Sei sulla pagina 1di 27

CnaprER Sx

Decision Engineering

fD revious chapters have subordinated questions about the


.f quality of decisions or their improvement to understanding
decisions and decision processes as individual and social phe-
nomena. This chapter considers some aspects of the former
questions. It examines a few basic complications that arise as
decision makers try to achieve intelligence.
Claims of intelligence have been used to justify various deci-
sion procedures, including both anticipatory rationality and his-
tory-based rule following. Advocates of rational action empha-
size the intelligence provided by explicit calculation of future
consequences. They decry the blind conservatism of rule fol-
lowing. Advocates of rule followirg, on the other hand, em-
brace the intelligence captured by accumulated experience rep-
resented in rules. They criticize the excessive informational and
cognitive requirements of rationality.
The argument between the two sides echoes an old debate in
social philosophy between those who see wisdom in willful, cal-
culated action and those who see wisdom in the lessons of histo-
ry as coded in traditions. The argument is not settled in this
chapter. On the contr&Íy, the discussion here suggests that
221
222 A PRTMHR oN DucrsroN MAKTNG ':

Decision Engineering 223


many forms of decision making, including
both rationality and
rule followiilg, ate yefy-l procedures for were deemed good. Alternatively, actions are sometimes de*
decision making, but
no form guarantees intelligence. scribed as "tational" or "procedurally ration al" if they were
That conclusion is emuéooed in a simple made by a procedure that assesses expected consequences and
introduction to de,
cision engineering. section 6.l considers chooses actions that are expected, on averago, to lead to de-
the problems of defin.
ing decision intelligence in terms of the sired outcomes. Similarly, an organization is sometirnes de-
oütromes decisiong
produce and the complications of
making the tradeoffs rs. scribed as "learning" or "substantively learning" if its perfor-
quired when outcomes are realized
across ii*"
mance improved (regardless of the process by which that
in the form of both symbolic and ruurruntive and space and improvement was realized) and is sometimes described as
returns. Sectiong
6'2' 6'3, and 6'4 examine some of the
possibilities for enhancing "learning" or "procedurally learning" if it used inferences
decision inteltigence, considering
,o*. aspects of how decision drawn from previous outcomes to adjust behavior (regardless
engineering can improve adaptiu-rr"rr,
tt , use of knowledgo, of whether those inferences led to adjustments that improved
and the creation of meaning. outcomes).
The confusions between process and outcome definitions of
6.1 Defining necision Inteiligence
rationality, learning, and other forms of decision making are
understandable. It seems natural to believe that decision mak-
Decision engineering is dedicated to ers will generally do well if they choose actions that are expect-
producing decisions that
ate intelligent, but the definition
of intelligence is often left u'- ed to lead to desired consequences. Similarly, it seems natural
clear' students of decision making osciliate to believe that outcomes will improve if a decision maker ad-
between process
and outcome definitions of intellifence justs behavior by avoiding actions that in the past have been as-
and have never been
able to resolve satisfactorily ,o*J difficult sociated with bad things and reinforcing actions associated with
issues associated
with key tradeoffs underlying the definition good things.
of good outcomes.
The problems are deep enough and
have been known long Convictions that good outcomes will be achieved if decision
enough to suggest that hopes for makers follow a particular preferred process underlie much of
their early resolution are
probably misguided. the normative literature on decision making. In standard trea-
tises extolling the beauties of analytical decision making, the
preferred procedure is rational calculation. In standard treatis-
6.7.1 Outcomes and processes
es extolling the beauties of learning, the preferred procedure is
Discussions of decision making and experimentation and incremental change. In standard treatises
decision engineering are
plagued by confusions over rh¿
mulripte meantü;;ili"re1i- extolling the beauties of fundamentalist ethics, the preferred
gentr" "tationalr" "learningr" and
various similar terms. fn this procedure is literal consultation of sacred texts. The possibility
book, and particularly in ttris chapter, that following the rules of decision theory, learnitg, or funda-
distinctions are made
among these terms. The distinctions-though mentalist ethics may sometimes lead to poor outcomes is ob-
not necessarily
any particular nomenclature-are
essential to discussions of scured by a definitional sleight-of-hand that equates outcome
decision engineering. intelligence with process reliability.
Words like "tational" and "learn ing" Observations of decision making, as well as more precise
ate used in two quite
distinct ways' Past actions are sometimes specifications of the models, suggest that these presumptions
described as ,,ratie-
nal" or "substantively ration al" if they are not, in general, warranted. The links between processes and
,rrrrlted in outcomes that
outcomes cannot be assumed. They, must be demonstrated.
224 APRIMERoN DECISIoN MAKING
Decision Engineering 225
This chapter examines questions of whether
and when decision of changing its standard rules and responds to positive feedback
processes are likely to lead to good
outcomes and the possibili- by decreasing the chance of changing its standard rules (regard-
ties for improving rhose ptor"ú"r. The
intelrig""r"ái in action less of whether the learning led to improvement in performance).
is defined in terms of itsbutcomes. Anaction
is defined as intel- Those definitions make rationality and learning qc ante con-
ligent if, after an the results.are-in (including
porriul. .t ung* cepts, properties of procedures and not defined by their out-
in preferences and identities), it has satisfied
the wishes of rele- comes. Rational procedures may lead to good outcomes or to
vant parties. In this view, intelligence is un
u por, iincept. In- poor outcomes. The remainder of this chapter examines the op-
telligence cannot be determined until the results
are known. An portunities and limitations of various possible procedures for
outcome-based definition of inteiligence
excrudes ü making decisions in organ izatioils, including ration ality, learn-
the form: "she actedintelligentty b-ut the ""*""i,
outcomes were poor.,, irg, multiple actor conflict, and ambiguity.
An outcome-based definition of intenigen." ,roi onlv
intelligence arr erc posr assessme"i¡ut alio
rur.r,
makes irrtáIig"n.,
subjective. since the intelligence of an
action is ¿etermined by 6.7.2 The Tiadeotrs of Intelligence
the value of its outcomes, it depends on
the pt"r"r"*rs and
identities of the rerevant actor ór actors. Normall¡ If intelligent decision making means decision making leading to
of course,
an action is associated with a specific decision outcomes that contribute to a decision maker's well-being or to
decision makers, and its outcome is assessed -ui", or set of
the well-being of a group, organizatior, or society, three diffi-
in terms of the val-
ues of those decision makers (or those
for whom tnel acteo as cult questions need to be addressed: (1) To what extent should a
In assessing multiple actor actions, however, it may be decision process and a decision be judged by their contributions
1gg_"t*).
difficult to speciff the relévant decision makers, to action and the consequences of action as opposed to their
and the rele_
vantvalues may be obscure. Outcome intelligenc", contributions to the de¡¡elopment of meaning and an interpre-
u, u ,"rult,
maybe indeterminate. tation of life? (2) How should outcomes that are distant in time
be weighted in the overall evaluation relative to outcomes that
. Rationality, learning-based rule following, coalition forma-
tion, imitation, and temporar sorting decislón are near? (3) How should the wants, values, and needs of differ-
treated herc as processes. For example, a -"kiü are alr
ent individuals and groups be weighted? None of the questions
decision is d"escribed
as "rational" if it is made by a process has an obvious answer.
that foilows ,turráuio pro-
cedures for choosing among alternatives
in terms of eapecta-
tions about future consequences. Decisions SYMBOLS AND SUBSTANCE
by a firrancial'orga-
nization are rational, in these terms, if the
organization rnat<es a Decision making allocates scarce resources and achieves some
systematic analysis of the probable future
consequences of pos- of its importance in that way. At the same time, however, deci-
sible alternatives and seleóts that artern"i*
pected value (regardless of whether the
rrrái-J-i"r r"- sion making also displays values and interprets life. Decision
decisions read to good . making is a central part of modern Western ideology. It is
outcomes).
linked to key concepts of the Age of Reason, such as intentional
Similarly, a decision process is described
as being a form of human control over destiny and human will. Decision processes
"learning" if it involves changing behavior i"o"."ituuy
in re- are imbued with symbolic content. They exhibit and reassert so-
sponse to feedback about performance.
Thus, a manufacturing cial beliefs, dramatize commitments to a faith of deliberate and
organization is described aslearning if the
effective human action, and provide opportunities for making
by"rgit"ri",,"¡p"r¿,
to negative feedback on its performánce
inlrearinj tn" .tun., individual statements that fit an individual into that faith. Deci-
226 A pRtMlJRoN DuctstoN MAKINc
Decisk¡n Engineerin¡¡ 227
sion making is organized to share stories,
e4planations, and Alternatively, decision engineers can see the development of
symbolic meaning as well as to take action.
adequate rituals and symbolism as a primary responsibility. The
For example, decision makers often seem
. to treat the gather-
ing and use of information as part of the pur.rrii
symbolic aspects of decision making can be embraced and elab-
oiiymbolic orated as reflecting a fundamental aspect of decision intelli-
meaning rather than as part of the resolution
of decision uficgf. gence. Decision making can be seen as developing myths and
tainties. Gathering information and making
decisions-are sig. interp retations of life and as modifying the diffuse beliefs and
nals and symbols of competence. The porr"iion
uná-"*t iuition cultural understandings that make events comprehensible.
of informarion symboüzés (and oemónstratej;;
With that in mind, decision engineering may want to be con-
legitimacy of decision.makers. A good
decisián .ut", "bü;;is one cerned with improving the effectiveness with which decision
who makes decisions in a propet way,
*rro Jilit, lapertise symbols develop and maintain social beliefs about choice, rea-
and uses genera'y accepted infórmatión.
tations among decision makers stimulates
c"-p"ii,.n iá,,"pu- son, power, conflict, and intelligence. The object is to improve
pro^orr"tio' uo¿ aon- human commitment to rationality and identity, to build bases
spicuous displays of information well
beyond what is.rrJ to ,r. for community and diversity. I)ecision rituals can be construct-
solve substantive uncertainties. oeóision
information and do not use it; ask for more
il;;;- gather ed and performed to accomplish such ends. The main impedi-
arra igrror" it; make ment is an inability of decision makers to think about meaning
decisions first and rook for rerevant
informatián afterward; as fundamental, thus an inability to focus on the design of ritu-
gather and process a great deal of information
that has littre or als and the organization of symbols.
no direct relevance to decisions.
The qnnbolic significance of decision
making can be seen as a
corruption. There is rittle doubt that rhe,y*uJrir- COMPARISONS ACROSS (NE,STED) TIME
results in misalrocation of decision energy,
áii""irion,
in manipulation of Decision outcomes unfold over time. The short run is nested in
the innocent by the sophisticated, and
in "armam"n,r ru."rll J, the long run. Many actions that contribute to short-run well-
purveyors of technorogical and organizationar
guog"t, i"ou". being ate deleterious in the long run, and vice versa. Moreover,
competition in their own versionr ár conspicuous
information preferences and identities change over time, partly as a result of
consumption. From this standpoint, ,
_decision "rrg;";;;4"r", taking actions. Are outcomes to be evaluated in terms of pref-
want to resist the introduction of symbolic
sion processes or b"f:l the makin! of
.t"-""irlnio or.i erences and identities that existed at the time of the decision or
decision, rrü-¡L"i, .r- in terms of those that exist at the time at which the effects of the
fects. The former is difficurt to acco"mplish
on a unilaterar basis, ,
decision are realized?
but the latter offers some hope. It is ómedmes
possible to seg- The complications of weighting consequences that are dis-
regate the symbolic and substantive
aspects tributed across time constitute a prime topic in both the psy-
For.exgmde, information gathering unc "r¿á.lri""tunng. chology and the economics of choice. Psychologists of individ-
symbolic significance, is sómetimei butter"d ""J;rio;il1;;""ry
fro- ,rr";ur.ing ual development have long noted the difficulties involved in
of choices. choices are frequently negoti""¿
,n" .on- delayed gratification, in forgoing current pleasures in order to
text of e4plicit decision proi"r."i, wh-ere ""irrá"are more
the rituals lay the basis for future pleasures. Individual eaters implicitly
"¡;;;;; *;
h"uytY symbotic. observatior;
es indicate that decisiotr pro."rLs exercise
üffi''oioirrr_ compare the immediate delights of eating chocolate cake with
probrems more the more uncertain future costs of weight and health problems.
oftg.n than they solve them, and that
the substantive choices Subsequently, they make the same comparison from the van-
made are frequently made outside formal
decision t;;";r, tage point of time. The problem is defined as one of learning to
228 A PRIMER ON DECISION MAKINC :

Dec:isktn llngineerln¡¡ 229


give more weight to anticipations of delayed pain and pleasure
are made. It can be approximated, however, particularly if deci-
and to give less weight to immediate pleaiupr and pains.
sion makers attend not only to the unfolding consequences of
The same problem is discussed ir the literature of econoffi-
their actions but also to the unfolding effects of their actions on
ics, particularty in consideration of the choice between
current their preferences and identities.
consumption and savings (future consumption). As numerous
critics of delayed gratification have obsenrld, there is no ileces-
sary reason for giving more or less weight to expectations COMPARISONS ACROSS (NESTED) DFCISION MAKERS
of fu.
ture pleasures and pains then to current ones. Thóre are Students of decision making intelligence have ordinarily been
pathologies of saving as well as pathologies of consumption.
In uncomfortable with the idea of intelligence when applied to de-
stand atd economic treatments, the choice between savings
and cisions involving multiple inconsistent actors. Intelligence has
consumption depends on some discounting factors that are
generally been treated as requiring consistent preferences or
partly based on individual values (weights¡ aáO parrly based
on identities. The well-being of a decision maker may be inconsis-
the fact that savings accumulate earnings.
tent with that of others, as well as with the well-being of an or-
Economic discussions emphasize ro*paring present and
future ganization or a larger community of which the decision maker
costs and benefits in terms of their "present uul.res." The
idea of is a part.
present value calculations is to make present and future
expecta* A limited form of intelligence has been found in the Pareto-
tions monetarily equivalent by estimating the amount of current
preferred criterion. Where an action will not hurt anyone (itt
money, properly invested, that would grow (ot decline) to
the his or her own terms) and will help at least one person (in his or
level of the future expectation by the time the future has
arrived. her own terms), taking the action seems unquestionably more
Since present value calculations are normally made in terms
of intelligent than not taking it. The classic Paréto procedure is a
money, rather than the value to the decision maker of the goods
form of unanimity in which all persons vote "yes," "no," or
money buys, they avoid the more difficult complications involved
"don't care r" andaction is taken only if there afe no "no" votes
in evaluating the future subjective value of future realizations.
and at least one "yes" vote. Bilateral voluntary exchange is one
Comparisons across time are particularly difficult because
such procedutre.
preferences and identities are not stable. They are changing,
Given the severe restrictions imposed on choice by Pareto
and changing in part as a result of decision making. It is fre-
decision rules, some social welfare theorists have sought to de-
quently argued that the intelligence of an action should be
an erc fine procedures for weighting the preferences and identities of
ante concept, that is, that a decision should make sense from
various people. For most decision theorists, however, interper-
the standpoint of the preferences, identities, and time values
of sonal comparisons of value seem to have create unresolvable
the decision maker at the time of the decision. Such a position,
problems. How is it possible to decide whether gains to one per-
however' has the disadvantage of ignorirg the obvious fact
that son, measured in terms of that person's values, are greater or
many decisions that arejudged "intelligent" ex ante will subse-
less than losses to another person, measured in terms of that
quently be assessed as "unintellige nt" when all of their
out- .
person's values? Modern students of choice are agreed that or-
comes and effects on values are finally realized.
dinary individuals and societies do, in fact, routinely make such
Alternatively, it is possible to argue that the intelligence of an
comparisons, but there is no agreement on a justification or
action is ultimately an ex post concept, to be assessed in terms
of procedure for doing so.
future values and experiences. In that case, decision intelli-
The ways in which the organization of attention provides an
gence cannot, in general, be determined at the time
decisions implicit solution to these problems was discussed earlier in the
230 ApRtMERoN DECrstoN MAKTNc
Decision Engineering 231
context of inconsistency in multipre actor decision
making. socialization of new organizational members and moderate turn*
Multiperson decision -fng procedures such as
exchange,
bargaining, coalitions, and poler achieve some over sustain needed variability in individual rules. Thus, the
of their succesg socialization and turnover rates that ate desirable from the
by being embedded in a syslem of lirnited
attention. If everyono
could attend to everything, and did so, the possibilities standpoint of the development of the organizational code are not
for irrec- necessarily desirable from the standpoint of every indiüdual in it.
oncilable conflict, that is, for the failure of decision
procedures
to achieve acceptable choices, would be greater
than'they are.
Incomplete attention to i¡rconsistencLs in pr"i"r""ó"s
and 6.7.3 Decision Making Myopia
identities is aided by mechanisms that concear
the oontradic.
1i9nl. organa;ations are divided into departments, andlabor is The dispersion of consequences over time and space is a fünda-
divided among specialists, thereby reduóing mental complication in defining decision intelligence. Actions
the likelihood th;;
inconsistencies cutting across divisional oi taken here and now have consequences here and now, but they
specialty iirres wilr
impinge on decision making. Inconsistencies *"--igrrorra also have consequences somewhere else and some time later.
through ignorance. In this *u¡i-it"d attention Since the distant consequences, in particular, are not easily pre-
sacrifices grob.
al consistency for locar consiri"n.y. since not dicted, tradeoffs across time and space invite some biases and
all irr.oori.t"ñJrs
are evoked, loosely coupled decision making prejudices.
systems operate
successfully in the face of potential confliciwien
those that
seek integration fail. The old folk wisdom ouT oF srGHT our oF MrND
is: Good t"n.", make
good neighbors.
The assessment of intelrigence across individuals Although moralists and decision theorists are sometimes con-
and groups cerned with the dangers of sacrificing immediate and close in-
I fuÍlgt complicated by fuir nested character. Decision mak-
ing indiüduals are parts of decision making groups. terests to those more distant, deploring excessive savings or
óecision misguided .parochialism, they generally see larger perversities
groups are parts of decision mat ingiiganLations.
TS"e
cisionmaking organizarions are parrs of dJcisián
oe- in biases favoring realizations that arc close in time and space.
eties. outcomes that are favorabti to in¿ividuars -;ki;; soci-., The immediacy and clarity of the present and the nearby tend
'
vorable to the groups, organizations, or societies -uy,roiu"-iu- system atically to disadv antage the spatially and temporally dis-
of-which they tant. The symbolic and substantive pleasures of the process of
are parts; and vice versa.
decision making and its immediate, ne arby consequences come
. For example, consider the problem of deciding on socializa- to dominate.
tion and turnover rates in anárganization. a"
orluniruiiorr ro- The bias toward effects that are clear and close undoubtedly
cializes its members,^ instructing them in
the Jrg*iruiional introduces a tendency for decision making procedures to be in-
code.. convergence of indiüduJbeüefs
and an ori*irutionrt
cgde generally useu both for indiüduals appropriately inattentive to important concerns that are fuzry
5 and for ín oiguniru-
tion. Indiüduals profit.from learning the rules and distant. Correcting the bias is, however, complicated by the
tion quickly. The organaationalcodJprofits from
or trre ;ü#;- factthat favoring the clear and the close is sometimes necessary
individ-ual ex_
periments with new beliefs that turn to survive. In that sense, 9t least, concerns about long-term'and
out to be usefur. However,
a serious threat to the effectiveness global intelligence must always be subordinated to valid con-
of the system is the possibili-
ty that indiüduals will adjust to,an organizational cerns about short-term and local intelligence.
code before
the code can learn from them. A comblnutio,.ái In any event, the story of bias against the distant is told about
t"lurl*üso* both indivieluals and social systems.'Individuals seem often to
232 APRIMERoN DECISIoN MAKING
Decision üngineerilry4 233

sacrifice their long-run interests for short-run pleasures. Brief


Three important modern variations on the theme of reckon-
gastronomic, alcoholic, pharmacological, symbolic, and sexual
pleasures seem irresistible at the moment despite their conspic-
ing are found in theories of divine judgment, economic ex-
ploitation, and environmental tampering. The basic idea of di-
uous longer-run costs. Individuals indulge i" practices when
young that have adverse consequences when- they are old, vine judgment is that any temporal, earthly rewards and
penalties for actions taken here and now have compensatory re-
When consequences are dispersed among individuals, the de-
wards and penalties in a heavenly eternity. The basic idea of
tions that produce pleasures for some often cause pain to oth-
economic exploitation is that spiritual or physical wealth in one
ers. I)ecision makers seem to find it systematically easier to
place is made possible only through compensating poverty else-
empathize with the feelings of close relatives, colleagues, corr-
where, that richness for individuals and for states results from
patriots, and currently visible strangers than with theleelings
of redistributing limited resources. The basic idea of environmen-
more distantpeople. In short, theré is a tendency to respond to
present, local concerns and to leave the resulting long-iun
tal tampering is that human interventions into the ecology of
and the earth in order to better the lives of contemporcry humans in
more distant problems for subsequent or more distani de;isi;;
some parts of the world inexorably reduce the quality of life for
makers. For example, the budget director who presided over
later and more distant humans. The common thread of all three
the largest expansion of the national defici; i; United Sor**
theories is the idea that every gain is a loss somewhere else.
history has been quoted as saying that the job of every subso-
quent administration "is to distribute and administer pa-in.', t The theme of progress is basically a theme of expanding re-
sources (or increasing efficiency in exploiting them). It finds
modern representations in theories of religious conversion, the-
APOSTLES OF RECKONING, APOSTLES OF PROGRESS ories of economic growth, and theories of scientific and techno-
logical development. The basic idea of religious conversion (ot
Making an assessment of decision intelligence involves a judg-
education) is that virtue can be expanded indefinitely. Each ad-
ment with respect to distant consequences of actions takenheie
and llow. In this regard, the deepest division among students
ditional person who is led to Tiuth increases the aggre gate
of amount of virtue in the world. The basic idea of economic
decision making intelligence is the division between those who
growth is that ailocations of disproportionate wealth to some
see decision making as affecting the distribution of symbols
and persons at one time increase the pool of resources over time
resources over time and space but not their total amount and
and thus ultimately the wealth of others. It is possible for every-
those who see decision making as affecting both the distribg-
tion and the total amount of symbols and r"io,rrces available. one to be made better off and for gains now to translate into
the possibility of even greater gains in the future. The basic idea
The former group might be called the apostles of reckoning,
Their beliefs are typified by a principle: The world's resources of scientific and technological development is that knowledge
ate fixed. Whatever gains are made at one time and in one and its practical applications increase over time, and those in-
place must be paid for at another time or in another place. creases lead to increases in aggre gate available symbolic and
Ulti- substantive resources. The common thread of all three theories
mately, there is conservation of symbolic and substantive assets.
The latter group might be called the apostles of progress. Their is the idea that resources are expandable, that gains for one
person can lead to gains for all others.
beliefs ate typified by an alternative prinriple: "itt" world,s re-
sources are expandable. By suitable inventions and exchanges
The prophets of progress are clearly less depressing than the
everyone can be made better off. Symbolic and substantive
prophets of reckoning. They offer hope that the whole universe
ár- indefinitely, rather than just a part of it briefly, can be made
sets can be augmented.
better. Although beliefs in the axioms of progress can be fcrund
234 A pRtMER oN DHCrsroN MAKTNc Decisktn hngineerlng 235
in many places and times, they seem to be disproportionately the decision environment. As a result, decision engineering is
ox-
pressed through the voices of the young, the Wbit, and tied closely to an understanditrg of fundamental issues in theo-
the rich.
The discouragements of age, life away-from the mainstream ries of adaptation.
of
organ izational and technólogical etáUoration, and persistent
poverty all make progress ttt* more fantastical than believable.
understanding the bases for divisions between the apostles 6.2. 7 Adaptive Inefficiencies
of reckoning and the apostles of progress is useful for monitor-
itg contempor ary debátes over religion, economic systems, sci. Traditionally, the search for adaptive decision intelligence has in-
ence and technology, and the environment, but a resolution volved trying to discover decision procedures that lead to unique
of and stable decisions that are (by some plausible criterion) best in
the debate still leaves the criteria for intelligence ill defined.
For whether one accepts a vision of reckoning or a vision of a particular environment. If such ptotrdures can be specified,
progress, one is still left with problems of intertemporal they may be appropriately described as "intelligent" or "effi-
and in- cient." Such claims are often made for rational decision proce-
terpersonal One is still left with probtems of the
9o*parisons. dures and for history-dependent rule followirg, for example.
unstable and endogenous nature of preferences and identities.
And one is still left with the necessiry of deciding whether deci- Assumptions of efficiency aÍe) however, suspect. Decision
sion making is to be seen as an atenafor developrng interpreta- processes may sometimes lead to decisions uniquely required
tions of life or an arenafor directirg life.
v r by a decision situation, but they often will not. In fact, modern
Faced with the intractability of these problems in defining understandings of adaptive processes identify several ways in
in- which decision processes result in outcomes that ate not implic-
telligence, decision engineering for the most paftignore$ them.
It asks how decision making intelligence can be improved even, it in environments or are not unique. Those inefficiencies in
in the face of uncertainty about exactly where inteiligence lies. matching the environment stem from some very general fea-
The implicit presumption is that some techniques of i*prove- tures of adaptation that ate shared by mary decision processes :
ment are robust vis-á-vis alternative conceptions of intelligence. !. Lags in matching.Adaptation takes time. Although it is
The implausibility of the presumption is cóunterbalarrrrd by possible to imagine that decision processes improve the
its
necessity. In such a spirit, the remainder of the chapter match between decisions and the environment, there is no
exarr*
ines three sets of problems involved in decision engineering. gvarantee that complete convergence will be achieved by
The first is the problem of improving adaptiveness, of making any particular time. If the environment is changirg, there
a
more consistent match between decision makers and their ,rriri- is not even any assurance that adaptation will be fast
The second is the problem of using knowledge, of enough to improve the match.
:Pl*.ents.
bringing what is known and knowable to bear on decisiorrJ. The 2. Multiple eqwilibria. Most theories of decision making aÍe
third is the problem of creating meani"S, J using decision theories of local adaptation. They assum e a process in
making as an occasion for expanding p.rréptions, constituting which considerations close in time and close in cognitive
selves, and eliciting human commitmént. or social distance domin ate those that are more distant.
Such decision making is essentially "hill-climbing,". re-
6.2 Improving Adaptativeness sponditrg to local feedback, and is subject to becoming
stranded at loc al (rather than global) maxima.
Decision processes strengthen their claim to intelligence 3. Path dependency. Decisions and outcomes in a particular
when
they improve the match between decisions and the demands environment depend not only on that environment but
of
236 A pRTMERON DrcrsroN MAKtNc
Decision Engineering 237
also on previous environments and the ways
in which they
have been e4perienced. The historical p"irt effects on subsequent history. Though the course of develop-
Á"L"s some ment is explicable in terms of a comprehensible process, its re-
outcomes unrearizable in the future, including
some previ.
ously realized. Reratively unlikely events, ií aliz,ed course is difficult to predict.
trr"y occur,
change the structure in permanent ways.
4. Networlcs of diffusíon óutcomes deplnd
on the ways in 6.2.2 Exploration and Exploitation
which information spreads. Information structuil,
i.áute
some decision makers and produce outcome;
;aittb""bi; A cen ttalconcern of ad,aptive intelligence within a path-depen-
to elements of isolation und tr"g#;'tiiJ-"i.", ,r* dent, meandering history is the reiation between the *pl;
outcomes of a decisio.n p-rocess r"nritiu" io pur ration of new possibilities and the exploitation of old certain-
^making
terns of connection in informati,on networks, to changes in ties. Exploration includes things captured by such terms as
information technology, and to the *iiir *rrirr, infor- search, vatiation, risk taking, experimentation, play, flexibility,
"ur"
mation is incorporated by decision makers. discoverY, and innovation. Exploitation includes such things ;;
5. Munat adaptaion Decision
-d";;;up, ,o their envi_
ronments at the same time as their eivironments
refinement, choice, production, efficietrcy, selection, impleárrr-
are tation, and execution. Explicit choices between expíoruiill urrd
adapting to them. They discover what p"tlr
ronment are e4proitabre, but the e4ploitability "r-iit"
*"i- exploitation are found in calculated decisions aboit alternative
of tttl investments and strategies. Implicit choices are buried in many
ronment changes as a result of their having "nui features of decision making rules and custoffis, for example, in
áiscovereo it.
They receive social approval or disapprouui
u, tfuil; procedures for accumulating and reducirg slack, in searrn t"f"t
of making particular
{ecisions, but the level and practices, in the ways in which targets are set and changed,
disapproval shifts as the "i6ñ;J;
number of other decision *"k*; and in incentive systems.
making similar decisions rt r"r-. In rational models of choice, the relation between explo ita-
outcomes"i;;;;;i
"rru"g"r.
adaptation are likely to lead to rtubl""rá that are tion and exploration is basic to theories of rational üarch.
predicted neither by the initial environrnent
nor uf u re- .j Choices must be made between gaining new information about
alternatives and thus improving future returns (which suggests
t.trij,lli; of adaptation. Decision makers are tied inro allocating patt of the investment to searching among uncertain
communities of other decision makers and
their decisions. alternatives), and using the information currently lvailable to
Their histories are intertwined by competition, improve present returns (which suggests concentrating the in-
tion, and other forms of interaction. History L"mót
"oop"lu-
seen as simply a product of one decision
u, vestment on the apparently best alternative). The ptoUtem is
makér and his or complicated by the possibilities that new invest*"rrt alterna-
her. own exogenous enüronment. Decisior,
their environments coevolve. -uL"^ uno tives may appear, that probability distributions may not be sta-
ble, or that they may depend on the choices made by others.
The idea that history is a loca'y adaptive, In theories of limited rationality, discussions olexploration
branching process
with multiple equilibriá is a centrál i"utur" and exploitation emphasize the role of targets in reguláting risk
of modern theories
of change. There are irreversible branctres, t ence taking-
Ttd allocations to search. The usual assumption is ttrat
p;rl']álp""_ risk taking and search are inhibited if the most preferred alter-
dence and decisive minor moments. rrr"
seem almost chancelike in their rosolution,
ti"r.rr-í"i"tJ"tr" native is above (but in the neighborhood of) th; target. On the
yet decisive in it
"i, other hand, risk taking and search are stimulated lf tne most
238 A PRTMER oN DECtsroN MAKTNc Decisktn Engineerin¿4 239

preferred known alternative is below the target. Because of the that returns from the two options vary not only with respect to
role of targets, discussions of risk taking and search in the limit- their present expected values but also with respect to their vari-
ed rationality tradition emphasize the significance of the adap- ability, their timing, and their distribution within and beyoncl
tive character of aspirations themselves. the organization. Processes for allocating resources between
In learning and selection models of rules, discussions of ex- them therefore embody intertempo ral, interinstitutional, and
ploration and exploitation are framed in terms of the twin pro- interpersonal comparisoils, as well as risk preferences.
cesses of variation and selection. Elimination of inferior foims, The difficulties involved in making such comparisons lead to
routines, or practices is essential to sunrival, but so also is the complications in speci$ring appropriate tradeoffs, and in achiev-
generation of new alternative practices, particularly in a chang- ing them. Defining an appropriate balance is made particularly
ing environment. Because of the links among environmental tui- difficult by the fact that the same issues occur at several levels
bulen ce, decision diversity, and competitive advan tage, the effi- of a nested system-at the individual level, the organizational
ciency of any decision process is sensitive to the relation between level, and the social system level. Achieving an appropriate level
the rate of exploratory vañation reflected by the practice and is rnade particul arly difficult by dynamics that tend to trap deci-
the rate of change in the environment. For exampfá it has been sion makers in cycles of accelerating exploration or exploitatioll.
argued that the persistence of garbage can decisión processes in On the one hand, exploration can become a trap. If failure
organizations is related to the diveriity advantage mry provide usually leads to exploration and exploration usually leads to
in a world of relatively unstable environments, wñ.tr páiá¿ with failure, a decision maker can be trapped in a cycle of explo*
the selective efficiency of convention al rationality. ration, trying one new thing after another without spending
long enough exploiting any innovation to secure the gains from
experience that ate necessary to make it fruitful. When decision
6.2.3 Finding a Balance
processes lead to a string of inadequately exploited experi-
Ordin ary experience teaches that social institutions often suffer ments, they are likely to be improved by interventions that in-
from a failure to exploit what is known. Enthusiasms for cre- hibit exploration.
ativity, new ideas, and change lead decision makers to be lax in On the other hand, exploitation can also becorne atrap. The
assuring that the plumbing works , that telephones are or- essence of exploitation is the refinement and extension of exist-
swered, that the logistics are well managed. Glorification of ex- ing competencies, technologies, and paradigms. Its returns ate
ploration obscures the fact that most new ideas are bad ones,- positive, proximate in both time and space, and predictable.
most changes are detrimental, and most original inventions are The essence of exploration is experimentation with new alter-
not worth the effort devoted to producing them. Decision sys* natives. Its returns ate uncertain, distant in both time and
tems that engage in exploration to the exclusion of exploitation space, and often negative. Strategies of exploitation that lead to
are likely to find that they suffer the costs of experimentation locally positive outcomes are likely to come to domin ate over
without gaining many of its benefits. They exhibit too many u1- exploratory strategies that are globally better but locally inferi-
.

developed new ideas and too little distinctive competence. Con- or. That is not an accident; it is a consequence of the temporcl
versely, systerns that engage in exploitation to the exclusion of and spatial proximity of their effects, as well as their precision.
exploration are likely to find themselves failing to discover and Those traps to exploitation stem not from stupidity but from
develop new capabilities and nelv opportunities. In a changing learning. Refinements and improvements in competence asso-
world, they are likely to become obsol,escent. ciated with improving standard paradigms, conventional knowl-
I-Jnderstanding the choices and improving the balance be- edge, and established methods provide local gains that are com-
tween exploration and exploitation are complicated by the fact pelling. As they develop greater and greater competcnce in
240 A pRrMIlR tlN DlrcrsroN MAKTNc ':,

Decision Engineering 241


using existing technologies, knowledge, routines, forms, or
Decision makers try to involve themselves in the social pro-
strategies, however, decision makers become less and less
will- cess of forming and transmitting knowledge in order to trans*
ing or able to change to newer ones that offer longer-run
supe- late that knowledge into effective action. From time to time
riority. They become better and better at an infeiior practice,
they make mistakes or are misled. They leave some things un-
In this sense , &t least, ad,aptation is self-destructive. When
known or incorrectly known. Knowledge is often elusivé, and
adaptive processes lead to greater and greater competence
at using it is often difficult, but ordinarily the benefits from knowl-
existing procedures, they uré Hkely to be improuro ui irri**e1-
edge make its pursuit worthwhile. Difficulties in gaining and
tions that protect or stimulate exploration.
using knowledge are mitigated by improving techniques for ex-
tracting knowledge from experience and from others.
6.3 Using Knowledge
6.3.1 Extracting Knowledge from Experience
Decision processes presume the exploitation of knowledge.
For
example, rationality involves anticipating the future conse- It is well known that individuals are not particularly good inter-
quences of present actions, as well as future preferences preters of evidence, and their limitationi underminelhe intelli-
for
those consequences when they occur. The ability to use gence of decision making. They mak e avariety of mistakes and
L"on¿-
edge to anticipate consequences and establish preferences simplifications. They learn lessons inadequately, recall memo-
for
them is essential. In a similar way, rule-based action demands ries incompletely or incorrectly, estim ate futures inaccurately.
coherent adaptation to the past. The adaptation of rules They learn superstitiously, assigning causal significance to ac-
and
procedures to experience involves forming inferences tions that are correlated with effects but not affected by them.
about the
world from the events o{ history. The aUiity to use knowledge They remember history in ways heavily dependent on their cur-
in making those inferettcét validly is essential. rent beliefs. They confound their wishes and hopes with their
Knowledge is a social construction. ft is developed and expectations. They ignore useful information out of concern
certi-
fied within social institutions. Those institutions help to that it may reflect deliberate falsification.
make
knowledge both valid and reliable, valid in the sense that
it por- Some features of the processes of human inference have
trays reality correctly and reliable in the sense that it is been reviewed in earlier chapters. The limitations on inference,
shared
and reproduced among knowledgeable people. There however, are not simply human frailties. They are also due to
are rules
for certi$ring knowledge, and those rules are maintained in structural features of the environments in which decisions are
so-,
cial institutions of research and education. The rules made. Decision makers face four principal structural problems
and tt .ir
integration into institutions make knowledge inaccessible in understanding either the future or the past: the paucity, re-
to
unilateral manipulation. It is not easy for in¿ividuals or small dundailcy, ambiguity, and strategic nature of information. Mis-
groups to dictate what is believed. On the other hand, takes of inference in the face of those problems do not simply
knowl-
edge changes' Often knowledge at one time becomes produce random error. They also produce systematic biases.
ignorance
and prejudice at another, as anyone who has read fiftyrear-old
textbooks in any field can testify. Those changes result PAUCITY OF INFORMATION
from
tensions between old knowledge and new chadánges
to it and Predicting the future or learning from the past requires infor-
ate orchestrated in such a way that knowledge raiely
wanders mation adequate to assess the causal structure of the environ-
very fat from broad socially acceptable und*tsturrdings.
ment. Decision makers try to understand their environments on
242 A pRtMHR oN DECrsroN MAKTNG
Decision Engineerin¡4 243
the basis of inadequate information. History is not generous
with observations. Sample sizes are small. Furth.r*oñ, differ. tion of smaller events, each of which can be experienced.
ent samples The process of elaboration is a matter not so much of
lrawn from historical experience are not easily making good statistical estimates of a few variables as it is
comparable. 'Pooling across events involves pooling relatively
dissimilar things.
of accumulating and integrating detail.
2. Multiple observers and multiple interpretations. Decision
For example, consider the case, discussed in Chapter L, of situ-
ations involving very low-probability events of very high colso.
makers can try to interpret experience in more ways.
quence" Examples might include important innovative discover.
Given the cost of increasing sample size, increasing the
number of interprctations of observations will often result
ies or large-scale nuclear disasters. Most decision makers will
never in greater net return than will increasing the number of
experience the occurrence of any particular very low-probability
obsenrations. Improvement comes from reducing mea-
event. Experience is likely to lead them to believe that itt. chancá
surement error rather than samplitrg error.
of that event is even smaller than it is. As a result, ordinary e{po.
3. Hypothetical histories. Decision makers can try to experi-
rience is likely to lead a decision maker to be pe*.6.ry pes,
ence more of the events that did not happen. Using rich
simistic in the case of low-probability positiv, (e.g. üreak descriptions of actual histories and near-histories to devel-
through innovative discoveries) and peryersely"urrrtsoptimistic in the
case of low-probability negative events (e.g. nucle ar accidents).
op models of the processes of history, they can simulate
alternative histories and estimate the likelihood not only
System atic sampling bias is also introduced by promoiing
successful decision makers to positions of power and influenrJ
of the realized history but also of other histories that
might have happened.
Successful decision makers are inclined tó attribute their
own
past successes to their own past actions and thereby tend Each method attempts to deal with what appears to be a
to ex-
aggerate their own capabilities for avoiding risk and producing small sample problem by squeezing more information out of a
successful outcomes in the future. By retaining and pto*otin[ single case. In effect, they are criticisms of standard (large-sam-
successful individuals and removing or ignoring individuals
who ple) methods, which tend to extract only a single bit or two of
have failed, decision making systems gener ate strong pressures information from a complicated event.
toward interpretations of the past that reinforce rotrn¿ence in ',
being able to overcome obstacles and to avoid pitfálls. Such
op- REDUNDANCY OF INFORMATION
timism can sometimes lead to self-confident behavior that is
self-confirming. But it can also lead to foolish adventures. Information about the world is produced by past actions in it.
Overcoming the paucity of information involves both trytng Past actions tend to be repeated, leading to observations that
to increase sample sizes, which is not always easy but is usually are concentrated in the same or similar situations. As has been
viewed as uncontroversial, and trying to extract more inforrlr¡* detailed in the earlier discussion of exploration and exploitation
tion from unique cases, which is often seen as more controver- and of risk takiilg, decision makers generally repeat successes.
sial. Three suggestions have been made for improving the When decision makers repeat successes, their experiences be-
infor-
mation content of history as it is found within it . small sampi** come increasingly redundant. They gain littte ."périence doing
of natural experience: other things. Decision reliability limits evidence to phenomena
1. Rich histories. Decision makers can try to experience
in the neighborhood of current practice.
his- Overcoming the redundancy of information involves increas*
tory more richly. Any particular historical ,u.nt is a collec-
ing the willingness to experiment. One way of increasing experi-
244 A PRTMHR oN DncrsroN MAKTNC
Decision Engineering 245

mentation is by adding noise to measures


of perforrnance. per- variables uncontrolled. The outcomes of history occur within an
formance feedback that is noisy leads
to arbiir aryfailures, thu' endogenously changing environment that complicates the mak-
to experimentation. The same kitr¿ of result
may be achieved by ing of valid inferences about it. Measuretnent procedures are
sustaining other elements of confusion
into the system-break. poorly specified and are subject to substantial subjectivity.
downs of social ization, memory, and
social control. When new Overcoming the ambiguity of information involves trying to
technologies or new organ izational forms
make retrieval easier improve the experimental design of experience. When deciiion
or harder, they disturb the balance between
remembering his. makers change things, they often make several small changes at
tory and forgetting it and change the rateof
experimentation. the same time, Such incrementalism has much to recommend
A second way to increase experimentation á*ong it, but it is not a very good strategy for gaining knowledge in a
successful
decision makers is to increase the rate
at which aspirations ad. noisy world. fncremental decision makers find it difficultlo dis-
just to success. The same performance
will be interpreted as ei. entangle the effects of multiple small changes in a complex en-
ther success or failure depending on a decision
maker,s aspira. vironment. The effects of one change confound the effects of
tion level. Raising aspirations increases
the failure rate and thug other changes, and the smallness of each change means any ef-
the experimentation rate.Investigations
of the effects of adap. fects it may have produced tend to be lost in the confusions of
tive aspirations indicate that u.ry rapid
or very slow adaptation history. Such problems suggest a strategy of decreasing the
of aspirations to past experiencb is Lsualty less
desirable than number and increasing the scale of changes that are made.
an intermediate tate,of adaptation and
that the rate of aspira. They also suggest a strategy of slowing the rate of adaptation
tion adjustment has long-run i*plications not
only for an indi. to experience. In a world where there is both considerable noise
vidual decision maker but also f* populutio"r
of ,h.*. in observations and simultaneous adaptation by others, fast
As was obsenred in the discussion ór tirt
taking in Chapter L, learners tend to track noisy signals too closely and to confuse
however' the relation between success (o,
failure) and experi. themselves by making changes before the effects of previous ac-
mentation is more complicated than the
simple ..failure pro- tiOns are clear. Those disadvantages of fast adjustments are
duces experimentation'i hyporhesis. For
failure may have deadening effects rather
ou*;1";;;;istent often overlooked in the face of enthusiasms for fast learning as
than stim,rlating a tool of intelligetlce. The translation of premature inferences
ones; interpretations of outcomes may ,

persistently exaggeratá into action is likely not only to lead to decision mistakes but
performance, thus reduce the failur
rít";
may be stimulated not only by failure " but and experimentation
also by the personal
also to confuse subsequent inferences. Patien ce, in combination
with decisiveness, is likely to yield gre ater gains than an alterna-
security and illusions of invulnerability
produr*á uy pirrirt*nt tive strategy that combines quick learning with incrementalism.
success' As a result, managing explorution
by contiottirrg pat-
terns of success and failure is more compticated
than it iright
appear at first blush.
6.3.2 Extracting Knowledge from Others

AMBIGUITY OF INFORMAIION Many of the capabilities of individuals are connected to their


abilities to profit from the experiences, knowledge, and prac-
Natural experience is difficult to interpret.
As history unfoldsn tices of others. Many of the iapabilities of orgu-trizations are
the drawing of inferences is seriously
.trrrpromised by the diffi- connected to their abilities to store knowledge in rules that are
culty of isolating causal effects from random
forces. The experimental designs of
or extraneous widely accessible. As a result, improving decision processes in*
hisrory leave ;;;;;-"anr volves understanding how individuals and organ izations appro-
246 A I'RIMEII oN DuctstoN MAKINc /'t

Dec'ision l;,rt¡¡ineerin¿¡ 247


priate the knowledge of others and what
the traps and limits of whether a decision maker is "overinvesting" in knowledge? Li-
appropriation may be.
braries have many books that are infrequently used; p*.urch
laboratories generate knowledge that has no impact; and man-
PROFITING FROM THE IDEAS OF OTHERS agement information systemt tutry files that ur. rurely tapped.
Imitation leads to shared practices and beriefs, Does society "overinvest" in library books? In research? In in-
but the intelri. formation? Optimizingon investments in knowledge is particu-
gen:e of the progell depends not only on
the rffi-uno .o,o. latly troublesome becá.ttt the costs and benefits of ntó*t.Jg.
prehensiveness of diffusion but also
on the "qíuity; lf an, in. ate distributed quite differently over both time and rfur". Hulu-
formation or.practice being diffused. If bad
iC"u, ,ir"ái easily, ing knowledge when it is needed often requires an investment
knowledge will be shared, but the sharing
*il1""t r"'"d-io inrdti in knowledge that is not known to be needed at the time it is ac-
gence. If imitation devalues the ideas;,
;;;;r;;;;ling dif.
the sharing of knowredge w'l read to decreases quired. The returns from knowledge may occur in a
fart of the
1::d,
macy rather than increases.
i' regiti. system quite different from the partwhere the costs arepaid.
A basic advantage gf knowredge is that Knowledge-based abilities to imitate provide both a riegative
bad ideas spread incentive for exploration (because of the difficulty of rupirrring
mo.re ragidry among the ignorant- than
among the informed, the gains from one's own discoveries), and a simultaneous posi-
and good ideas sprea{ moie rapidry among
the ignoranr. In particurar, go'oo io"ur -Jffi
thJ informed than tive incentive (because of the resultirg ease of capturing the
knowledge necessary to theiicomprehension
ñ;."d r ,n. gain's from the discoveries of others).' When the positive ef-
or uiiiiiation is fects of spillover exceed the negative, decision makers should
not widely shared. This creater .o-ptri.utions
for o"rirr invest in the development of irrelevant fundamental knowl-
ersseeking to use good new ideas glnerated -ur..
ersewhere and for edge. In this case, the investment in fundamental knowledge is
societies seeking to ensure the deielopment,
storage, un¿ re. not directed primarily toward making discoveries or inventing
trieval of knowredge about possibre technorogi;i;;?;iicy
ar- new policies but to developing the knowledge base required for
telnaiivrs in the absence of a current demanifor
themi'- profitittg from policies and discoveries made by others.
Individuals, business fi rms, schoor s, *J
are able to make effective use of new ideas
ñbli. ú',rüuu.ru.i* There are two relatively stable equilibria in this story of
that are coisistenü knowledge inventories and absorptive capaclt¡/. In one equilib-
with their existing competencies and technorogies.
makers have not invested in knowledg"
If áJ;ffi; rium, everyone devotes substantial resources to exploring for
not able to evaluate, adopt, and adapt new
i"*n?orl"o ,*r-", *o new ideas. Most of the returns of exploration for each inOivi¿-
ideas. fuónng n"r. ual, however, do not come from that individualis ¿iscoveries-
knowledge, they may adopt irr" ro.- but
".rrury not the sub- which are unlikely to be significant enough to warrant the in-
stance of new concepts (most reóently things rit" , q""rity
cles," 'Just-in-time,, inventory and deírvery
.ir- vestment. The returns come from building capacity to absorb
systems, and ,.total
quaJit¡ management',). To iacilitate the spiead oi i¿*, the occasional signifi cant discovery by oth.tr. Those returns
anO from knowledge and absorptive capacity sustain the exploration
tlteil t$etr^availability, individuars and oiganization,
á"u.rop that occasionally produces the discoveries required. In-this situ-
stockpiles of knowledge. organizations,
r"r iJJest in ation no individual decision maker has an inclntive to decrease
anarysis,"."--pi.,
knowledge acquisitionlrtrgysr, poriry
riur*i"r, exploration' nor cana new entrant gain by forgoing exploration,
e4pert systems, and file cabinets of contingeoi "-perts,
u.tion piunrl That is true even though the direct returns of, on.ls own explo-
Determining an optimal lever of investment
in knowredge in- ration are less than the costs. The equilibrium is sustain*á by
ventories is not easy. How can it be determined,
for exairple, the advantages of imitation and the inability to scpar¿rtc thc
248 A pRTMER oN DEctsroN MAKtNc /+

Decision Engineering 249


processes that build
fn9wt9lge and absorptive capacity from
the processes that underlie disóovery. behave in a way that makes them legitim ate decision makers.
In the second stable equilibriuá no one-invests in e'plo. Occupying similar positions in a nehvork makes imitation more
ration. since the returns óf e4ploration are p.i*"riü,eturns likely, and imitation across status differences is more likely to
from absorbing ideas generated'elsewhere, occur from high status to low status than in the other direction,
those returns are in.
significant if no one else.is engaging in The spreading of rules follows a natural dynamic. In the most
e4proration. Ár-tong ae
no oRe else is in e4ploiati,on, thlre i, ir;J"q"ate in. common situation, increases in the number of adopters of a
centive for any "".gug.r$
indiüdual párticipant_or potenti"l ;;;;: rule increases the likelihood of adoption by any partióular deci-
trant-to do so. sion maker who is not currently using the rule. The more uni-
Thlknowledge inventory and absorptive capacity versal the rule, the greater the value in using it. In such a situa-
story sug.
q:rr tlut the problems.of intetigent irniration are i"i"rlyiñ. tion, each current user of a rule gains legitim acy from each
dividual problems but also systerñ problems. other user. This gain provides a strong incántive to encourage
a"y ,yr1"* of de.
cision makers is at a serious disadvantage imitation, and users seek to convert nonusers. In some situa-
if it i, tiupp"o ut an
equilibrium with low investment in accumulating tions, however, the relation between legitim acy and the spread
knowledge.
The advantage of the high-investment equilibrium of rules is more complicated. Secrets, for example, have the
comes in the
first instance from the increased ease of appropriating properfy that they lose value as they gain converts. Fads have
and
using good new ideas effectively and, the propefiy that, up to a point, an increase in users increases
as a Uypr'oO,rót, th" g"nrr-
ation of occasional good ideas. rr tn" high-investment the legitim acy of each, but beyond that point additional users
alterna.
tive is the better o1g, ir.gannot easily be-achieveJ decrease the appeal.
tlrtugr, uu-
tonomous action of inte[igent indiviáual
actors ,turii"g]-* .
low-investment equilibrium, MATCHING SITUATIONS TO REPERTOIRE INVENTORIES

PROFITING FROM THE RULES OF OTIIERS When decision making is organized by a logic of consequence,
problems and alternatives are ordinarily taken as given, and the
students of decision making have studied information problem is primarily a matter of establishing the
the spread of prac. .r

tices, forms, and nonns thróugh populations likelihood of various possible consequences conditional on
of decision mak-
ers. They emphasize the imporiance to uny-giu"" choice of one of the alternatives. That is the conception under-
áeririon
Tuk"l of acting in a way that-is appropriat;.'Th"; ;;;r;;; lying the design of many decision support systems. It empha-
that decision makers who follow appropriate sizes information about the future in order to allow choice in
rules gain
m1? as decision mafers, and that tégtimacy racilitatEs legiti- the present.
s,rrvivar.
The spread of rules, forms, and ir;;;;tiniu, Such uses of information are certainly commor, but they are
p.oseq by edict d gou."T-"1r"r
f,ractic;. i*- probably less characteristic of decision makers than a som.*ttut
"g"ri.i*;#;
sional ol tr.age associations. It soñetimes - or*.r-
resurts from contacts different mode. Rather than look for information about conse-
among decision makels, This copying is quences to resolve a choice among alternatives, decision mak-
ra"ilitut; uy iüru.-
ment in a network and influen""a u¡iporition ers scan their environments for surprises and solutions. They
in thaíne¡vork.
TTlo decision makerswho *" ir"ltn* often do not reco gnize a "problem" until they have a solution.
decision makers use those networks "*ÉáAü;,#"rt,
not only to improve infor- They operate in a recognition/appropriateness mode. They
mation flows and channers of influence,
but árro to iiu* tá* ro look for a match between a situation and practices, preferences,
and identities they have in their repgrtoire library.
250 A pRTMER oN DnctstoN MAKTNG
Decisk¡n Engineering 25 I
In such a mode, identities and response repertoires
are ordi- than to scanning and interpreting ambiguous environments;
narily taken as given, and the infor¡natio'probl.*is
one of recognizing a situation. A rure-folr"i¡"g irimarily tied less to evaluating a specified set of alternatives than to a
á""isín -ut * wide spectrum of knowledge relevant to actions impossible to
is more interested in timely information
about the state of the anticipate precisely; less likely to illuminate the consequences
world¡1s it unrords ,r'u" i"
ñtouuui-rü,""Ji"u;J o'JJrro" ,". of known alternatives for known goals than to recognize situa-
jure. The presumption is tirat appropriate action"i" wili ¡" r.norn
immediatel¡ once rhe siruarion is ráu"at"a. tions and pafu them witn- established routines.
Á, ,"¡;ñ,;;t
toring supporr systems emphasize earty w-arniG^;;;- "
portunities and dangers over estimates
ór ¿irtuot"posriuititi"r,"p. ledge
A. mgnitoring *ylt", ;;. ,h" environmeni
events that require a change in performu".";;;;;;lril;;;
fo, current
Intelligence is inconceivable without knowledge, but knowledge
on networks of informers, gossip, disorderly
quick, clear signals of surpñs"r. Á .onr"qu"n."
iliil;rt";;;; also poses problems for decision intelligence. In particular, itls
urr"rr_.nt in. worth noting that substitutes for knowledge can interfere with
formation system tries to irganizesystematic aa

improving knowledge , that knowledge does not always yield ad-


information rele-
vant to understanding possibre fútúres. vantage in a competition for primacy, and that the potiti cal na-
n rrtri"",
ür, ""';;;itrir:
orderly information, und deriberatio".-n"ri'
tion systems are useful to decision intelligen.";il;""rü;;;:
;; tr;**.. ture of knowledge gives advantages to some decision makers at
the expense of others.
signed to do the first will not do avery
good job on thl ,iffi;
and vice versa. since information-monitoring KNOWLEDGE SUBSTITUTES
systems tend to
receive less attention than information-ana$si-s
ryrt"-. t";h.;: ,,
As was noted in the discussion of competency traps, compe*
IiT:l de.cision, ,r," pu"opü;
communication,
f;;i informatioo ,"r.i"#u"n'o tence at one technology or procedure can substitute for, and
as well as the technology of iniormurt*;;lt;
to be oriented to the latter. therefore inoculate against gaining competence at and using,
An information stratery of monitoring for surprises another, possibly superior, one. In the same w?y, when other
implies available capabilities can substitute for knowledge, knowledge
that the response repertoiies of a decisiú
to deal wirh whatever situation., -uk"."ur" udü;;;; is less likely to be pursued. Knowledge substitutes make knowl-
U" ¿isc"rneá. fñ;i;i;;;;
Tuy
implies a second aspecr to decision-informatülir¿" edge less valuable, at least in the short run. People who can see
lowing decision *i"_for_ " are, by virtue of that competence, less likely to learn sign lan-
-ui"r, want to ,*p"rJ;;;u,, .;#;t*- guage. T\Mo conspicuous forms of knowledge inoculation are
tions, they must continuafly rpdatá
tneii capaoiliai", r"."r.- found in technique and power.
sponse, gathering information and
training in;;;;#;;
that may be needed in rhe future. Tho."
iíventoriel oi pro- Substituting Technique fo, Creativity. Much of knowledge is Ítr*
ttuT:j-oly a view of information trrails ;;;";ft;inlt or u
goodlibrary rhan thar of a system poinred ranged hierarchically, proceeding from basic principles or para-
cific decisions.
,;;;;;lñ;;;": digms through intermediate propositions to immediate ob-
As a resurt, ideas about.the engineering servables. As decision makers seek to increase knowledgb, they
of decision making deal simultaneously with several levels of their knowing. For ex-
sfiou]a includ e "e arry warning sys"temr
" u"i rtG;iJiuritrru,. ample, students learn how to function within a given paradigm
timely response and effectiu" i-itution.
They prouuuty nero to and simultaneously to learn what paradigm to use. The latter
be oriented less to anticipating uncertain
ruió" .onrlqurn.., learning is sometimes called higher-order learning, double-loop
252 A pRTMERoN DccrsroN MAKTNG
Decisit¡n Engineering 253
learning, insight, creativity, or consciousness
raising. The for. sponding to it. Many adaptive strategies involve monitoring the
mer might be called gaining competence
or techniqui. environmento understanding its causal structure, storing iñfer-
New higher-order-knowledge is partic.'tarty
- vaiuable but is ences drawn from that understandirg, and retrieving thé impli-
also a particularry difficult kiná of
knowledge"," g"i, rr"m ei. cations of those inferences at appropriate times &fld ptur*s. De-
ther education or personar e¡perience.
Norriar prJrrrr", of ed. cision makers who engage in such activities have abilities that
ucation tend to strengthen uoiur"rr"r,
principles and paradigms and to protect
*.iJü".r"pr"¿ Uuri, are augmented by their experience in dealing with an uncertain
"f against alternative
frames' A well-educatéd physicist L, g"ogrupher but possibly comprehensible environment. The skills are learn-
knows welr what is accepted-as knowledge
is one who able and vital to intelligent action.
within those fierds. Not all decision makers develop talents in such activities,
similarly, e4perience gldn*ily protects
higher-order beliefs. however. Some decision makers are buffered from attention to
As e4perience is accumulateO, imirov"rn"ntr
¿n a schema confuse
i' techruquewith, the environment by a combination of pig-heaJe¿rr"r,
among schemata. In the ";;i;.t.
".-p;rtJ;about how to function
course of augmenting knowledge They believe they know what is right uid pursue their course
a set of principles or frames, deiision
within without attention to signals from the environment. In rate
makers ;;;;;e their
willingness and ability to consider alternative;;;;áig;. cases, they turn out to be right and conquer the world. fn even
tater cases' their luck continues into their next revelation.
. In this way, the process of becoming more competent is serf-
destructive' Michael poranf, commenting Other decision makers are protected from the environment
on one of his contri- by their power: Domineering parents with children; domineer-
butions to physics, observeá: "I would
,r"*, have conceived my ing bosses with subordinates; domineering paradigms with sci-
theory ler alone have made a great"n
r"rrd;t I had entists; domineering firms with other firms in their industries;
been more fam'iar_with majoi developme"tr "il il;;yris that domineering countries in the world order. The powerful do noí
were taking place. Mgreover, my initial
ig"orunr" oi;-ü;;;;: attend to the environment, because their power makes them
f-ul, false objections that were raised
ugui-nrt -v i¿"", prJtected
those ideas from being nipped in the immune to its threats. They become the environment, fotrirg
bud.,,3
For most decision others to adapt to them. They develop skills at unilateral lead]
-ukétr, therefore, the probrem of multi- .,
ership, &tcommand. It is a familiar story in interpersonal life, in
tnowing is a probrem of e4ptoring alternative
fgvel frames in science, in markets, and in world politics.
the face of the well-established urr¿ *"nr"
cognizecr[rt rrique A conspicuous problem with such power, however, is that it is
they have at using an existing one. In practíce,because
nique is an inoculation againlst a paradigm tech- poor preparation for loss of dominance. Children grow up; sub-
shift, rather little ordinates revolt; new paradigms arise; new firms challeng¿ mar-
higher-order learning is sécured trom
experience or from or- ket position; national preeminence declines. Decision makers
derly searches for knowledge. FundÁentar
assumptions come who have little experience with obsen¡irg, understandirg, and
from paradigm peddle* uná paradig;
pofitics, ,o.áipro."*.,
of persuasion and reinterpreiation iiat reacting to changes in the environment loie the capabitiijiio Oo
are part of the trade of
teachers, preachers, writers, and consultant..
-il; so. Their power to impos e aworld undermines knowirg how to
process' as has often^been -gurus, cope with a world that cannot be unilaterally controlled. As
observed, invorves fitful transitions
from one stable set of beliefs to anotúer. their skills at imposing environments grow, their skills at adapt-
ing to an environment atrophy. Subsequent losses of power tum-
Subs:jtutinF Power f,o1 fearytns.,Adaptive ble decision makers into worlds requiring skills that itr.y lack.
decision making is There is no magic solution to the inoculations against knowl-
predicated upon skill at underslanoing
trre e;;;;;#;;,._ edge provided by technique and poúer. Efforts tg strengthen
254 ApRrMERoNDpcrsroN MAKING
Decision Engineering 255
performance through improving technique
or power interfere swer is complicated, for it depends on a more careful specifica-
with efforts to redefne broad prLcipres. It
is a variation on ong tion of the kind of knowledge involved and its precis. *if*cts on
of the'mosr ramiliar features or iir"
dilemma-the cornperencSr rrap. It is quite
"aplo.utiánÁaproitation the right-hand tail of the performance distribution. But knowl-
nique or power within an'estaühsrreo system
pÁ;Li;?r tech. edge that simultaneously increases average performance and its
r.""*r.oge to reliability is not a gvarantee of competitive advantage. As a re-
become great enough to preempt higher_ord", "i t r"wl"¿ge or
capabilities sult, the role of the pursuit of knowledge in achieviig and sus-
taining competitive prim acy is mixed.
DILEMMAS OF COMPETITIVE DISADVANTAGE
THE POLITICS OF KNOWLEDGE t

societies, institutions, and individuals compete


with one anoth- Intellige,nt decision making requires coping with the political
er, using their knowl^qdge in purr"tot
competitive aJ.rurrtage. uses and consequences of knowledge. On the one hand, the use
The c-o¡sequences of .hung"r in knowledg;-;
iir" Juli'or onu of klglwledge makes decision makérs dependent on knowledge
competitor depend on changes in the knñedge
in ways thar maylgt be immediately obvious."i,
h;ffi;;"; providers, therefore vulnerable to manipulation. On the othér
was observed in Chapter 1, knowleáge
p*iJ"Ur, ,, hand, knowledge itself is not neutral politically. Knowledge
has a mixed effect on
competitive advantagl. r"áu"o; fl;;;ffi;il;,lil;
therebv reduces variabitity, un¿ tJJ"rt;;,
frames approaches to problems and shapes decision possibiíi-
ties itt wáys that favor io-, alternatives and interests ;ñ;;-
costly in a competition f";;rir"".)
-' ;"uuiuüffi* u, advantage others.
Suppose rearized performance on a particular
occasion is a Strategic l{ature of Knowledge. Decision-related knowledge is
draw from a probabilitv distribution of p";i;;";:"ü"
characterize rhis distribution uy some
.un r_arely innocent. It is tainted by the way it is generated uná Uy
-";r*. ;i;;;;";;p.*
formance, which might be caileá "abrlity;' the context in which it is presented. Most knowledge is subject
and some measure of to strategic misrepresentation. As a result, systemr of informa-
variabilirv, which *iglrt u" .áil"á--"inreliabilü;¿;t";"i"
knowledge are typically associatedr"ith tion and knowledge ane instruments of po*ót that favor those
improürrg uu"rug;i# i who can control'them at the expense of those who cannot.
formance_and reliability, thus with increasing
the"*eun u"o¿ ¿r- Those problems become paiticularly severe in modern con-
creasing the variance of rhe performan""
crease in reliabiliry restricrs ih"
oírttiuuil;;:il;- texts where specialization in knowledge and short lead times on
gained from knowledge, particuúy
.o.p"dñ;;;;;;"^;; ;; decisions cause decision makers to máke spot contracts for the
;ñ;;; ;;ffi
petitors is targe- Mo1é próciser¡ ff "r"; r"*_
í';;;;;'"iüüuóia,,r,- provision of decision-relevant knowledgá. As organizations
move toward knowledge and competence inventories that are
lffect :f
r"d:"tlg the right-hánd tail of the diitribí,ion, in-
creased knqwledge may easily decrease represented by markets in contract services, rather than a col-
the chance ¿iüü uru
-uulrug, lection of in-house skills, the politics of knowledge acquisition
even though it increases
::::9-11Yulsome
pertormance. :o-p:,itors,
highry variabre competitor i, ul-ort certain
and utilization becomes particularly germane to intellig"ttt de-
to be best. cision engineering.
The question is whether one can do exceptionally , Procedures for overcoming the strategic nature of knowledge
well, as op- ate key elernents of classical politi cal theory treatments of the
posed to better than average, without
reaving trr" .onñne, or role of experts and of modern economic theories of informa-
conventional action and soeialry certified
knowredge. ct" on- tion. The latter have been considered briefly in section 3.3.3.
256 A pRtMER oN DECtsloN MAKTNc Decision Engineerin¡¡ 257
DeQision makers are sensitive to their sions, thus avoiding the updating of basic beliefs with conterl-
wlnerability to the lack
of innocence in knowledge and wary of relying porary information. Because the information used in decision
on'strategically
manipulable evidence. Everyone knows thai making is strate Eic, it is treated as unreliable and less impor tant
alales representa.
tive has incentives. for exaggerating a productrs for decision making. Decoupling strategies are often cháracter.
good'qualitiee,
Everyone knows a mánager Jr worker has incentives fot istic of socially peripheral individuals or groups. They feel suffi-
trnderestimating how-that
long a ¡óu witt rake or rró* *.i ir will ciently estranged from prime sources of information that they
cost when seeking to have a job approved, reject all external clairns to knowledge.
and rrt ot"r"rti*ui.
how long it will take or hów
ing -""rr it will cost when negotiat.
ing a budget for ajob once approved. The l{onneutrality of l{nowledge. Politi calproblems with knowl-
Beca'se everyone knows suih ttrings, everyone edge extend beyond the conscious manipulation of knowledge
makes adjust-
in evaluaring t_h" eüdence. e""í u".áur" ;6;;'"rr; by knowledge providers. Knowledge itietf plays politi cal fa_
lents that everyone
Knows knows such things, sales representatives, vorites. That is well known with respect to tecirnólogical unO
:iTq:* and.workers are involved iia poten,t"llt;4pt;;i"J scientific knowledge. The comp uterlinformation revolution, for
escaratron of biases and counter-biases. examPle, has considerably disadv antaged public policies of indi-
consider, ror eámple,
tfe langulse of praise i" ;ü;il-iiiurttv, *rr"r" ii" io"aning vidual privacy. The development of nuciear weapons of mass
of unstinting enthusiasm.is ambiguous, ft destruction changed the politi cal position not onty of nations
than
"magnificent" or "genius" is treatEd (correctry) ";yú;;less
u, uo inruit.-'t*' but also of groups within nationr. Th, nonneutrality of knowl-
. The pathologies are familiar ro unyor" *ír;
the uses of information and knowredge where pr"r"i"rr"".
il.;;ri.áoeo edge is also well known with respect to politi cal and social
knowledge. Research on individual voting behavior has trans-
identiti es are- inconsirte"t. ráceáliü. rrr" possibility
o, ,

' formed not only politi cal campaigning but


cally manipulated information, decision makers "t
r,r*e'i also the whole struc-
,ry-* Lnr.r* ture of democratic responsiveness, to the detriment of citizens
that the preferences and identities of knowledg" who favor a Burkean perspective on representation.
consistent with their own ft_"y r""t ,o ñ;td;;;;
of knowtedge proüders or their
assure ;#irr;íái-ff The political biases of knowledge can be illustrated by look-
Th"'r""ir""y ing at American economics and sociology. The biases ol inter-
";;riü
$owledge proüders is a goal of many ;;ó;;, "r est here do not stem from the factthat American economists, as
sionalism in which the maik of a professional "ipr.írrr- a group, and American sociologists, os a group, exhibit quite
ir r""n á, ír" ,u-
pabilitv to render advice uncontarninated
bt;;;;l¡áut its different politi cal pafty preferences. The biases corne from the
*:::l ::nse
quences. Desires .o u.rur" ir,-J nv"iiy
tron providers make decision makers rikery
.i i"ir-.- nature of economic and sociological domains of knowledge, the
way they are organized, the factors they consider, the visions of
to cLnsult rriends or
relatives, rather than comparably trained
;rh";;;;il;; ;; humankind they assume, the questions they formulate, and the
make tho¡e wfo pr9üd" üo*r"ág"-ffiü';ñffi;; Lr rr- way they frame answers to them.
sourc:.s
\oyu],{ of course, has ihe pot"ntiufb-Offirtioouf Economics imagines the world as a system of voluntary ex-
complication that ', it may shield decision iffi;;;;; change among actors who are endowed with wants and re-
information that might t tt
-*éÁ - sources and who seek to satisfy their own wants by trading their
to ,""ogrrir" -iri"t"r.
"- try to de.oupt" it
decision"tpmakers own resources for the resources of others. It is a conception
--^tr:-u,ivel¡
rrom cependence on knowledge providers. "mselves that favors decisions oriented to improving the conditions of ex-
In the course of de-
coupling thems elves from sp ec-inó mo*ié
er, they tend ro decouple tñelr decisions
ág" ;;;"á.* il*.u- change, assuring that no mutually satisfactory trades are over-
looked. It tends to fteat the initial conditions of exchange
fróm nno*lrió. On,
response is to use "intuition" or "prior (wants and resources) as given. As a body of knowledgc, cco-
estimates,'to mafe deci-
258 A PRIMER ON DECISION MAKINC
Decisktn Engineerin$ 259
nomics is an any of decision makers who prefer
to take individ- meanings are often interpretations of fate and nature, but they
ual wants, identities, and resources as inviolate.
Decision mak- are human constructions, and decision processes are one of the
ers who are conrerned about the initial
distribution of wantg sites within which the constructions take place.
and resources in a system of exchange
body of knowledge tárr n"tpi.ri.-
wilr finJ
".o*rii., u, ,
on the other 6.4. 1 Expanding Awareness
of^Sociology,
norms, berongings, and lul-d, imagines the world as a system
identitiá that sustairi lñJ-r"lu* Decision processes use language and symbols to create and
through sociatizitio; ;,J;;;ffi;,;"jiHJ: ,..,TJ a con- communlcate meaning. They are threatened by two dangers in
;;;;áC; il ffi:;
ception that favors decisions ori"rri"á
belongings, and identities. It ,"nJ. iá the use of language. The first is that explicit meaning will be lost,
treat wants and resources that what one person intends to say will not be understood by
as subject to sociar contror. As a body"rt""J"lüil;ilgy;; another, that specific knowledge will be degraded and manifesi
an
fly. of decision makers who prefet ,o .rr"ig" Iiii'iou"r
wants,,identities, and resources. Dócision feelings obscured. Concerns over this danger domin &¡e those
makers irrr.*" ,""- theories of cornmunication in decision making that draw their
cerned with the problems of arranging
among individuals i1 a sociery will finá
efficien,-;#;;;; inspiration from the transmission of electiical signals. In
knowledge less helpful.
óciorogy ;r;;"dy;¡ metaphors taken from such theories, optimal communication is
The differences are not the standard ones portrayed as the precise reproduction by a receiver of a se-
of specialization. It quence of overt meanings coded and transmitted by the sender"
is not so much rhar econo-irr-;;;;ociology
look at different The second danger is that impticit meaning will be lost, that
lH::,i:^.,.Ii:they
lh:y
rook at tr,",u'"" ;tó;írr;;ü. ii,"
have of looking at the world transraté
oir- what one person says without being entirel¡r conscious of saying
T1"n'yays into deci_ it will not be he atd, that specific knowledge will not be entaigeá
slon advantages and- disadvantages for
others. In a similar;uy, and interpreted, and that feelings will not be elaborated. Th.
any body of knowledge fits tne n"ee¿s
not fir orhers. Medical knowledge otsu"t"J
orr"'n" i**.o ¿o* quest for order excludes the ambiguous features of life and
"r?
;;;;;l-rrlu,ing those human reactions to them that are at the heart of imagina-
sick people is differenr in its decifiá,
knowledge organized around keeping
ññ-,lo;;;;; -"¿i.ur tion and artistic comprehension. Sometimes expanding under-
hlalthy p"opi"-fro* Ur- standintg involves taking things that seem clear and making
coming sick. Hisroricar knowle cge orlun¿"J
state leads to different decisio?
;;;;ri""
árl r"'J them ambiguous.
;ou*#;;;;;;Jorical Concerns over this second danger domin atethose theories of
knowledge organized around internationai;;;;:;*; communication in decision making that draw their inspiration
organized around
$o¡v.ledse from poetry and the arts. The ernphasis is on creating a;d inter,
decision implicarions flom legal "o"."piio"r
knowl"Og""ir"*;'lr"?¿in"r.nt
oi;un;;*uroun¿ preting evocative ambiguiry. Evocative ambiguity usés language
conceptions of ,Justice.,. fórmer privil"!", f"iirir,"rr, ;1";;
T" vigilantes,
and judges; the ratter privileges
(or other media) to stimulate rneaning. nicn meani"gr are
rJ¿orrl"á-:-*i"r. drawn from the resonances and deep structure of languege,
meanings that ate not transparent to the writer or speaker but
6.4 Creating Meaning ate generated from the particular language chosen. Communi*
Decision-making presupposes meaning, cation, in these terms, is the art of crafting such evocativeness,
an understanding of In this spirit, T S. Eliot once commented on a critic's discussion
tfe way things are and might be, a basil for engaginf
discourse about whar is possible an¿
otür, in of his poem "The Love Song of J. Alfred Prufrock" by writing
wtrat t as náp!eñ"¿. ihorc that the analysis made by the critic "fvas an attempt to find out
260 A pRtMERoNDgcrsroN MAKTNG
Decision Engtneerlng 261
what the poem meant-whether that was what I had
meant it to The aspirations are) of course, too grand for the instruments
mean or not. And for that tr was grateful.,,a
at hand, and it would be foolish to imagine that the design and
Propounding a theory of artistic imagination may
be a bit be- use of decision making processes can result in solutions to such
yond the present book, but it shourd be noted that
artistic ancient problems. It may be helpful, however, to judge decision
Teaning draws on a spirit different from that of information making in terms of its evocativeness, its power to provide not
theory. It uses euo"atiue ambiguity to expand u*ur"r.rr.
just confirmation of familiar orders but also intimations of al-
Evocative ambiguity is far from nóir" o, arbitiary.v-uoi"
poet'creates rneaning without fully comprenendin!
rr* ternative orders, not just communication of what is known but a
the mean. transformation of what is knowable.
ing that has been created, but the words ire chosericarefully
to
elicit the imagination of ranguage. poetry and art encourage
the
simultaneous adoption of a üsión and the ,".ogoition 6.4.2 Constítuting the Self
;¡its un.
reality. They affirm life in the face of absurdityl
They u." Among the many aspects of meaning that ate shaped within de-
fortable with multiple, contradictory meanings and "o..
with the si. cision making processes, few ure more important than the un-
multaneous truth and farsity of beliefs. In a siri,ilar
way, áecision derstandings individuals have of their preferences and their
makers create ambiguity not to confuse but to
,ti-"1ái", identities. Values, goals, wants, and other conceptions of the
obscure meaning but to discover it. communications
ui""oiio
,"n- self emerge from the process of making decisions. Facilitating
structed to gain ac-crs to the imagination and
to knowledge intelligent transformations of preferences and identities is a
carried in words and visual stimuli. rn ttre world of
evocative de- prime task of decision engineering.
cision making, memoranda become forms of poetry;
plans be- Theories of limited rationality and rule following tend to
oome forms of sculpture; meetings become forms
ofiheater. take preferences and identities as given. I)ecision makers are
.In order to promote evocative ambiguity ro u' upp-fiiut,
place in its repertoire, decision enginJering ptouutit assumed to be endowed with wants and conceptions of selves
snourd and to act upon them, but the sources of those endowments ate
ground part of its character in thóries oitrirtory junirp*-
dence, language' culture, art, and criticism. Historiá
left unclear. If the origins of preferences and identities are con-
interpre- .. sidered at all, they are generally treated as buried in some inex-
tation involves the e4ploration of possible histories
un¿ o"áur- plicable cultural and sócial process of soc ialization and devel-
standing the efficiencies and inefficiencies of the
aecumulation opment that occurs primarily in childhood.
of history in tradition and belief. Theories of regar interpreta- -
tion consider the ways in which the meanings oflaws This sometimes leads to a curious difference in the ways the-
uoo't"gut ories of decision making treat adults and children. Adults are
concepts evolve through an interplay of séH-interested
pres- assumed to have well-defined preferences and identities. Their
sures and contemplative elaboration of meaning.
Theories of decisions are predicated on those givens but do not affect them.
language examine the ways in which communiátion
e4proits Children, on the other hand, are seen as developing prefer-
the structure of language to capture and impart
meaning that is ences and identities. Their decisions are seen as wayrl" which
fully comprehended neither uy ttre writer nor uy tte"reuoer.
Theories of culture e4plore thó ways cultural they develop (or engineer) themselves, moving from a condi-
o"*topr*nt ,"- tion in which preferences and identities are ambiguous and in-
flects adaptation to, and enactment'of, a changing
sñuotic en- consistent to a condition in which they aÍe clear and consistent.
üronment. Theories of art and criticism treat
engineering not as a passive or manipulative
i.o? íniorÁution The distinction made between adults and children is too
á.tiuity i" . deci- sharp. The preferences and identities of adults, like those of
sion scheme or a program of leárning, but as an
instrument of childror, can be molded within decision making processes. De-
appreciative interpretation.
cision makers can discover values, aspirations, ánd self-concep-
262 A pruMER oN DEcrsroN MAKTNc "r

Decision Engineering 263


tions in the process of making decisions and experiencing their
consequences. The aspirations that allow decision a belief in intuition strengthens the case for actions that are
makers to
distinguish successes from failures can themselves be trans. otherwise indefensible.
formed in the process of decision making. By obseryirrg 3. Tieat ltypocriry as a transition. Hypocrisy is an inconsistenry
their
own actions, decision rnakers can learn what they want between behavior and asserted preferences or identities. It in-
and who
they ate. When decision makers feel responsibie for their de- curs opprobrium both, because it reflects inconsistency and be-
tions or feel in a positive mood, they areparticularly inclined cause it appears to combine the pleasures of vice with the re-
to
see an outcome as successful and to define their preferencss wards of virtue. The onus on hypocrisy, however, inhibits
and identities in a way thatis consistent with it. foolishness. A decision maker with good-sounding talk may be
Suppose decision making is treated as a way of creating a person experimenting with being good in other ways. It may
pref-
erences and identities at the same time as preferences
and iden.
be more sensible to encouruge the experimentation than to
tities are tteated as a basis for decisions and their justification. condemn it.
In order to use decision making as a conscious basis for con- 4. Tieat memory as an enemy.Rules of consistency and coher-
structing the self- decision makers have to combine logics ent selves require memory. For most purposes good memories
of
consequence and appropriateness with a technology of make good cñoices. Memories accum"futl .*peiience and p.r-
fóo¡sh-
ness. They need to think about action now as Ueing
taken in
mit learning. But the ability to for Eet, or overlook , ffi&y also be
terms of a set of unknown future preferences or identities. useful. A decision maker who tuttñot recall what was dorc yes-
They
need ways to do things for which they currently have terday or does not know what others are doing today may act
no gooá
reason. In that sense , ?t least, they need ro*.iimes
to act be- within a system of consistent rationality or rule-following and
fore they think. nevertheless act foolishly.
The engineering problem is to find ways of helping decision 5. Tleat experience as a theory. Learning is based on a series of
makers experiment with doing things for which ifr.y have conclusions about history that people have invented to under-
no
good reason, to construct their conceptions of themselves. stand experience. fnterpretations of history, and thus experi-
Five
things have been suggested as possiblé: ence' can be changed retrospectively. By changing the interpre-
tation of history now, decision makers can revise what they
L. Treü the self as a hypothesis. Conventional thinking
about learned earlier and reconstruct self-conceptions.
decision making allows doubts about everything except
itte on*
thing about which there is often the greátest áoubt-ttre r.lf. Each of these suggestions represents a way to suspend tem-
Suppose decision making is defined ui a time for testing
ufr* porarily the operation of a system of reasoned consistency.
n-afive preferences and identities. Decision makers
can treatde- They are also potentially dangerous. They make the greatest
cision making less as a process of deduction or negotiatio" sense in situations, which occur in much decision áaking,
un¿
more as a process of gttttly upsetti"g where there has been an overlearning of the virtues of togics Ji
;;";;;r"pii"ns of what is
desirable or appropriate. consequence and appropriateness.
2. Treat intuiioi ot real. It is not clear what intuition is. per- A second requirement for a technology of foolishness is
haps it is simply an excuse for doing whar decisi";;;rrr'*unt some strategy for suspending imperativesJoward consistQncy.
to do when they cannot explain *hy they want to do ir. prthaps Knowing which of several foolish things might be done is not
it is some inexplicable way of consulting memories or ideas that enough. There is still the problem of actually doing it. The most
are inaccessible to standard theories ofthought. Whatever
it is, natural answer is that decision makers escape consistenry
through playfulness. Play allows action that is "unintelligeilt,"
t /)'

264 APRIMERoN DECISIoN MAKING


Decision Engineering 265
o'irrational,"
"out of chara ctef," or "foolish.,, Such action gx,
plores alternative ideas of possible purposes identitY, or determining the consequences of a decision and at-
and alternative vi. tributing consequences to actions, is a complicated problem in
sions of identities while retáining a basic commitment
to the il', historical analysis. As a result, the commitment ,rér.rsary to
cessity of order and seriousness. In ord", toi
irrdividual decision make and implement decisions is undermined. Not only is it
makers to function effectively, orga niulations
and societies need hard to make intelligent decisions, it is hard to say that thínking
to maintain both playfulness and óonsistency u,
urpects of intel. about the making of decisions is a sensible *áy of thinkin[
ligence. Org anizaiions can be plffieven
when the participantc about life. The ideology of choice conflicts with the realities oT
in them are not. organ izationi piuy is encouraged
by ,",,,fi** knowledge.
relief from control, coordination, and comm,rrr-irutión.
It is not necessary to embrace the more extreme forms of this
view to recagnize that knowledge can be inimical to making de-
6.4.3 Eliciting Commitment cisions. Knowledge seems to increase questions at afaster tate
than it increases answers. It provides too many qualifications,
Decisiol making ngtmally presumes commitment, recognizes too much complexity. One American President said
the willing-
of degision makers both to devote time and r"rrg1, that he wished he could find a one-handed adviser. Most of his
it::
ciding uttg jo accept responsibilify for the uncer grií
to de.
corso. advisers would tell him one thing and then say "but on the other
quences of their actions. Designing effective hand" and add several contradictory obseryations. The more
decision processes
involves creating a motivatioñal basis for such knowledge they had, the more complicated the world seemed
a cornmitment
that is consistent with the human spirit and does to thern. They saw the difficulties, ttt. interconnectedness of
not create
more problems than it solves. things, the second- and third-order effects of simple actions.
Decision making seems to invite a different kind of style
KNOWLEDGE
lr1lE AT\TT\ a/rrrrr^rY
AND ACTION from that of knowledge. Decisions require cla úty, closure, and
The usual modern motivational basis for great action confidence. As a result, decisive action comes more easily from
lies in the ignor ant than from the wise, more easily from the short-
hopes for great consequences, but those hopes
are persistently ,i
sighted than from those who anticipate the long ru¡. Hamlet
undermined by the many ways in which históry
frustrates them. bemoaned the way the "native hue of resolution ir sicklied o'er
The more one knows about the world, the less inclined
one is to with the pale cast of thought"s-the contradiction between
embrace a vision of delibe rate human control
over history;; thoughtfulness and action. This cliché of literature-that the
the less one is willing to make the commitment
demun¿áá from pursuit of knowledge is inconsistent with making decisiols-
decision makers. It is a comTonptace observation
and experience that they undermine confidence "b""r-aging has some basis. Knowledge creates both arrogutrré and doubts.
in the efficacy Arrogance supports action, but doubts undermine it.
of decision making, that only the youthful, the i".f;;;rsd,
and the very successful retain a beiief that ttrey The contrast between the timorousness of knowledge and
can affect des- the assertiveness of decision can be seen in decision ñraking
by theii own actions. Experience and knáwledge
'ity ambitious into cynics and .rrroutug.-rittt.t convert meetings. Meetings often begin by consulting knowledge, bt
the passivity
'ávvlf Lvr or
v the making sure that experts ate heard and considerationS are
crass pursuit of immediate personal pleasure.
Saying when, where, how, ot *hy u decision aired. But before long, meetings become instruments of confi-
occurs t¡lpically dence building. Participants reassure each other that the deci-
becomes more difficult as one' understands
decisions better. sion is right, that it is supported by logic and by evidence. With-
Showing whether a particular decision fits into
a particular out such reassurance and the collective curtain it draws on
266 APRIMERoN DECISIoN MAKING Decision Engineerlng 267
:

individual knowledge, informed decision making is Those rituals and ceremonies of decision making are rein-
hard, for
knowledge accentuates uncertainty about the efficacy forced by interpretations of the career experiences of decision
of action.
makers. People who are successful normally discover ample
SUSTAINING ACTION grounds in their experience to sustain a belief in their signifi-
cance. The stories they tell develop a story line that promotes
Despite the encroachments of knowledge, one thing such a belief among readers and listeners. Hierarchicál promo-
that seemg
obvious about the human estate is th; persistence tion systems in which individuals compete for decision making
of human
ambitions for significance. Myths of importance, the authority on the basis of past performances make it likely thai
idea uf ui.
sion and its implementation, the idea of meaningful individuals with decision making authority will be relativeiy un-
action, and
the idea of human will are not only ancient traditions encumbered by doubts about their capabilities for makin g adif-
but also
conspicuous parts of contempor ary human faith. ference. People who are promoted in a hierarchy are g*lr"rally
Support for
decisiveness in decisions in actions normally comes more resistant to false beliefs in impotence than to false beliefs
from one of
three different sources: hopes for consequence, pursuit in control. SimilarlY, an educational system organized by pro-
of iden-
tity, or arbitrary willfulness. Each contiibutes to maintaining motion on the basis of success predisposes those who are pro-
commitment, and each has its difficulties. moted to believe in possibilities for human control.
The usefulness of mythic stories of decision maker signifi-
Hopes fo, Conseql¿ence.Some people come to believe that therÉ cance does not depend entirely on their truth. Any rush ñ so-
is, in fact, a role for decisive ñ.rrnan action in determining phisticated doubt about the possibilities for significant action
the
course of history. Some find hope in a general sense should not ignore the extent to which a belief in heroic conse-
that human
destiny is within human controi. otheis find hope quences can sustain commitment to meet unreasonable de-
in the way in
which path-dependent, branching histories are sensitive mands on decision makers. "Thke away an ordinary person's il-
to
small interventions at particular poittts. Those beliefs lusions," says I)r. Relling, "and you take away happiness at the
may be il-
lusions' as many students of histoty would say they same time." u
are;but they
are beliefs that are conspicuous and are supported The issues extend beyond individual dreams of happiness,
II by many
other students of history. however. There are potential social costs in a loss of confidence
A few of the individual and social mechanisms by which deci- in human significance by decision makers. Consider two general
sion makers develop confidence in their own capabilities types of errors a decision maker might make in assessing the
for af-
fecting action have already been described. When great importance of intention al actions in controlling history. A deci-
cor'-
mitment demands great expectations but history cannot sion maker might mak e a "false positiv e" error of coming to be-
provide them with clarity, myths of heroes are invented. lieve in considerable personal control over outcomes when, in
Stories
are told in which the flow of history is attributed fact, that control does not exist. Such a belief would lead to (fu-
to the inten-
tions, strategies, and virtues of indiüdual decision tile) attempts to control events, but it would not otherwise af-
makers. The
routines of decision making are orchestrated to confirm fect results. Alternatively, a decision maker might make the
those
visions of significance. fnformation is gathered "false negative" error of comittg to believe personal contról is
and reported to
symbolize the importance of decisioá makers.
Meeiings are not possible when, in fact, it is. Such a belief would lead to self-
held to symbolize that decision makers are in control. confirming withdrawal from efforts to make a difference. The
Memo-
randa, procedures, and rhetorics of action symbo social costs of the first type of error seem small relative to the
lizethe impoi-
tance of decision making. second. Given a choice, most people would probably prefer a
268 A PRIMER oN DECISIoN MAKING
DecislonEngineering 269
:

society that errs on the side of making false positive errors in


assessing human significance rather than talse negative errors. human being. Action is sustained by socialization into a sense
Heroic myths constructed on ambiguous hislories present of its necessity.
one important difficulty, however-the vulnerability of decision obligations form the basis of commitment within old tradi-
making to intelligent doubt about the myth. I)ecisión makers tions of honor, duty, self-respect, and self-actualization. Those
of traditions also have problems. They have often been used to
judgment and intelligence find it hard to believe unconditional,
ly in their own significance. Great expectations become ques. justiff stupidity, obstructionism, and tyranny. But they offer a
tionable, and enduring the frustrations of decision making counterbalance to exclusive reliance on a logic of consóquence.
seems to require either blindness to the ambiguities of histoü Decision making can be seen as driven lesJby personal incen-
or an irrational acceptance of costs withoui likely benefid. tives or heroic e4pectations than by obligation to a sense of self.
Hence the standard dilemma of intelligent decision makers:
They must, choose between maintaining a belief in a myth of Arbitrary willfulness.In some traditions of human action, com-
history that they see as dubious and abandoning their commit. mitments to seek a better world do not depend on clear visions
ment to the significance of their roles. of preferences or identities, and willingness to try does not de-
pend on confidence in success or in obligations to self. These
Pursuit of ldentity. The dilemma posed by belief in decision effi- bases are, from this point of view, inimical to commitment, be-
cacy is real, but it depends on the assumption that great cause human commitment should be an arbitrary e4pression of
action will. It is precisely because decisions are without consequence
requires great expectations. There is another grand human tra-
dition for justi$ring great action, one that .rrr.o,rples heroic and are irrelevant to identity that they have significance. By act-
commitment from hopes for consequence and connects it to a ' ing decisively in decision making without any legitimate reason,
conception of self. The vision is one of obligations rather than the decision maker creates meaning in an arbitrary e4pression
expectations, of a logic of appropriateness rather than of conse- of aesthetics or faith.
quence' and of a sanity of identity rather than rationality. En- This üsion of arbitrary action is reflected in the Kierkegaar-
thusiasm and commitment stem from a willingness to embrace dian observation that any religion that can be justified is hárdly
the claims of identity, from a desire to achieve consistency with a religion. The conventional word is ,,faith,,, an irrational, sef_
an internal sense of self. less commitment that achieves its status precisely by its incon-
Decision maker identities are learned in the same way other sistenry with reason or with self. The medical/legal word is .,in-
social identities are learned, though instruction, imitatión, and sanity." The decision making structures that have traditionally
elaboration of what it means to be an acceptable decision sustained faith and insanity are the structures of family, ethnici-
maker. Those identities are grounded in culturál commitments ty,. religion, and nation.

to individuation through the making of choices and the accep- From this perspective, the irrelevance of decisions and the
tance of responsibility for them. Personhood is associated wiih inconsequential role of human intention in human destiny are
making decisions: choosing an education, an occupation, a mar- not justifications for inaction but a basis for rejoicing in action.
fiage, a family, a place to live, and a time to die. dhrn this com- If decision making were to make sense becaúse of itr conse-
mitment to life as the making of choices permeates identitieso quences, then it would be nothing more than economics. Ir de-
the denial of decision making is tantamount to the denial of ' cision making were to make sense because of its consistenry
self. To question the existence of decision options or to exhibit with an identity, then it would be nothing more than sociology.
passivity in the face of them is to reject ónr'* identity as Bu.t if decision making makes sense neither as a pursuit óf
a
objectives nor as the fulfillment of 'ah identity, then it repre-
270 A pRTMERoN DEcrsloN MAKTNc Decislon Englneertng 27 I
sents the ultimate declaration of individuality,
the right to sion process will yield decisions that can be unambiguously de*
freedom from concern about consequenr* scribed as intelligent. History is cluttered with decisions and de-
uná identity, the
proclarnation of an aesthetic of human
life. Don euixote said: cision processes that seemed to offer short-run improvements
"For a knight-errant to make himself crazy ro, u-r"uron war- in some domains but in retrospect created greater problems in
rants neither credit nor thanks; the point the long run or in other domains. And history is filled with evi-
' is to be foorish with_
out justification.,' 7 dence of the irrelevance of human actions. fn the long run, de-
cision makers are all dead and the species is extinct.
THE PURSUIT OF INTELLIGENCE Such a catalog of despair might be seen as an invitation to
abandon decision making. Hopes for magic in new technologies
The indeterminacies of decision intelligence and or philosophies of decision seem to be as unrequited as they are
the com-
plications in achieving it make the-purs,rit
of decision intelli. unquenchable. The issues ate ancient. Their resolution is elu-
gence frustrating. Arthough individuáls
and societi", iuu" tong sive. Knowledge about the difficulties involved in defining and
struggled with issues of-barancing the short achieving virtue through decision making leads to pessimism
run and the lon!
run, coping with the_ shifting nature of preferen.", about the possibilities for intelligent action. The means by
unJ identi-
ties, balancing benefits to some with which individuals justify action in the face of such problems
costs to others,
ing the symbolic and substantive elements "rú;;;;-
of decisions togeth- " might well seem to make the problems more severe. ff sustain-
er, modern theories of colrective decision
,'utiog ou"rli-" u* ing action involves colluding in illusions of control, accepting a
conspicuous for their failure to resorve
,.r.r, piolt"*r-io
definition of decision intelligence. The problims tt, social ideology of human identify that is disturbingly idio-
nously changing preferenceJ and identities
;i;"d";;: syncratic to modern Western society, or embracing one or
uó"rd thE another unreasoning faith, it is easy to conclude that action is
qeagh of any current formulations. And the symbolic "r"- c'ontent not worth sustaining. The alternatives manage to give passivity
decision making is alternately treated as
of
corrupt, paramount, or a good name.
irrelevant. A book on decision making might possibly conclude with
Efforts to improve the practicar procedures such an apostasy, but this one does not. Such an ending would
. of intelligent de-
cigign making without woirying abóut a precise be a mistake for two conspicuous reasons: First, the problems
definitiJn of in-
telligence seem similarly frustráting. A fruitful are difficult, but they are not beyond human comprehension
balance between
e4ploration and e4ploitation is peisistently and amelioration. It is possible to apply thought to the prob-
upset Uy Ayrru*i.,
of adaptation that lead decision-makers to lems of intelligence, to profit from an understanding of how de-
excess of one
othel' Gaining knowredge from e4perience or from or the cisions happen in order to make them happen better. Much of
others is
persistently confounded by timitations in the present book has been devoted to laying a foundation for
-efforts individual ana oigani
zationalcapabilities. And to develop usefur te"rrnoiogies such improvements. Foundations do not keep the rain out, but
of foolishness and poetic ambiguity to comprement they offer some small basis for imagining that a roof is possible.
technoro-
gies of reason and information .Iurity
.""-,á runl"," r"i""r- Second, if human commitment depended on a comprehen-
tial ideological resistance. sion of life and on hopes for consequence, cornmitment would
Any reasonable contemplation of that record of be less beautiful to contemplate than it is. The elegance and
failure
erodes confidence in the idéa that decision beauty of human life is augmented within a vision of decision
making uná tt
pectations of meaningful action,that it " "*-
impries i.-pru"ri¡1, making, and the human spirit is elevated. The idea of decision
routes to a moral rife. It is hard do be confiáent making gives meaning to purpose, tq self, to the complexities of
that árv ¿.ri-
272 A PRrMriR oN DHCrsroN MAKTNc

social life. It ennobles as it frustrates. So it is not despair


that is
appropriate but a reminder from Sénancour: "Man is perish-
able. That may be. But let us perish resistirg; and if nothingness
is what awaits us, let us not act in such a way that it
is a just
fate." 8

And if the earthy world of decision making occasionally leads


us to the divine world of Ibseil, Cervantes, Kierkegaard, and Sé-
nancour, let us be grateful. But let us not stay in that
world too
long' It is hard enough to make sense of the simpte things
with-
out discovering they are really not as simple as they look.

Potrebbero piacerti anche