth (parrhesia, speaking the truth freely and without reserve) from rhetoric, as
the modality of speaking in order to convince without regard for truth. In words
that could not be more poignant in our own times of ‘post-truth’ elections and
bogus referendums, which make even post-modern cynicism look naive, Foucault’s
raises the question how an individual becomes a subject, by asking
in which manner the individual constitutes itself in the act of speaking true,
and how it is constituted by others as a subject holding a true discourse […].
(Foucault 2011)6
Rather than the creation or revelation of truth, (alethurgy), the praxis of

parrhesia is thus a modality of speaking freely, an alethurgical form. It distinguishes
itself form rhetoric, but is even more different from flattery, because to speak
freely, to hold nothing of the truth back, implies the risk provoking a violent
reaction. For Foucault, Socrates is the parrhesiast who pursues this task ‘even
when he is threatened with death’ and to his last breath (Foucault 2011, 36).
It is thus in this Foucauldian sense, as an act of parrhesia, as an act of courage,
that Socrates uses the metaphor of the helmsman to hold up to the young and
powerful Alcibiades the mirror of ‘the most disgraceful sort of stupidity’, which at
once not only proves him unfit to govern but also gives Alcibiades the opportunity
to acquire the awareness necessary for his incumbent responsibilities:
Socrates: Well, if you were sailing a ship, would you be out there wondering
whether to put the helm to port or starboard, and wavering because you didn’t
know? […]
Alcibiades: I’d leave it to the skipper.
Socrates: So you don’t waver about what you don’t know if in fact you know
that you don’t know. […] Don’t you realize that the errors in our conduct are
caused by this kind of ignorance, of thinking that we know when we don’t know?
[…] the sorts of people who don’t think they know how to do things make no
mistakes in life, because they leave those things to other people. […] Well, who
are the ones making the mistakes? Surely not the ones who know? […] Since
it’s not those who know, and it’s not those who don’t know and know they don’t
know, is there anyone left except those who don’t know but think they do know?
[…] So this is the ignorance that causes bad things; this is the most disgraceful
sort of stupidity […]. (Plato 1997, 574–75)
Alcibiades is unfit to govern, not because he is young and lacks knowledge, but
because he cannot conceive that there may be unintended consequences of his
actions. His arrogance is the fruit not of lack of knowledge, but of his ignorance
of this lack of knowledge. The risk of wavering, of putting the helm to port or
starboard too abruptly, of underestimating the powers he modulates through the
use of the tiller thus stands for Alcibiades’s lack of knowledge of the unintended
consequences of his actions. Alcibiades lacks knowledge of the limits of his own
competence: not only will he fail to predict the risk of unintended consequences
of his action, his arrogance relies on the fact that he cannot even fathom this risk,
in contradistinction with Socrates, who uses the dialectical method as ‘feedback’
in full knowledge of the risk he incurs that Alcibiades may rise to power and take
revenge for the humiliation.
Alcibiades lacks self-control, because the self-control Socrates speaks of
comes from both the knowledge of one’s own power and the knowledge of
one’s own ignorance. What Socrates leads him to understand is that Alcibiades
wants to be master, when he is only the product of power: his opinion sways one
way or another according to persuasion. Unlike the well-adjusted action of the
helmsman, Alcibiades’ opinions are driven passively, subject to the powers that
prevail, rather than being the expression of a subject, asserting itself in the face of
contingency. In other words, the information Alcibiades needs in order to govern
is both knowledge of the art of governance, of what is the good and just conduct,
and knowledge of contingency, in the form of knowledge of his own ignorance.
Wiener’s return to the metaphorical relation between the governor and
the helmsman allows us to compare Alcibiades’ wavering and confusion
as an inexperienced political governor, to the wild oscillation or hunting of
the ill-adjusted tiller. Also Sand and Ratey’s definition of the ‘catastrophic
reaction’, whereby any attempt to overcome the ‘mental state of noise’ ‘only
adds to confusion’, can be understood, like the overshooting of the rudder,
as an overcompensation, causing amplification of turbulence and ultimately
breakdown. By analogy, noise becomes a political, ethical and epistemological
problem in equal measure: lack of knowledge of one’s ignorance makes noise an
epistemological problem of the docte ignorance; power without the knowledge of
unintended consequences makes noise a political problem of good governance;
and, the relation between contingency and self-governance, finally, makes noise
an ethical problem of the just conduct.
Taken together, the mathematical requirements of a critical definition of the

limits of controllability, and the cautionary role of the helmsman metaphor in
the Platonic dialogue, allow us to bring a nuance to either the purely utilitarian
definition of control or the entirely negative association of control with the
abuse of power. The notion of control is insufficiently understood when it is
seen exclusively as leading to a totalitarian exercise of power – even though
there is no doubt that the knowledge of control theory is put to efficient use in
mechanizing human decisions and rationalizing human action with the objective
of efficiency and frictionless exercise of power. Yet, what is lost, if the cybernetic
concept of control is reduced to the idea of totalitarian domination without
noise, is the nexus between knowledge and ignorance, in the form of systematic
epistemic humility: the knowledge of noise, of non-linear dynamics and of the
mathematical limits of control. The important status of noise in cybernetics thus
allows us to add a nuance to the cybernetic paradigm, often overlooked when the
focus is – rightly – on the confluence of technological, political and commercial
control in an information intensive and globally networked economy (Mersch
2013). The point is not to soften the critique of cybernetic ‘control’ society, but
to think freely, in order to rethink control as self-control in the form of epistemic
humility before unintended consequences of actions. In turn, cybernetics itself
is an incomplete paradigm, if it is concerned only with adjusting the trajectory
of a self-directing missile and fails to consider the problem of the just act, as a
problem involving contingency, ignorance and unintended consequences.
In light of the cybernetic notion of control and interpreted in view of
the Alcibiades dialogue as self-control through knowledge of one’s own

ignorance, the problem of noise thus acquires an ethico-political dimension
that is inseparable from the epistemological problem it poses. It requires us to
distinguish the manifold objects of the definition of noise (acoustic, cognitive,
biological, thermodynamic noise, etc.) from the normative power to regulate
complex systems endowed with recursivity or feedback.
The individual trapped in the ‘mental state of noise’ is, like the protagonist
of a drama, bereft of certainties: incapacitated to judge the unpredictable
nature of stimuli as either useful (information) or potentially harmful (noise)
noise becomes equivalent with an undecidable state. Noise here becomes an
unthinkable freedom of choice. The individual is nevertheless required to
assert a decision, and to assert it on the grounds of nothing but this necessity
to either appropriate this overwhelming experience as his or her own or face
break down.
The catastrophic reaction to this state of indecision is the compulsive

imposition of order, closure and the precipitation into meaning. For the
‘chameleon poet’, like Keats, but also for the resistance philosopher, like
Canguilhem, and by default for every scientist working at the edge of what is
known and can be proven, the willful confrontation with uncertainty becomes
the occasion for a self-grounding of art, science or reason. What the individual
has lost in the ‘mental state of noise’ are the physiological a priori of the healthy
nervous system, which filter incoming stimuli according to pre-conscious
criteria of pertinence, but also according to the adaptability of a posteriori rules
which experience had consolidated over time as pre-set attitudes. Such pre-set
attitudes feed into our cultural preconceptions and even into our epistemic a
priori, notably the sense of duration and the sense of self. However, when these
are powerless in the face of contingency, life calls for new norms of living.
Reason too, calls for new a priori when its paradigms are called into question
by radical uncertainty. This was the case with Lobachevsky’s answer to the
impossibility of proving Euclid’s fifth postulate, which occasioned the invention
of non-classical geometry with infinite dimensions. This recasting of geometry
continued to throw waves in the epistemic field, precipitating the crisis of
foundation of mathematics and the pluralization of logic.
In this sense, noise, understood as maximum uncertainty, is what calls forth
and hence precedes the normativity of reason, i.e. the judgement according to
which uncertainty is valued as informative or discarded as spurious.
This is why we can now interpret the problem of noise, its illegitimate and
therefore abnormal character, in light of Canguilhem’s enigmatic dictum about

the priority of the abnormal:
Consequently it is not paradoxical to say that the abnormal, while logically
second, is existentially first. (Canguilhem, 1991, p. 243)
We can now think of noise in terms of a fundamental epistemological

contingency, a state of suspension or indecision, from which reason emancipates
itself with acts of self-grounding. As the groundlessness that necessarily precedes
our own rational self-grounding, noise is is no longer marginal to philosophical
discourse, no longer reducible to mere error. Noise can, instead, be thought as
as a fundamental philosophical problem: as the groundlessness that necessarily
precedes reason’s act of self-grounding.
What is at stake with the question of noise, is ultimately a vital and
epistemological normativity, an emancipatory act of self-grounding, that is
conditioned by no ready-made control law, grounded in nothing but itself.
A cybernetics worthy of its noble philosophical heritage, a cybernetics of the

just act, would thus not only be a cybernetics capable of speaking truth to power.
The cybernetics of a just act, capable of harnessing powers far greater than itself,
is one that cultivates the knowledge of its own ignorance, rather than boasting
with the mechanical reduction of complexity; one that cultivates control as
self-control and fine-tuning of doubt, rather than as an excuse for the hubris of
domination.
The transdisciplinary proliferation of concepts of noise, far from merely
indicating error and perturbation everywhere, may be the index of a renewed
and more confident scientific, artistic and philosophical culture of qualified
uncertainty.
Notes
1 Thanks to Inigo Wilkins for drawing my attention to this.

2 On the ongoing cultural and economic shift away from a tradition of deep focus
towards the requirement of polyfocal ‘hyper attention’, see Katherine Hayles, ‘Hyper
and Deep Attention: The Generational Divide in Cognitive Modes’, Profession,
pp. 187–199.
3 L’instinct de conservation peut apparaître comme un trait essentiel de l’organisme,
bien qu’en réalité, la tendance à la conservation soit un phénomène de maladie, de vie
‘qui décline’ (Goldstein 1983, 355).
4 (Maxwell 1868).
5 Although the Alcibiades dialogue was traditionally attributed to Plato, D. S.

Hutchison argues in his introduction to the dialogue that similarity with later
Academic doctrine and the simplicity of the dialogue point to an Academic
philosopher writing in the 350s or just afterwards and that it displays similarity
notably with the Aristotelian idea of the Magna Moralia that self-knowledge is best
gained through a philosophical friendship in which we see ourselves, as in a mirror
(Plato 1997, 558).
6 My translation (Foucault 2009; Foucault and Davidson 2012).
Bibliography
Abramowicz, Lisa. 2017. ‘A Synthetic CDO by Any Other Name Is Still Risky’.
Bloomberg Gadfly. Accessed 18 July 2017. https://www.bloomberg.com/gadfly/
articles/2017-02-03/a-synthetic-cdo-by-any-other-name-is-still-risky
Allen, Robert, ed. 2000. The New Penguin English Dictionary. 1 edition. London and
New York: Penguin Books Ltd.
Allen, Robert. 2001. ‘Information’. In The New Penguin English Dictionary. London:
Penguin Group.
Alliez, Eric. 2010. ‘What Is – or What Is Not – Contemporary French Philosophy,
Today?’ Radical Philosophy. Accessed 31 May 2017. https://www.radicalphilosophy.
com/article/what-is-%e2%80%93-or-what-is-not-%e2%80%93-contemporary-
french-philosophy-today
Alloway, Tracy, and Michael Mackenzie. 2013. ‘Boom-Era Credit Deals Poised
for Comeback’. Financial Times. Accessed 18 July 2017. https://www.ft.com/
content/1d2d33d8-5ba5-11e3-848e-00144feabdc0
Altmann, Jürgen. 2001. ‘Acoustic Weapons – A Prospective Assessment’. Science &
Gobal Security 9: 165–234.
Anderson, M. 1988. The American Census. A Social History. New Haven: Yale University
Press.
Ashby, Ross. 1962. ‘Principles of the Self-Organizing System’. In Principles of Self-
Organization: Transactions of the University of Illinois Symposium, edited by H.
Von Foerster and G. W. Zopf, 255–78. London: Pergamon Press. http://csis.pace

.edu/~marchese/CS396x/Computing/Ashby.pdf
Atlan, H, and A. Katzir-Katchalsky. 1973. ‘Tellegen’s Theorem for Bond Graphs. Its
Relevance to Chemical Networks’. Currents in Modern Biology 5 (2): 55–65.
Atlan, Henri. 1979. Entre Le Cristal et La Fumée. Paris: Seuil.
Attali, Jacques. 1985. Noise: Political Economy of Music. Minneapolis: University Of
Minnesota Press.
Ayache, Elie. 2015. The Medium of Contingency: An Inverse View of the Market by Elie
Ayache. London: Palgrave Macmillan.
Bachelard, Gaston. 2003. Le Nouvel Esprit Scientifique. Paris: Presses Universitaires de
France – PUF.
Bachelier, L., 1900. Théorie de la spéculation. Annales Scientifiques E.N.S.
17, 21–86. Accessed 29 January 2018. http://archive.numdam.org/article/
ASENS_1900_3_17__21_0.pdf
220 Bibliography
Badiou, Alain. 1993. ‘Y-a-Til Une Théorie Du Sujet Chez Georges Canguilhem?’ In
Georges Canguilhem. Philosophe, Historien Des Sciences, 295–304. Paris: Albin Michel.
Badiou, Alain. 2005. Being and Event. London: Continuum.
Balibar, Françoise, Roland Lehoucq, and Jean-Marc Lévy-Leblond. 2005. Qu’est-Ce Que
La Matière ? Paris: Editions le Pommier.
Barber, Kenneth F. and Jorge J.E Gracia. ed. 1994. Individuation and Identity in Early
Modern Philosophy, Descartes to Kant. Albany: State University of New York Press.
Bawden, David, and Lyn Robinson. 2009. ‘The Dark Side of Information: Overload,
Anxiety and Other Paradoxes and Pathologies’. Journal of Information Science 35 (2):
180–91. doi:10.1177/0165551508095781.
BBC. 2008. ‘Company Defends Anti-Teen Device’. BBC: Company Defends Anti-Teen
Device. Accessed 12 February. http://news.bbc.co.uk/1/hi/wales/7240306.stm
Bell, E. 1986. Men of Mathematics. 1st Touchstone edition. New York: Simon & Schuster.
Benzecri, J. P. 1982. Histoire et Préhistoire de l’analyse Des Données. Paris: Dunod.
Berglund, Brigitta, and Thomas Lindvall, eds. 1995. ‘Guidelines for Community Noise.
Document Prepared for WHO’. Archives of the Centre for Sesory Research. Stockholm
University and Karolinska Institute 2 (1): iii–140.
Bijsterveld, Karin. 2001. ‘The Diabolical Symphony of the Mechanical Age: Technology
and Symbolism of Sound in European and North American Noise Abatement
Campaigns 1900–1940’. Social Studies of Science 31 (1): 37–70.
Bijsterveld, Karin. 2003. ‘City of Din: Decibels, Noise and Neighbors in the Netherlands,
1910–1980’. OSIRIS 18: 173–93.
Blanché, Robert. 2009. L’axiomatique. Paris: Presses Universitaires de France – PUF.
‘Bluebrain | EPFL’. 2013. Accessed 11 March. http://bluebrain.epfl.ch/
‘BMJ Careers – Doctors’ Own Mental Health Issues’. 2017. Accessed 31 May. http://
careers.bmj.com/careers/advice/Doctors%E2%80%99_own_mental_health_issues
Boutin, Aimee. 2015. City of Noise: Sound and Nineteenth-Century Paris. Urbana:
University of Illinois Press.
Brassier, Ray. 2007. ‘Multitudes Web – 22. Genre Is Obsolete’. Genre Is Obsolete.
Accessed 7 March 2013. http://multitudes.samizdat.net/Genre-is-Obsolete
Brillouin, Leon. 1949. ‘Life, Thermodynamics and Cybernetics’. American Scientist 37:
554–68.
Brillouin, Leon. 2013. Science and Information Theory: Second Edition. 2nd revised
edition. Mineola: Dover Publications Inc.
Bunz, Mercedes. 2012. Die stille Revolution. Berlin: Suhrkamp Verlag Gmbh.
Cage, John, Dieter Daniels, and Hartware-Medien-Kunst-Verein (Dortmund). 2012.
Sounds like Silence: John Cage 4'33' Silence Today: 1912–1952–2012. Leipzig: Spector
Books.
‘Cahiers Pour l’Analyse (An Electronic Edition)’. 2017. Accessed 29 May. http://cahiers.
kingston.ac.uk/
Cahiers Pour l’analyse N° 9/Genealogie Des Sciences. 1968. Seuil/travaux du cercle
d’epistemologie.
Bibliography 221
‘Calculus of Variations | Mathematics’, 2013. Encyclopedia Britannica. Accessed 9 June

2017. https://www.britannica.com/topic/calculus-of-variations-mathematics
Campbell, Sarah. 2008. ‘Now Crime Gadget Can Annoy Us All’. BBC. Accessed 1
January 2015 sec. UK. http://news.bbc.co.uk/1/hi/uk/7759818.stm
Canguilhem, Georges. 1991. The Normal and the Pathological. Translated by Carolyn R.
Fawcett. New edition. New York: Zone Books.
Canguilhem, Georges. 1993. ‘Le Cerveau et La Pensée’. In Georges Canguilhem.
Philosophe, Historien Des Sciences. 1st edn., 11–36. Paris: Albin Michel.
Canguilhem, Georges. 2000. A Vital Rationalist: Selected Writings from Georges
Canguilhem. New edn. New York: Zone Books.
Canguilhem, Georges. 2009. Le Normal et Le Pathologique. Presses Universitaires de
France - PUF.
Capurro, Rafael, and Birger Hjorland. 2003. ‘The Concept of Information’. Annual
Review of Information Science and Technology 37: 343–411.
Castellengo, Michele. 1994. ‘Les formes sonores’. In Les sciences de la forme aujourd’hui,
edited by Yves Bouligand, 127–39. Paris: Ed. du Seuil.
Cavaillès, Jean. 2000. Sur La Logique et La Théorie de La Science. Paris: Vrin.
Coll. INSEE. 1987. Pour Une Histoire de La Statistique. Edited by J. Affichard. Vol. 2.
Paris: Economica.
Collectif. 1979. Colloque de Royaumont: Theories Du Langage, Theories de
l’apprantissage. Paris: Seuil.
‘Control Theory | Mathematics’. 2017a. Encyclopedia Britannica. Accessed 8 June 2017.
https://www.britannica.com/topic/control-theory-mathematics
Cowan, J. D. 1965. ‘The Problem of Organismic Reliability’. Progress in Brain Research
17: 9–63.
‘Cpa9.17.Cuvier.Pdf ’. 1968. Accessed 10 June 2017. http://cahiers.kingston.ac.uk/pdf/
cpa9.17.cuvier.pdf
Crutzen, P.J., 2002. Geology of mankind [WWW Document]. Nature. https://doi
.org/10.1038/415023a
Cummings, Dominic. 2017. ‘Dominic Cummings: How the Brexit Referendum
Was Won’, 2017. The Spectator. Accessed 1 May 2017. https://blogs.spectator.
co.uk/2017/01/dominic-cummings-brexit-referendum-won/
Cusick, Suzanne. 2006. ‘Music as Torture’. TRANS - Revista Transcultural de Música -
Transcultural Music Review, no. 10. http://www.sibetrans.com/trans/articulo/152
/music-as-torture-music-as-weapon
Cuvier, M. Le Baron G. 1837. Histoire Des Progrès Des Sciences Naturelles : Depuis 1789
Jusqu’à ce Jour. Bruxelles: Société Belge de Librairie. Accessed 14 February 2018.
https://archive.org/details/histoiredesprog02cuvi.
Desrosieres, Alain. 2006. ‘Statistique’. In Dictionnaire d’histoire et philosohie des sciences,
4th edition, 1008–24. Paris: puf.
Diaz Nafria, Jose Maria. 2010. ‘What Is Information? A Multidimensional Concern’.
Triple C 8 (1): 77–108.
222 Bibliography
Dodig Crnkovic, Gordana, and Wolfgang Hofkirchner. 2011. ‘Floridi’s “Open Problems
in Philosophy of Information” Ten Years Later’. Information Open access (2): 327–59.
doi:10.3390/info2020327.
Eigen, M. 1971. ‘Selforganization of Matter and the Evolution of Biological
Macromolecules’. Die Naturwissenschaften 58 (10): 465–523.
Eisler, Zoltan, Janos Kertesz, and Fabrizio Lillo. 2007. ‘The Limit Order Book on
Different Time Scales’. Proc. SPIE 6601 Conference Volume 6601 (Noise and
Stochastics in Complex Systems and Finance): 66010G.
Elliott, Larry. 2017. ‘Bank of England Warns of Complacency over Big Rise in Personal
Debt’. The Guardian. Accessed 29 July 2017, sec. Business. http://www.theguardian
.com/business/2017/jul/24/bank-of-england-household-debt-bank-credit-card-car-
loans
Enzensberger, Hans Magnus. 1977. Der kurze Sommer der Anarchie: Buenaventura
Durrutis Leben und Tod. Roman. 19th edition. Frankfurt am Main: Suhrkamp Verlag.
Esposito, Elena. 2011. The Future of Futures: The Time of Money in Financing and
Society. Cheltenham and Northampton: Edward Elgar Publishing Ltd.
Floridi, Luciano. 2002. Philosophy and Computing: An Introduction. 1st edition. London:
Routledge.
Floridi, Luciano. 2010. Information. Oxford and New York: Oxford University Press.
Floridi, Luciano. 2014. The Fourth Revolution: How the Infosphere Is Reshaping Human
Reality. 1st edition. Oxford: Oxford University Press.
Foucault, Michel. 1989. ‘Introduction’. In The Normal and the Pathological. Translated by
Carolyn R. Fawcett, 7–24. New York: Zone Books.
Foucault, Michel. 2009. Le gouvernement de soi et des autres : Tome 2, Le courage de la
vérité – Cours au Collège de France. Paris: Seuil.
Foucault, Michel. 2011. Der Mut zur Wahrheit: Die Regierung des Selbst und der anderen
II. Vorlesungen am Collège de France 1983/84. Translated by Jürgen Schröder. Berlin:

Suhrkamp Verlag Ag.
Foucault, Michel. 2012. The Courage of Truth: The Government of Self and Others II:
1983–1984. Translated by Graham Burchell. Edited by Arnold I. Davidson. Reprint
edition. New York: Palgrave Macmillan; Picador USA.
Frosch, John. 1983. The Psychotic Process. New York : International Universities Press.
http://archive.org/details/psychoticprocess00frosrich
Gelman, Andrew, and Cosma Rohilla Shalizi. 2013. ‘Philosophy and the Practice of
Bayesian Statistics’. British Journal of Mathematical and Statistical Psychology 66:
8–38.
Gibson, Kirsten, and Ian Biddle, eds. 2016. Cultural Histories of Noise, Sound and
Listening in Europe, 1300–1918. Burlington: Routledge.
Glass, David C., and Jerome E. Singer. 1972. Urban Stress: Experiments on Noise and
Social Stressors. New York: Academic Press.
‘Global Economic Prospects | Data’. 2017. Accessed 19 July 2017. http://data.worldbank.
org/data-catalog/global-economic-prospects
Bibliography 223
Goldstein, Kurt. 1948. Aftereffects of Brain Injuries in War: Their Evaluation and Treatment;
the Application of Psychologic Methods in the Clinic. New York: Grune & Stratton.
Goldstein, Kurt. 1983. Structure de l’organisme. Paris: Gallimard.
Goodman, Steve. 2012. Sonic Warfare: Sound, Affect, and the Ecology of Fear.
Cambridge, MA: The MIT Press.
Greenacre, Phyllis. 1952. Trauma, Growth, and Personality. New York: Norton.
Grene, M. 2007. ‘La vie des sciences et les sciences de la vie’. In Conceptions de La
Science : Hier, Aujourd’hui, Demain. Hommage à Marjorie Grene, edited by J. Gayon
and R. M. Burian, 23–36. Paris: Ousia.
Grossman, Sanford J. and Joseph Stiglitz. 1980. ‘On the Impossibility of Informationally
Efficient Markets’. American Economic Review 70: 393–408.
Hacking, Ian. 1990. The Taming of Chance. Cambridge: Cambridge University Press.
Hainge, Greg. 2012. Noise Matters: Towards an Ontology of Noise. London: Continuum
Publishing Corporation.
Hallward, Peter and Knox Peden ed. 2012a. Concept and Form, Volume 1. London and
Brooklyn: Verso.
Hallward, Peter and Knox Peden ed. 2012b. Concept and Form, Volume 2. London and
Brooklyn: Verso.
Harris, Ben. 1979. ‘Whatever Happened to Little Albert?’ American Psychologist 34 (2):
151–60. doi:10.1037/0003-066X.34.2.151.
Harrison, Charles, and Paul Wood, eds. 1999. Art in Theory, 1900–1990. An Anthology
of Changing Ideas. 1st edition. Oxford and Cambridge, MA: Blackwell.
Hegarty, Paul. 2007. Noise/Music: A History. New York: Continuum.
Herbert, Martin 2016. Tell Them I Said No. Berlin: Sternberg Press.
Hirsch, G. 1976. ‘Langage et Pensée Mathématiques’. In Langage et Pensée
Mathématique, 42–62. Luxembourg: Actes, Colloque International. Centre
Universitaire de Luxembourg.
Hydaralli, Saeed. 2012. ‘Noise: An Inquiry into Its Formal Properties’. In Reverberations,
edited by Michael Goddard, Benjamin Halligan, Paul Hegarty, 221–32. London:
Continuum.
‘Information’. 2014a. Oxford Dictionaries. Oxford University Press. http://www
.oxforddictionaries.com
‘Information’. 2014b. Merriam-Webster. London: Merriam-Webster, Encyclopedia
Brittanica Company. http://www.merriam-webster.com/
Jaen-Marie Pierrel (first) dir., William eds. del-Mancino, Laurent Gobert, Jean-Marc
Humbert, and Etienne Petitjean. n.d. ‘Nécessité’. Centre National de Ressources
Textuelles et Lexicales (CNRTL). http://www.cnrtl.fr/definition/nécessité
Janich, Peter. 2006. Was ist Information?: Kritik einer Legende. Frankfurt am Main:
Suhrkamp.
Judah, Hettie. 2017. ‘Don’t Look Now: The Artists Who Turn Their Backs on the World’.
The Guardian. Accessed 25 June 2017, sec. Art and design. https://www.theguardian.
com/artanddesign/2017/apr/20/tell-them-i-said-no-artists-who-turn-backs-on-
world-martin-herbert
224 Bibliography
Keats, John. 1818. ‘Letter to Richard Woodhouse’. Accessed 10 August 2014. http://www
.john-keats.com/briefe/271018.htm
Keats, John. 1958. The Letters of John Keats, 1814–1821. Cambridge, MA: Harvard University
Press.
Kindleberger, Charles P., and Robert Z. Aliber. 2011. Manias, Panics and Crashes: A
History of Financial Crises. 6th edition. New York: Palgrave Macmillan.
Kyle, Albert S. 1985. ‘Continuous Auctions and Insider Trading’. Econometrica 53:
1315–36.
LaBelle, Brandon. 2006. Background Noise: Perspectives on Sound Art. 1st edition.
London: Continuum.
Larousse, Éditions. 2017. ‘Définitions: Entropie – Dictionnaire de Français Larousse’.
Accessed 10 May 2017. http://www.larousse.fr/dictionnaires/francais/entropie/30093
Lemaine, Jean-Marie. 1960. ‘Schachter Stanley, The Psychology of Affiliation’. Revue
française de sociologie 1 (4): 499–500.
Lévy-Leblond, Jean-Marc. 1976. ‘Us et Abus de Langage: Mathématique, Didactique,
Physique …’ In Langage et Pensée Mathématique. Luxembourg: Actes, Colloque
International. Centre Universitaire de Luxembourg.
Locke, John. 1975. An Essay Concerning Human Understanding, ed. P. H. Nidditch.
Oxford: Clarendon Press. III, 27.3, 409.
Lockwood, Dean. 2012. ‘Lovecraft’s Weird Ecology of Noise’. In Reverberations: The
Philosophy, Aesthetics and Politics of Noise, edited by Paul Hegarty, Michael Goddard,
and Benjamin Halligan, 73–83. London: Continuum Publishing Corporation.
‘MA Rotterdam, Proceedings City Council Rotterdam, Session 26 February 1914,
64–65’. in.
Malaspina, Cecile. 2012a. ‘The Noise Paradigm’. In Reverberations: The Philosophy,
Aesthetics and Politics of Noise, edited by Michael Goddard, Benjamin Halligan and
Paul Hegarty, 58–72. London: Continuum.

Malaspina, Cecile. 2014. ‘Epistemic noise’. Systema: Connecting Matter, Life, Culture
and Technology 2 (1): 48–58.
Manson, Phillip. 2014. The Philosophical Implications of Information Overload. Seattle:
CreateSpace Independent Publishing Platform.
Marcel Ducham p. 1917. ‘The Richard Mutt Case’, The Blind Man. New York.
Maxwell, J. C. 1868. ‘On Governors’. Proceedings of the Royal Society 100: 270–83.
Mersch, Dieter. 2013. Ordo Ab Chao - Order from Noise. Zürich: Diaphanes Verlag.
Monod, Jacques. 1973. Le Hasard et La Necessite. Essai Sur La Philosophie Naturelle de
La Biologie Moderne. Paris: POINTS.
Morange, Michel. 2005. ‘What Is Life? A New Look at an Old Question’. In Logic,
Methodology and Philosophy of Science, edited by Petr Valdes-Villanueva Hajek, Dag
Luis Westerstahl, 431–40. King’s College Publications.
Morange, Michel. 2006. ‘Information’. In Dictionnaire d’histoire et Philosohie Des
Sciences, edited by Dominique Lecourt et al, 618–21. Paris: Puf.
Morin, Edgar. 2008. La méthode : Coffret en 2 volumes. Paris: Seuil.
Bibliography 225
Moulin, Anne Marie. 2006. ‘Téléologie’. In Dictionnaire d’histoire et Philosohie Des

Sciences, 4th edition, edited by Dominique Lecourt, 1072–74. Paris: puf.
Neumann, J. von. 1932. Mathematische Grundlagen Der Quantenmechanik. Berlin:
Springer Verlag.
Neumann, J. von. 1956. ‘Probabilistic Logics and the Synthesis of Reliable Organisms
from Unreliable Components’. In Automata Studies, edited by C. E. Shannon and J.
McCarth, 43–98. Princeton: Princeton University Press.
Nunes, Mark, ed. 2012. Error: Glitch, Noise, and Jam in New Media Cultures. New York:
Continuum Publishing Corporation.
Pascal, Blaise. 1966. Pensées. Paris: Bordas.
Pascal, Blaise, and Michel Le Guern. 1987. Les Provinciales. Paris: Gallimard.
Piaget, Jean. 1999. Play, Dreams and Imitation in Childhood. London: Routledge.
Plato. 1997. ‘Alcibiades’. In Plato Complete Works, edited by J. M. Cooper, translated by
D. S. Hutchison, 556–95. Indianapolis and Cambridge: Hackett Publishing Co, Inc.
Porter, T. 1986. The Rise of Statistical Thinking. Princeton: Princeton University Press.
Preda, Alex, Mike O’Hara, and Ricky Treadwell. 2013. ‘Humans vs. Robots: Man Fights
Back’. HFT Review, High Frequency and Algorithmic Trading. Accessed 30 July.
https://c.mql5.com/forextsd/forum/109/alex_preda-hftr-final.pdf
Preda, Alex. 2017. Noise: Living and Trading in Electronic Finance. Chicago: University
of Chicago Press.
Prigogine, Ilya, and Isabelle Stengers. 2009. Entre Le Temps et l’éternité. Paris: Flammarion.
Regnault, Jules. 1863. Calcul des chances et philosophie de la bourse. Paris: Mallet-Bachelier
[et] Castel. Accessed 29 July 2017. http://archive.org/details/calculdeschances00regn
Ruyer, Raymond. 1954. La Cybernétique et l’origine de l’information. Paris: Flammarion.
Sands, S, and J. J. Ratey. 1986. ‘The Concept of Noise’. Psychiatry 49 (4): 290–97.
Schachter, S. 1959. The Psychology of Affiliation. Stanford: Stanford University Press.
Schafer, R. Murray. 1994. Soundscape: Our Sonic Environment and the Tuning of the
World. Original edition. Rochester: Destiny Books.
Schaufeli, Wilmer B., Michael P. Leiter, and Christina Maslach. 2009. ‘Burnout: 35
Years of Research and Practice’. Career Development International 14 (3): 204–20.
doi:10.1108/13620430910966406.
Schneirla, T. 1959. ‘An Evolutionary and Developmental Theory of Biphasic Processes
Underlying Approach and Withdrawal’. In Nebraska Symposium on Motivation, 1959,
edited by M. R. Jones, 1–42. Oxford: University of Nebraska Press.
Schopenhauer. 1851. ‘Über Lärm Und Geräusch’. In Parerga Und Paralipomena, 2,
517–19. Berlin: A. W. Hain.
Schrödinger, Erwin. 1945. What Is Life? The Physical Aspect of the Living Cell.
Cambridge: Cambridge University Press.
Schumpeter, L. 2010. ‘The Silver Tsunami | The Economist’. Accessed 18 July 2017.
http://www.economist.com/node/15450864
Shannon, Claude E., and Warren Weaver. 1964. The Mathematical Theory of
Communication. Urbana: University of Illinois Press.
226 Bibliography
Shenk, David. 1997. Data Smog: Surviving the Information Glut. San Francisco:
HarperSanFrancisco.
Simondon, Gilbert. 2005. L’individuation à La Lumière Des Notions de Forme et
d’information. Grenoble: Editions Jérôme Millon.
Simondon, and JohnGilbert, trans. Cecile M. and John R. 2017. On the Mode of
Existence of Technical Objects. Minneapolis: University of Minnesota Press.
Simondon, G., N. Simondon, J.-Y. Chateau, 2010. Communication et information: Cours
et conférences. Editions de la Transparence.
Smith, Mark M., ed. 2004. Hearing History: A Reader. Athens: University of Georgia Press.
Smith, Steven. 2002. ‘Statistics, Probability and Noise’. In Digital Signal Processing: A
Practical Guide for Engineers and Scientists. 3rd Revised edition, 11–34. Amsterdam
and Boston: Newnes.
Snell, James Laurie. 2012. Introduction to Probability. Providence: American
Mathematical Soc.
Stigler, S. 1986. The History of Statistics. The Measurement of Uncertainty Before 1900.
Cambridge, MA: Harvard University Press.
Taleb, Nassim Nicholas. 2008. The Black Swan: The Impact of the Highly Improbable.
London: Penguin.
Tamir, Eshel (ed.) 2005. Defense Update, International Online Defense Magazine. ‘Long
Range Acoustic Device - LRAD’. Accessed 1 January 2005. http://www.defense-
update.com/products/l/LRAD.htm. LRAD Corporation website. http://www.lradx.
com/site/content/view/15/110/
Tetlock, Paul C. 2006. ‘Does Noise Trading Affect Securities Market Efficiency?’ http://
www.columbia.edu/~pt2238/papers/Tetlock_Noise_and_Efficiency_09_06.pdf
Thomas, Gavin. 2012. ‘Sonic Device Deployed for Games’. BBC, Accessed 1 January
2015, sec. London. http://www.bbc.co.uk/news/uk-england-london-18042528
Thompson, Christopher. 2015. ‘Synthetic CDO Volumes Double amid Hunt for Yield’.
Financial Times. Accessed 18 July 2017. https://www.ft.com/content/cb8d80d8-a323-
11e4-9c06-00144feab7de
Tonnelat. 1996. Thermodynamique et Biologie. Paris: Maloine.
‘Transcript of IMF Seminar – The Risk of a U.S. Hard Landing and Implications for the
Global Economy and Financial Markets’. 2007. Accessed 15 July 2017. http://www.
imf.org/en/News/Articles/2015/09/28/04/54/tr070913z
Voegelin, Salome. 2010. Listening to Noise and Silence: Toward a Philosophy of Sound
Art. New York: Continnuum-3PL.
Volcler, Juliette, and Carol Volk. 2013. Extremely Loud : Sound as a Weapon. New York:
THE NEW PRESS.
Von Humboldt, W., 1836. Über die Verschiedenheit des menschlichen Sprachbaues
und ihren Einfluss auf die geistige Entwickelung des Menschengeschlechts. Berlin:
Königliche Akademie der Wissenschaften.
Von Humboldt, Wilhelm, and Albert Leitzmann. 1903. Gesammelte Schriften. B. Behr.
Accessed 18 December 2014. http://archive.org/details/gesammelteschri03berlgoog
Bibliography 227
Walter, Christian, and Michel de Pracontal. 2009. Le Virus B, Crise Financière et

Mathématiques. Paris: Seuil.
Watkins, Calvert, ed. 2011. The American Heritage Dictionary of Indo-European Roots.
3rd edition. Houghton and Mifflin: Harcourt Publishing Company Inc.
Watson, John B., and Rosalie Rayner. 1920. ‘Conditioned Emotional Reactions’. Journal
of Experimental Psychology 3 (1): 1–14.
Weizsäcker, Carl F. von. 1994. Aufbau Der Physik. München: DTV Deutscher
Taschenbuch Verlag.
‘What Are Conduits, SIVs and SIV-Lites?’ 2017. Accessed 16 July 2017. http://
uk.reuters.com/article/factbox-conduits-sivs-idUKNOA13480720070911
‘White Phosphorus: Weapon on the Edge’. 2005. Accessed 16 November 2005. http://
news.bbc.co.uk/1/hi/world/americas/4442988.stm
‘WHO Guidelines for Community Noise’. 1999. Geneva: World Health Organization.
Accessed 17 June 2014. http://www.who.int/docstore/peh/noise/Comnoise-1.pdf
Wiener, Norbert. 1961. Cybernetics; or, Control and Communication in the Animal and
the Machine. New York: The MIT Press.
Wiese, Wanja, and Thomas K. Metzinger. 2017. Vanilla PP for Philosophers: A Primer
on Predictive Processing. Philosophy and Predictive Processing. Frankfurt am Main:
MIND Group. http://predictive-mind.net/papers/vanilla-pp-for-philosophers-a
-primer-on-predictive-processing/getAbstract.
Wilkins, Inigo. Forthcoming. Irreversible Noise. Falmouth: Urbanomic.
Wilson, Paul. 2012. ‘Physical Spectatorship: Noise and Rape in Irreversible’. In
Reverberations: The Philosophy, Aesthetics and Politics of Noise, edited by Benjamin
Halligan, Paul Hegarty, and Michael Goddard, 122. London: Continuum.
Wilson, Robert W. 1978. ‘The Cosmic Microwave Background Radiation’. Accessed 6
August 2014. www.nobelprize.org http://www.nobelprize.org/nobel_prizes/physics/
laureates/1978/wilson-lecture.pdf
Winograd, S., and J. D. Cowan, 1963. Reliable Computation in the Presence of Noise.
Massachusetts: The MIT Press.
Zimmermann, Rainer. 1988. ‘Ordnung Und Unordnung - Zum Neueren
Determinismusstreit Zwischen Thom Und Prigogine’. Lendemains, 50: 60–73.
Index
ABCP. See asset-backed commercial Bayesian data analysis 100

papers (ABCP) behaviour, defined 197
Abramowicz, Lisa 127–8 Bell, Eric Temple 58
absurd pessimism 130 Bernoulli’s law 1
accidental entropy 198 beta-blockers 204
accidental information 97–101 big data 2, 10
acoustic noise 143, 149, 153, 156, 170 Bijsterveld, Karen 144
additive white Gaussian noise (AWGN) 99 historical analysis 154
afferent noise 98, 127 biosemiotics 11
Alcibiades 213–18 bit 40
lack of knowledge 215 Black-Scholes option price model 123
wavering and confusion 215 Blanche, Robert 57–9
alethurgy 214 Boltzmann, Ludwig 4, 27, 31, 36, 200
Aliber, Robert 130 entropy 32, 69, 93, 95
alternative mathematical models 130 formula to physical reality 29
Altmann, Jurgen 157 Brillouin, Leon 4, 17, 35, 65–9, 76, 182
ambiguity, margin of 32 negation of entropy 71
analogy 2 neologism 80
‘analytical’ philosophy 88
Anglo-Saxon humanities 87 Canguilhem, George 10, 12, 87–90, 184–5,
Anglo-Saxon philosophy 87 191–2
Anglo-Saxon theory of knowledge 87 concept of normativity 191–2
anxiety 195–8 Cantor, Georg 58
Aristotle 44–5 Cartesian Method 10

Arno Penzias and Robert Wilson 109 catastrophe theory 89
artificial intelligence 208 catastrophic reaction 175, 180, 181, 187–9,
art of governing 213, 215 195–7, 204
Ashby’s law 112 definition of 215
asset-backed commercial papers (ABCP) Goldstein’s concept of 180, 191
126 to noise 199
astigmatism of intuition 79–83 psychiatric definition of 199
Atlan, Henri 10, 35, 75, 112 Cavaillès, Jean 88, 106, 116–17
audible noise 158–9, 162 CDOs. See collateralized debt obligations
audiometers 145 (CDOs)
AWGN. See additive white Gaussian noise Central Limit Theorem 99
(AWGN) chemico-diffusional coupling 81
axiomatic set theory 59 choice
freedom of 57–60
Bachelard, Gaston 8–9, 88, 117 logic of 57–60
background noise 109 clash of cultures 158
Badiou, Alain 9 Clausius, Rudolf 4
basis of noise 71–7 clonidine 204
Index 229
closed systems 81 cybernetics 1, 2, 5, 11, 16, 68, 170, 209,

closure 187–9 211, 213
CMOs. See collateralized mortgage control of noise in 204–5
obligations (CMOs) inclination 170–1
cognition, processes of 207 information model 69
cognition, process of 171 ‘ironmongery’ 89
collateralized debt obligations (CDOs) origins of 207
126–7 second-order 5, 205
collateralized mortgage obligations theoretical influence of 206
(CMOs) 126 theory of 205
colonialization 172 Cybernetics (Wiener) 16
combined sound energy 151 cybernetic theory 171
complex systems reliance on 171
basis for 207
contemporary modelizations of 207 declining life 187–9
theory 211 deductive redundancy 57, 59, 82
confusion 169–77 Desrosieres, Alain 136
contemporary humanism 44 determinism 200
contemporary philosophy 199 Diaz Nafria, Jose Maria 23, 24
context of psychiatry 203–4 discourse, ‘de-differentiation’ of 180
continental philosophy 86, 87 discursive ambiguity 17–18
contingency 215 discursive interpretation 17–18
control 203–6 dispersion of energy 3
contingency vs. 203 docte ignorance 215
cybernetic concept of 216 Duchamp, Marcel 104–5
cybernetic notion of 203, 216 dynamical state of system 209
of noise pollution 155
controllability 216 eclecticism, impression of 171
limits of 210 ecology of code and vibration 161
control law 209 economic development 149–50
control mechanisms 207 ecosphere 2, 4
mathematical analysis of 208 efferent noise 98, 127

control theory 207, 209, 210 Eigen, Manfred 81
cybernetics’ origins in 207 Eisler, Zoltan 119, 141
emergence of 208 electronic signal transmission 19, 31, 62
limits of 207 emotional agitation 204–5
mathematics in 210 empirical noise 167–8. See also noise
uncertainty in 211 empirical reality 8, 45, 49, 81, 94
Copernican revolution 46 energy in thermodynamic terms 4
cosmic background radiation 109–13 entropy 4, 15–16, 39–40, 50, 62, 67, 73, 83
cosmic expansion 111 accidental 198
cosmic noise 109 Boltzmann statistical concept of 95
cosmos 111–12 definition of 32, 69, 136, 199
Crutzen, Jozef 44 dissipation of energy 111–12
crystallization 72 as ‘Freedom of Choice’ 23–6
process of 72 of information (see information
cultural theory 144 entropy)
culture 20 informational value of 79
Cummings, David 6 of message 15, 98
cybernetician 112 negation of 66–7, 80, 112, 182
230 Index
and noise 28, 202 ignorance, knowledge vs. 216

perception of 80 imitation 188
physical 27–8, 132 immorality of speculation 127–8
statistical 3, 93, 95, 200 impairment, hearing 155
environmental noise 156 inaudible noise 159
Enzensberger, Hans Magnus 115 inconceivable freedom of choice 198
epistemic humility 8, 140 inconceivable multiplicity of choice 197
epistemological noise 7–10, 33, 60, 69, 95 individuation
epistemological normativity 106, 207 empirical aspects of 45
epistemological obstacles 8–9, 64 process of 43–50
epistemology 11 theory of 48
essential human trait 181–4 industrialization 143
etymology of noise 213 inforgs 2
Euclid’s geometry 58 information 8, 16, 41, 45, 117, 133, 134,
exercise control 202 141
extrinsic noise factors 175–6 concept of 4, 5, 17, 69, 71
definition of 16, 17, 23, 134, 140
Floridi, Luciano 1, 11, 93, 105 entropic conception of 15
Foucault, Michel 10, 214 entropic definition of 61, 76
revaluation of error 89 epistemological equation of 134
freedom of choice 16, 17, 28–30, 33, 39, ethics 11
51, 53–4, 57–60, 69, 74–6, 82, 95, iconoclastic definition of 20
182, 188, 197, 198, 202 vs. noise 61–2, 198
French epistemology 87, 88 paradigm 2
Freudian notion of regression 189 philosophy of 11
physical concepts of 35–7
Galton, Francis 140 politics of 6
gambling behaviour 122 process of 39–41, 43–50, 63, 98
Gaussian noise 98 qualitative definition of 47
Gelman, Andrew 97, 100–1 quality of 49
Goldstein, Kurt 175, 195–8 quantity of 67
catastrophic reaction 191, 192 ready-made 103–7

clinical observations 175 Shannon’s definition of 90
Goodman, Steve 160 statistical conception of 16
governance 213 systems 11
political 214 Wiener definition of 16, 82
Greene, Marjorie 59 informational potential 82
information culture 11
Hainge, Greg 11 information entropy 3, 4, 15–19, 24, 26–9,
hearing impairment 155 31, 33, 36, 48, 51, 53, 62–3, 64,
consequences of 155 66–69, 71–7, 83, 90, 103, 105, 109,
helmsman metaphor 207–11, 214, 216 182, 188, 189, 197–8
Herbert, Martin 30 of complex systems 77
homeostasis 71, 170 Shannon’s definition of 23, 68, 76
physiological concept of 205 information process 54
hormone noradrenaline 204 information theory 1, 5, 11, 16, 143
humanity 44 noise 1–2
hylomorphism 44–5 infosphere 2
hyperbolic geometry 58 internal chaos 167–77
hyper complex systems 76 internal disposition 175, 176
Index 231
intrinsic noise factors 175–6 Maxwell, James Clerk 93, 208

intuition 79–83 MBS. See mortgaged-backed security
astigmatism of 79–83 (MBS)
irreversible/irreversibility 7, 201 Mead, Margaret 133
Medical Protection survey 177
Keats, John 9–10, 181 mental health, assessment of 177
negative capability 182–4 mental state of noise 93, 169, 170, 173–4,
Kertesz, Janos 141 176, 179, 181, 183–4, 187–9, 195,
Kindleberger, Charles 130 203–6, 215–19
knowledge 133 definition of 188
about noise 145 message’s entropy 98
academic organization of 12 metaphor 7
Anglo-Saxon theory of 87 critical use of 8
discipline of 141 of helmsman 207–11, 214
domain of 94, 136 of noise 8
forms of 2, 106 role of 7–8
vs. ignorance 216 metastability 71–3
lack of 215 molecular entropy in thermodynamics 27
theory of 6–7 moral dichotomy 121, 129
krach 130–1 moral philosophy 121, 125, 129
Krugman, Paul 130 Morange, Michel 201
Kuhn, Thomas 93–4 Morin, Edgar 109, 111, 113, 115
kybernetes 207–11 mortgaged-backed security (MBS) 126–7
motor governors 207–8
labour, division of 141 MTC. See Mathematical Theory
Lacanian psychoanalysis 172 of Communication (MTC);
lack of knowledge 215 mathematical theory of
law of errors 1, 137 communication (MTC)
law of possibilities 1
Leriche, Rene 171 necessity 51–5, 57
Lillo, Fabrizio 141 conceptual presence of 52
linguistics, structuralist analysis of 172 negation of entropy 16–17

Lobachevsky, Nikolai Ivanovich 58 negative capability 9, 181–5, 189
Lockwood, Dean 161 definition of 181
Logical Positivism 59 negentropic attitude 197
logic of choice 57–60 negentropic function 188
Long Range Acoustic Device (LRAD) 157 negentropic process 112
LRAD. See Long Range Acoustic Device negentropy 3–6, 33, 36, 65–9, 74, 76–7, 80,
(LRAD) 82–3, 182, 184, 188, 189
new communication technologies (NTCs)
magisterial discourse 116 105
Manichean dualism 106 Newtonian physics 100
market variations, real causes of 121 noise 8, 79, 117, 150
Markoff process 39 abatement 143–8
mathematical formalization 17 amplification of 205
mathematical noise 167–8 avoidance of 173
mathematical theory of communication basis of 71–7
(MTC) 3–4, 15, 24, 66, 67 catastrophic reaction to 191–3, 199
theoretical relevance of 52 concept of 167–9, 170, 173, 199
maximal entropy 28, 48, 71–2, 80, 82, 189 control of 211
232 Index
definition of 162, 169, 216 normativity 106, 191–2, 217

dimensions of 144 norms of living 217
Dutch opinion on 146 no-touch torture 158
entropy and 18–19, 28, 202 NTCs. See new communication
epistemological aspect of 197 technologies (NTCs)
etymology of 135, 213
in finance 119–32 On the Mode of Existence of Technical
in gap between narratives 115–17 Beings (Simondon) 20
impact on learning 156 openness 110, 177, 188, 189
information and 2, 198 ‘essential human trait’ of 181
information potential of 110 open systems 81
knowledge about 145 ‘optimization’ for control 209
mental state of 173 order 199–202
metaphors of 7–8 from noise 2
military history of 11
multiple sources of 156 parallel postulate 58
participation of 75 parrhesia 214
politics of 6 Pascal, Blaise 1
problem of 150 Penzias and Robert Wilson 109–10
psychiatric notion of 172 perception, cultural codes of 160
quantification of 152 Petty, William 136
quantitative measure of 150, 152–3 phenomenological astigmatism 83
scientific and cultural understanding philosophical discourse 7–8
of 15 philosophy, analytic objectives of 86
scientific and technological mastery Phyllis Greenacre’s 1952 study 179
of 93–4 physical analogy 32
sources of 110 physical entropy 15, 27–8, 32, 33, 54–5
speculative dimension of 167 physical potential, nature of 81
as spurious uncertainty 61–4 physical spectatorship 159
statistical 1 physics
stochastic models of 10 information and informational
in stock market 119 concepts of 35–8

theoretical polymorphism of 144 statistical method in 201
transduction of 93–6 Piaget, Jean 188, 189
of ventilation system 151 Picavet, Emmanuel 90
well-worn metaphor of 161 Planck, Max 28
Noise Matters (Hainge) 11 Plato 45, 213–18
noise pollution 2, 3, 89, 149–54 Platonic inspiration 87
concept of 152–3 Plato’s concept of form 44–5
control of 155 pluralization of logics 59
definition of 159 political governance 214
sources of 151 pollution
WHO report on 148 noise (See noise pollution)
noise toxicity 155–7 pollution, noise 2
definition of 155 post-modernism 86
noise traders 120–1, 123–5, 129 post-structuralism 86
non-audible noise 159, 160, 162 ‘post-truth’ elections 214
non-classical mechanics 93–4 potential information 29–33
‘non-Euclidian’ geometry 58 actualization of 33
The Normal and the Pathological potential, physical concept of 31
(Canguilhem) 10, 87, 191, 193 Pracontal, de Michel 129–30
Index 233
pre-anxiety tensions 179, 187 SEL. See sound exposure level (SEL)
Preda, Alex 122, 124, 125, 208 self-control 215, 216
predictability, advantage of 99 self-critical method 172
predictable noise 97–101 self-governance 215
probability 15, 200 self-organization 81
process of individuation 43–50 ‘self,’ presuppositions of 176
protocols of observation 140 self-regulating systems, Winer’s cybernetic
proto-mythological thinking 8 theory of 4
psychiatric illnesses 169 self-regulation 205
psychiatry 170 sense of governing 207
psychology 8 serendipity, instance of 97
psychopathology, forms of 173 servomechanisms 207
public health 149 Shalizi, Cosma Rohilla 97, 100–1
Shannon, Claude 3–4, 11, 17, 23, 24, 46–7,
qualities 139–41 49–52, 62, 69, 103, 137
quantity of information 23, 28, 36, 40, 51, audacity 15
65, 68, 76 concept of information 4, 25, 32–33,
quantum physics 79, 88, 93, 94, 112, 200 39, 46–7, 64, 68, 69
Quetelet, Adolphe 139 conceptual audacity 85, 105
statistical idealism 141 conceptual debt 16
cultural significance of 90
radio noise 109 definition of information 18, 20, 23, 25,
radio silence 144 35, 36, 103, 136
Ratey, John 171–4, 179–85, 187–9, 195, designation of information 27
203–6, 215 entropic ideas 18, 19, 25, 29, 35, 36, 66,
ready-made information 103–7 68, 85, 89, 90, 105
recuperated disorganizations 73 formula 40
redundancy 51–5 information entropy 93
deductive 57 information theoretical concepts 112
Regnault, Jules 121–3, 125, 129, 139 MTC 24
moral dichotomy 121 quantitative measure of information 32
rehabilitation 192 shift of emphasis 43

relative entropy 51, 189 Shapiro, Meyer 181
relative uncertainty 106 Shostak, Stanley 201–2
remote controls 207 signal-to-noise ratio 151, 155–6, 162
requisite variety 75 signal transmission 20, 109, 142
road and air traffic noise 144 mathematical theory of 3
role of teacher 214 perturbation of 110–11
Roubini, Nouriel 130 Simondon, Gilbert 20, 44, 45, 48, 49, 71,
Ruyer, Raymond 2, 80 80, 94, 95
epistemological appropriation 95
Samuelson, Paul 122–3 theory of individuation 47
Sands, Steven 169, 171–4, 179–85, 187–9, simple tautologies 59
195, 203–6, 215 social life, technologies in 145
Schachter, S. 180 sociology 8
‘schismatic morphogenetic’ process 111 Socrates 217
Schrödinger, Erwin 67, 80 Socratic maxim 85
nascent theory 68 sonic ecology 167
scientific discourse 21 sonic weapons 157–9
Scott, Peter 99 sound. See also noise
second order cybernetic system 5, 124 cannons 158
234 Index
culture 147 traffic noise 151

ecology 161, 162 transdisciplinary discourse 8
levels of 155 transformation, process of 45
military deployment of 158 true market value 122, 139
sound exposure level (SEL) 151
special purpose vehicles (SPV) 126 uncertainty
SPV. See special purpose vehicles (SPV) degree of 23
Stalinism 172 philosophical cultures of 86
statistical analysis 15, 17 recalcitrant cultures of 86
statistical entropy, formalization of 3 urban white collar workers 150
statistical noise 1, 2, 158
definition of 1 ‘vicious whir’ of sensations 179–85
statistical physics 139 vision, cultural dominance of 159
statistics 133–7 visual spectatorship 159
implications of 141 Volcler, Juliette 157–9
stereotypies 175 von Foerster, Heinz 112
stimuli disposition 176 von Humboldt, Wilhelm 90
Stoermer, Eugene F. 44 von Neumann, John 112
subjective noise 167–8
subjectivism 137 Walter, Christian 129–30
subprime crisis 119, 125, 130, 131 war on terror 158
symbolization 188 Weaver, Warren 16, 39–40, 51, 52, 63–7,
synthetic CDOs 126–8 97
systematic communication 63 definition of noise 103
systematic imitation 188 idea of noise 64
white noise 79, 80
taxonomy, Conrig’s system of 136 WHO. See World Health Organization
teacher, role of 214 (WHO)
technical noise 167–8 Wiener, Norbert 4, 16, 17, 69, 122, 205
technicity 20 concept of negentropy 47
technological innovation 143 cybernetic definition of information
teleonomy 201 69, 85

terror 169–77 cybernetic theory 17
Tetlock, Paul C. 123–4 definition of information 16, 47, 82,
theory of individuation 48 136
thermal noise 143 origins of cybernetics 208
thermodynamics understanding of information 135
molecular entropy in 27 Wilkins, Inigo 7
principle of 111 Wilson, Laura 159
Thompson, Christopher 126 World Health Organization (WHO)
Toscano, Alberto 9 148–51
toxicity of noise 159

Cecile Malaspina An Epistemology of Noise From Information Entropy To Normative Uncertainty

Caricato da

Informazioni sul documento

Titolo originale

Copyright

Formati disponibili

Condividi questo documento

Condividi o incorpora il documento

Opzioni di condivisione

Hai trovato utile questo documento?

Questo contenuto è inappropriato?

Copyright:

Formati disponibili

Cecile Malaspina An Epistemology of Noise From Information Entropy To Normative Uncertainty

Caricato da

Copyright:

Formati disponibili

Copyright © 2018. Bloomsbury Publishing PLC. All rights reserved.

Noise Matters, Greg Hainge

Foreword by Ray Brassier

BLOOMSBURY, BLOOMSBURY ACADEMIC and the Diana logo are trademarks of

First published in Great Britain 2018

Copyright © Cecile Malaspina, 2018

Cover design by Catherine Wood

All rights reserved. No part of this publication may be reproduced or

Bloomsbury Publishing Plc does not have any control over,

ISBN: HB: 978-1-3500-1178-6

Typeset by Integra Software Services Pvt. Ltd.

Goya Series: They Do Not Agree, 1997 courtesy of John Baldessari

Part 1 Concepts: Information Entropy, Negentropy, Noise 13

I How to Draw the Line between Information and Noise 15

II Entropy as ‘Freedom of Choice’ 23

III Information Entropy and Physical Entropy 27

IV The Idea of ‘Potential Information’ 29

V Physical Concepts of Information and Informational

VI Information as Process Rather Than Content 39

VII To Think about Information as a Process of Individuation 43

VIII Redundancy and Necessity 51

IX Logic and Freedom of Choice 57

X Noise as Spurious Uncertainty 61

XII Complexity on the Basis of Noise 71

XIII The Astigmatism of Intuition 79

XIV The Path of Despair 85

Part 2 Empirical Noise 91

I On the Transduction of the Concept of Noise 93

II Accidental Information, Predictable Noise 97

III Ready-Made Information 103

IV Cosmic Background Radiation 109

V Noise in the Gap between Narratives 115

VI Noise in Finance 119

VII Statistics: The Discipline of the Prince 133

VIII The Man without Qualities 139

IX Noise Abatement: The Dawn of Noise 143

X Noise Pollution 149

XI Toxic, Viral, Parasitic 155

Part 3 The ‘Mental State of Noise’ 165

I The Crossroads: Mathematical, Technical, Empirical

II Internal Chaos, Terror and Confusion 169

III The Vicious Whir of Sensations 179

IV Keat’s Negative Capability 181

V Closure to Noise and the Paradox of the Declining Life 187

VI The Catastophic Reaction to Noise 191

VII Anxiety 195

VIII Order 199

X The Helmsman Metaphor: Kybernetes 207

XI The Helmsman in Plato’s Alcibiades Dialogue 213

and George Canguilhem. By engaging these sources, Malaspina achieves a

From this follows what is perhaps Malaspina’s most striking insight:

Noise, beyond the reference to unwanted sound, thus reveals itself to be

conceptually polymorphous because it has never been about types, classes or

cannot be naturalized, objectified, catalogued or classified, this anomalousness,

more than it is either verifiable or falsifiable. It is ‘of ’ uncertainty because it is

judgement, constituting cognition against the backdrop of its dissolution, is a

1 Inigo Wilkins, Irreversible Noise, Falmouth: Urbanomic, forthcoming.

ILFI Simondon, Gilbert. 2005. L’individuation à La Lumière Des Notions de

METO Mode of Existence of Technical Objects, Gilbert Simondon, trans.

MTC Mathematical Theory of Communication, Claude Shannon and

NP The Normal and the Pathological, Georges Canguilhem, trans.

WHO World Health Organization, http://www.who.int/about/en

understanding of noise as inextricably linked to variation or even error goes