Sei sulla pagina 1di 8

Computers & Graphics 26 (2002) 13–20

Artists and engineers as cats and dogs: implications for


interactive storytelling
Chris Crawford*
2349 Sterling Creek Road, Jacksonville, Oregon, OR 97530, USA

1. Introduction: the two cultures problem neither of us could figure out what the other was talking
about. We were wise enough to accept the chasm
Forty years ago, the British philosopher Snow [1,2] between us in good humor; neither insisted on the
warned us of a dangerous trend in Western intellectual correctness of their own worldview. I hope our students
culture: its bifurcation into two warring camps, the appreciated the lesson of our cooperative mutual
sciences and engineering versus the arts and humanities. incomprehension.
Whereas, previously, these two spheres had intermixed
amicably, they were splitting apart and failing to
communicate. The danger in this, Snow warned, was 2. The lesson of computer games
that science and engineering give us power, but the arts
and humanities help us understand how to use it wisely. The sad history of computer and video games clearly
Snow’s warnings generated considerable discussion, demonstrates the nature of the problem. Born and raised
most of it concurring, but the cultural division between by computer programmers, these games have always
the two grand genera of Western intellectual tradition been the playground of technical people. The themes
only deepened. Scientists and engineers regard artists as they offerFblood-soaked battles, feats of hand–eye
fuzzy-heads who cannot handle the discipline of coordination, and obscure intellectual puzzlesFare just
rigorous thinking. Artists dismiss technical people as what the technical people love. But the artists and the
gearheads who just do not understand the human majority of consumers find such entertainments boring
condition. As the chasm between the two widens, each and distasteful. Computer games are now the confined
side’s comprehension of the other wanes. Few artists largely to young males. Women and most mature men
understand the simplest concepts of physics, and few shy away from these games.
technical people grasp the basics of artistic expression. Recognizing the magnitude of the problem, games
Nowhere has the gap between the two cultures been companies have long sought to break out of the rut in
more destructive to our efforts than with the problem of which they find themselves. One manifestation of this is
interactive storytelling. The problem demands the best the futile quest for ‘‘games for girls’’. The first
efforts of both artist and programmer, but the cultural expression of this effort was the attempt to make Pac-
divide between these two has doomed every effort to Man more appealing to women by putting a pink ribbon
failure. on top of his head and transgendering him to Ms. Pac-
My first lesson in this problem came 6 years ago when Man. Artists find such an effort pathetically laughable,
I worked with a brilliant Hollywood scriptwriter in but it was the best the engineers could think of.
teaching a course on interactive storytelling. We team- The attempts continued. A number of companies
taught the course, one of us teaching the first 80 min of attempted to reach non-technical players by producing
the 3-h weekly lecture, and the other taking the last text adventures, games without violence or strategy.
80 min. Each week, we would meet for dinner to go over However, these games were elaborate puzzles that
our notes and compare our content. These dinners were ultimately demanded wild guesswork combined with
fascinating exercises in mutual incomprehension. Each meticulous logic, a combination that never worked for
of us knew the other to be an expert in his own field, but non-technical people.
The dawn of multimedia in the early 1990s gave a new
*Corresponding author. Tel.: +1-541-899-9150; fax: +1- twist to the futile quest. A great many people believed
541-899-7926. that the ability to produce beautiful imagery and
E-mail address: chriscrawford@wave.net (C. Crawford). digitized sound would make the computer accessible

0097-8493/02/$ - see front matter r 2002 Elsevier Science Ltd. All rights reserved.
PII: S 0 0 9 7 - 8 4 9 3 ( 0 1 ) 0 0 1 7 4 - 1
14 C. Crawford / Computers & Graphics 26 (2002) 13–20

and entertaining to a broad spectrum of players. An 3. Case in point: Shrek versus Final Fantasy
avalanche of brainless product descended upon the
consumer, boasting great beauty that was, alas, only A sad and particularly illuminating example of this
skin-deep. Customers dutifully laid out money for attitude was played out for all to see at the box offices of
products that, they were assured, represented the Next theaters in the USA when, in the summer of 2001, two
Big Thing. After sampling these pathetic offerings, the animated movies, both created using computer graphics,
customers wisely never came back, and millions of were released. The first, Shrek, was clearly the work of
dollars of venture capital money went down the drain. people from the art side of the chasm. The computer
The programmers were utterly baffled by these graphics were made to serve the story. Shrek was a story
developments. How could anybody not be entranced first and foremost, told through a medium using
by such beautiful graphics, they asked themselves. They computer graphics. The technical side of the effort was
built ever more advanced graphics engines, demonstrat- competently executed, but no grand breakthroughs were
ing specular reflection, diffuse reflection, true point demonstrated. Nobody went to Shrek to see glorious
source shading, and all sorts of other mighty feats of graphics. They went to see a movie, and they got one.
technical prowessFyet the customers just did not seem Final Fantasy, on the other hand, was the cutting edge
to appreciate their accomplishments. The fact that the of high-performance computer graphics. The graphics
images in their games were soulless never dawned on the were impressive, all agree. Indeed, the primary selling
engineers. Computer games companies were dominated point of the movie was the computer graphics. We were
by technical people, and artists were relegated to minor breathlessly told that the heroine’s hair was modeled
support positions. This lopsided approach guaranteed with 6000 separate strands. All manners of major
that electronic games would never have the spark of life technical advances were demonstrated in the film. This
that animates artistic efforts. was clearly the work of technical people; in fact, the
There were a few serendipitous successes. Rob Land- director was a computer game designer. The graphics
eros and Graeme Devine, an artist and a programmer, occupied center stage of the design; story and character
formed a partnership and the chemistry between these were secondary considerations.
two was favorable. They built a great product, The The results played out clearly in the box office. Shrek
Seventh Guest, a game combining technical excellence was a huge hit, generating more than $250 million in its
with artistic flair, and the game was a huge hit. first 12 weeks. Final Fantasy, on the other hand,
Unfortunately, the chemistry between them failed, their bombed; after just two weeks, its box office receipts fell
subsequent products lacked the magic of the first, and below those of Shrek and came to a total of only $30
they eventually broke up in frustration. million. The critics were just as scathing as the box
Failures were more common than successes. Certainly office: they panned the movie for its poor story.
the most spectacular failure was Rocket Science, a Some might claim that this demonstrates the clear
company launched with much fanfare and even more superiority of the artists, and proves that engineers
money. Rocket Science was to be the perfect embodi- should stay out of the entertainment business. I agree
ment of the marriage between Silicon Valley and with this conclusionFwith one important proviso: when
Hollywood. It commanded the top talent from both a new medium is developing, and the technology is still
communities, it poured money into its products, and it in flux, then the engineers must play an important role.
promoted itself cleverly. Yet, when the first products Shrek worked because the artists were content to use
emerged from the Rocket Science labs, they were computer techniques that were not state-of-the-art. Our
duds. Rocket Science had committed the same own field of interactive storytelling is still in its infancy,
blunder that all the games companies were making- and the artists in this field must therefore accept the
Fonly Rocket Science’s version was the mirror image difficult task of forging an effective working relationship
of the game companies’. Where games companies with the engineers.
put engineers in charge of artists, Rocket Science put
artists in charge of engineers. The artists just did
not understand interactivity, and so Rocket Science 4. How to combine artists with engineers
produced games with glorious graphics and pathetic
play. All that money, all that effort, was a complete How can such a relationship be constructed? Perhaps
waste. if we could find individuals who have embraced both
And so today we have an entire industry for sides of the divide, people who can feel like artists and
interactive entertainment that has utterly failed to think like engineers, perhaps those people could solve
produce anything of broad appeal or merit. It has the problem. I agree that such people would be valuable
belatedly realized that story, somehow, is important, but to any interactive storytelling effort, but I do not think
it treats story as a component to be plugged into the that they are enough. I consider myself a scientist who
game rather than an end in itself. has approached the art of storytelling with respect and
C. Crawford / Computers & Graphics 26 (2002) 13–20 15

some appreciation, and I have made much progress in storytelling, is interactive, and interactivity is a process,
tackling the problem, but I do not consider my efforts not a datum. The specification of a process is nothing
alone to be adequate. I need specialists from each side of other than an algorithm. Artists who wish to creatively
the divide. Having collaborated with a published control interactivity must do so by creating algorithms.
professional novelist, Laura Mixon, I can offer some How much mathematics must an artist learn? Nothing
thoughts on how to handle the relationship between more than what teenagers learn in their first year of
engineer and artist. algebra: the algebraic properties of addition, subtrac-
The central task in managing this relationship is to tion, multiplication, and division. Square roots are also
establish the dividing line between each specialist’s area useful, but logarithms are unnecessary and trigonometry
of responsibility. In well-established fields of endeavor, is certainly too advanced for work at the current state-
we have developed clear specifications for each specia- of-the-art of interactive storytelling. I have devoted
list. For example, a great many talents must cooperate considerable effort to reduce the requirements for
to make a movie, yet the film industry has hammered algorithmic expression in my own interactive storytelling
out a system of rules that assign primary creative control system, and I have been unable to find any practical
over certain tasks to certain specialists, with overall system that avoids using these basic functions. They
control going to the director. The cinematographer can seem to be the irreducible minimum for algorithmic
contradict the director on certain matters; the sound expression.
specialist can weigh in with strong opinions that It is possible, of course, to eliminate procedural
command respect; and actors always have something considerations from the algorithmic concepts required
to say. But in the newly developing field of interactive of artists, and this goes a long way towards simplifying
storytelling, we do not have such conventions. the problems faced by the artists. Unfortunately, this is
I can offer a vague rule of thumb: any task, rule, or no grand leap forward; it is simply one of the benefits of
plan that can be clearly specified should be assigned to object-oriented programming languages. I would cer-
the engineer; anything that defies regularization should tainly urge all designers of interactive storytelling
be assigned to the artist. To put it in computer science systems to take advantage of these benefits.
terms, the engineers should handle algorithms and the
artists should handle data. This, however, raises a
fundamental issue. Algorithms are the heart and soul of 5. Development environments
computing. Data is the mere fodder of computing;
algorithm uses data as a steam locomotive uses coal. An I can now propose a framework for cleanly separating
artist who supplies only data merely shovels coal into the work of the engineer from that of the artist. That
the locomotive; the engineer actually runs the train. If framework is the development environment built by the
we grant all the algorithm-making power to the engineer and used by the artist. The capabilities and
engineers, then the engineers will always be the driving constraints of the development environment are defined
force behind interactive storytellingFa prescription and created by the engineer; the artist creates the artistic
sure to lead to dismal results. Ergo, the artist must work within that environment. This makes the working
wield genuine algorithmic power. relationship between artist and engineer concrete: they
This prescription runs afoul of artists’ deep-seated negotiate the feature set provided by the development
disdain for all things mathematical. Hopelessly en- environment. Standardized components of the system
tangled in two cultures prejudices, most artists refuse are programmed into the development environment by
to learn the mathematical language necessary for the engineer; artistic flights of fancy are realized through
algorithmic self-expression. I am saddened by this the artist’s expressions within that development envir-
betrayal of the boldness of the true artistic spirit; onment.
perhaps it demonstrates why great artists are as rare as At this point, an extended example will clarify these
great scientists. The historical fact is that for centuries, points. I built a development environment for my
great artists have mastered technical fields in order to interactive storytelling engine; it was used by science
advance their art. Indeed, in one technical field, fiction novelist Laura Mixon to create a storyworld. The
metallurgy, almost all of the technical advances up to history of that development environment illustrates the
the 19th century were the work of artists. The problems of disentangling the engineer’s work from the
mathematics of perspective was invented by the painters artist’s work.
of the Renaissance; the Golden Rectangle and its At the outset, the development environment (herein-
associated geometric relationships were the creation of after referred to as the Erasmatron [3]) provided control
Greek architects and sculptors. Harmonic ratios were systems for defining numerical values of variables such
first explored by musicians. as Greed, Libido, and Empathy for actors, or Affection
There is no alternative to the use of mathematics in and Trust for the relationships between them. We
this field, for interactive storytelling, unlike conventional quickly learned that the variable set was itself a major
16 C. Crawford / Computers & Graphics 26 (2002) 13–20

cause for discussion. Laura kept asking for more and and any intelligence that it can bring to bear can
more variables to address every conceivable situation. just as readily be brought to bear in a text-entry
We added many more variables such as LoveHunger system. What possible advantage can counterweigh the
and Nurturance. Later, we both realized that too obvious tactile disadvantage of the point-and-click
many variables only confuse the artist; we therefore system?
trimmed the list of variables. This led to long The answer lies in which comes first: user input or
discussions of the true merits and applications of each application of syntax-evaluating algorithms. In a text-
variable, and especially to discussions of overlap among entry system, the user must be free to type any desired
the variables. text; the system applies its syntax-evaluating algorithms
after the user has typed the text. This places the
computer in the role of judge, and the user in the role
6. The scripting language of supplicant. The poor, ignorant user hopefully types
his text and submits it to the computer for judgment.
The most complicated portion of the Erasmatron is The computer evaluates it and passes judgment: either
the scripting language used by the artist to control the the text is acceptable or it is rejected with a dismissive
decisions made by the actors. This scripting language is error message. The user is forced to dance to the tune
the algorithmic heart of the Erasmatron, for every story played by the computer. This is demeaning, demoraliz-
turns on the decisions made by the actors at each ing, and demotivating; many users are chased away
juncture. The artist requires a scripting language rich from programming by the harsh judgments of such
enough to express the nuance of an actor’s decision- programs.
making process, but the artist also needs a simplified In the point-and-click system used in the Erasmatron,
language within his or her grasp. These conflicting the syntax rules are applied before the user’s input.
requirements dictated the course of my work on the Rather than criticize the user for syntax errors, the
Erasmatron. Erasmatron simply does not allow the user to make
them in the first place. This requires more extensive
calculations on the part of the software. Whereas the
7. Point-and-click entry syntax evaluator in a text-entry editor must judge what
has actually been typed, its analogue in the point-and-
My first decision was to eliminate keyboard input for click editor must consider everything that might be
the scripting language. The egregious syntactic pickiness entered. The former routine simply runs down a tree of
of production languages like C++ surely substantiates possibilities to determine which leaves have been
the wisdom of my intent, but most programmers have activated, and then determines the correctness of each
difficulty imagining a practical alternative. My solution leaf. The latter routine must examine every single leaf in
was the use of point-and-click editing. Although the the tree and obviate those leaves that are inappropriate
scripting language is expressed on the screen in a textual in the current context. This constitutes a great deal of
format, it is edited keyword by keyword rather than more work for the processor. However, there is little
character by character. Thus, to edit a script, the artist additional workload for the programmer. Each and
clicks on a keyword, thereby highlighting it, and then every syntactical rule must be expressed in code, and
selects a new keyword from a set of menus, which therein lies the programmer’s workload. Applying the
replaces the selected keyword. Thus, an initial expres- syntactical rule, whether it be done few times or many, is
sion is expanded from within in a series of steps rather the processor’s workload, and I say, let us keep those
than composed in its entirety and typed. This type of CPUs humming! An idle processor is Murphy’s work-
entry system is often denigrated as overly cumbersome, shop!
but in our case, it seems to operate quite smoothly. Even One might object that, in a truly rich expressive
though the system comprises more than 400 different environment, the number of choices available to
keywords, they readily sort into intuitively obvious the user at any given juncture is so great that the
groupings that become the basis for menus. Assigning evaluation of all those choices would be prohibitively
the most frequently used keywords to a special button slow. My own experience with the Erasmatron puts to
bar at the top of the window greatly facilitated the rest such fears; with 400 keywords, the delay imposed
editing process. by screen refreshing is a hundred times greater than
Some readers, familiar with the many syntax-assistant the delay due to option evaluation. Moreover, even
features built into modern programming environ- natural language does not demand overmuch in this
ments, might wonder why a point-and-click editing matter: at any given point in the construction of a
system is preferable to a pure text-entry system. After sentence, the user has an average of only about ten
all, the point-and-click system is undeniably clumsier, choices that semantically and grammatically fit the
from a tactile point of view, than a text-entry system, context [4].
C. Crawford / Computers & Graphics 26 (2002) 13–20 17

8. Screen layout Every variable presented on the scripting screen is


drawn in its characteristic color, leading to a simple
This screenshot (see Fig. 1) illustrates the basic layout visual rule: you can only enter values of the same color
of the scripting editor. as the slot you are editing.
A minor innovation apparent in this screenshot: any Particularly unorthodox is the editor’s refusal to
screen location that is ‘‘hot’’ (i.e., on which a mouseclick permit the creation of any wholly new element by the
will make something happen) is tinted light blue. Why artist; existing elements are expanded or augmented
designers insist on keeping secret which screen objects instead. Should the user make an entry that calls for
are active is a mystery to me. some associated new element, then that new element is
Another important decision was to use strong but automatically created by the editor at the instant it
fixed variable typing into the scripting language. There becomes needed. It has always irritated me that, when I
are eight and only eight data types in the Erasmatron, create a new variable in a C++ program, I must
each with its own characteristic color: remember to enter the same variable in the variables list
at the beginning of the function.
Numbers (red) For example, when the user creates a new role, which
Booleans (black) in turn requires a new script, the editor creates an initial
Actors (blue) script with the minimum content required to function.
Props (pink) When the user adds a new choice to a script, the editor
Stages (brown) creates all the assignments required by that choice and
Verbs (green) inserts them into the script; the artist need only expand
Events (orange) on the minimal entries made by the editor. For example,
Groups (purple) suppose that the artist wishes to edit the inclination

Fig. 1. Script editing display of Erasmatron.


18 C. Crawford / Computers & Graphics 26 (2002) 13–20

formula for a verb: 10. Thinking like an artist, not an engineer

Inclination½ThisChoiceo ¼ 0 Procedural thinkingFthe ability to mentally step


through a program and imagine its behaviorFis a
An actor’s inclination to make this choice is initialized to mental perversion that all programmers must master to
zero. Suppose that the artist wishes to expand the value ply their craft, but it befuddles most decent, right-
0 to be the affection that the subject feels for the direct thinking people. An effective scheme for insulating the
object of the event. Instead of typing ‘‘Affection[Subject, artist from procedural thinking is the conflation of
DirectObject]’’, the artist selects the zero and chooses loops. The Erasmatron offers no looping structures to
the Affection token from the ‘‘Relationships’’ menu. the artist. Instead, it offers several ‘‘pick the best’’
The editor recognizes the Affection function as dyadic functions, such as PickBestActor, PickBestStage, and
and replaces it with the two unknown entries: PickBestProp. These allow the artist to specify two
selection criteria: a boolean criterion called Acceptability
Inclination½ThisChoiceo ¼ Affection½Who; ForWhom and an arithmetic criterion called Desirability. With
these two criteria, the function loops through all the
The underlining indicates that these two variables are actors, stages, or props and selects the item with the
unspecified values, and the editor expects the user to fill highest value of Desirability that also has a true value of
them in with entries. The blue color indicates that they Acceptability. The loop is implicit in the function, but
must be of type ‘‘Actor’’, and the text suggests the the concept of looping is never imposed on the artist.
meaning of the values. The artist can now click on these
to edit their values.
11. Use appropriate terminology

The example also demonstrates another useful design


9. Poison rule: use the artist’s terminology, not the engineer’s. This
demanded special discipline on my part. Time and again
Even sterner is my requirement that the engine must I found myself phrasing concepts in their computational
fail gracefully. If the artist has entered values that lead to terminology, thereby imposing my vocabulary on the
non-sensical results (such as a division by zero), or if the artist and deepening her confusion. Ofttimes a small
artist has failed to specify values, then the engine does effort would permit a quick change, such as when I
not crash or halt; it proceeds with the storytelling by changed my term thing (for an ownable or moveable
ignoring the tainted calculation and posting a ‘‘poison- object) to prop. The change required only trivial
ing’’, which can be reviewed by the artist after each test modifications to the program, yet made the concept
run. This eliminates the time-consuming and tedious immediately recognizable to the artist. Of course, the
sequential correction of minute errors before the term script is a happy coincidence of similar meanings.
creation can be examined as a whole. In one case, the use of theatrical terminology
One might rightly wonder how it is possible to ignore improved the conceptual clarity of the system. Stages
a tainted calculation and proceed with the overall in a movie have no spatial relationships among
calculation without in some way generating unreliable themselves; actors simply disappear from one stage
results. In most calculations, if an intermediate result is and reappear on another, without going through the
invalid, then the entire calculation is ruined and the tedious task of walking down corridors, through doors,
program must perforce abandon execution. In the case and up stairs. The dramatic laws of physics recognize no
of the storytelling engine, every calculation made is such thing as spatial continuity, yet I was all set to write
associated with a behavioral option available to a detailed algorithms for translating actors through a
character; should the calculation screw up, we simply spatially continuous universe. By thinking in terms of
ignore that option. The story does not crash, it merely stages rather than locations, I improved the storytelling
becomes a bit less interesting. engine and made its use more easily comprehensible to
For example, suppose that we have a story in which the artist.
the heroine decides to choose a husband. Suppose This was especially useful in the handling of the
further that the person playing the story has somehow scripting system. Rather than explain it as a program-
managed to kill off every eligible bachelor. When the ming language, I use the theatrical term role to designate
heroine comes to make her choice, the calculation must a particular dramatic context and the script assigned to
fail because there is nobody to choose. The calculation that concept. Thus, the artist defines for every action the
will therefore be poisoned and the engine will ignore the possible roles assignable to the various participants or
possibility of the heroine choosing a husbandFan observers to that action. Each actor then consults the
entirely rational solution to the problem. roles in sequence, examining the boolean React if
C. Crawford / Computers & Graphics 26 (2002) 13–20 19

expression. If that actor evaluates the React if expres- user; you must also indicate just how much power you
sion to true, then the actor executes the rest of the script. are delivering. Intelligently manage the perceptions and
I explain this to artists in the following terms: whenever expectations of your user.
an event takes place, there are a number of scripts
associated with that event; each script is in an envelope,
with a rule on front specifying whether the actor should 13. Conflicts between artist and engineer
open the envelope and carry out the script. The script
envelopes are passed around the actors on the stage, and These examples are in and of themselves unimportant,
if an actor finds that he satisfies the requirements to fill a and I do not offer them as feats to be emulated by
role, then he plays out that role using its script. others. Instead, they provide examples of a style of
Lastly, the simplest and most direct expression of this software design that helps engineers and artists to work
rule was to spell out words. Engineers have a penchant together. The engineer designs the tool that the artist
for rducng xprsns 2 smllst pssbl fm. Perhaps this was uses, and the artist makes requests for feature improve-
useful in the bad old days of 40-column video displays, ments. I have found that artist suggestions usually fall
but nowadays such economies save electrons to waste short of my ideals. Ofttimes the artist will attempt
more precious acetylcholine. DUA! (Do not use software design, suggesting some scheme or idea in its
acronyms!) Spell out your expressions! computational terms. Since most artists are lousy
programmers, these suggestions are often lame, but the
engineer should not dismiss them with the derision that
12. Unbridgibility the ideas themselves deserve. Instead, the engineer
should interrupt the technical description by asking the
Sometimes the chasm between engineer and artist is artist, ‘‘What are you attempting to accomplish with this
simply too wide to be easily bridged. In the Erasmatron, idea?’’ Empathetic probing will often reveal a worthy
I created an elaborate and, I blushingly say, quite clever idea underlying the atrocious technical suggestion.
debugging environment that gave the artist the power to For example, at one point, Laura desired to have her
operate the storytelling engine a single step at a time, actors attend a party. She wanted to be able to direct
watching its calculations step by step so as to discern the them to arrive at the location of the party at an assigned
nature of a problem. In practice, my artist seldom made time, and proposed a scheme for doing so. Unfortu-
use of my beautiful debugger, because it required her to nately, her scheme would have conflicted with the basic
think in procedural terms. Typically, when she ran into a engine architecture, which algorithmically controls the
serious problem, she would complain to me, and then I movements of the actors. I could easily have transferred
would use the debugger to trace the problem and report the engine’s control of actor movement to her, which in
back to the artist. I was never able to correct this clumsy turn would have required her to specify all actor
arrangement, but I have achieved partial amelioration motions in all situationsFan onerous responsibility
with an elaborate rehearsal feature. (Note that rehearsal that would have consumed far too much of the artist’s
is a term that storytellers will instantly recognize and time. The solution we hit upon was to give the artist
appreciate.) The rehearsal plays out the storyworld control of an ‘‘appointment book’’ for each actor, which
hundreds of times and then reports back with a in turn was consulted by the engine in moving the actors
compendium of problems. Defining those problems about. If the artist did not use the appointment book,
has been a major task, and continued effort will be actors moved under the normal algorithms; if she did,
required to expand and improve the detective work of then the actors responded to her dictates. This gave her
the software rehearsal agent. the power to control their motions without the
This brings me to a minor observation about agents. responsibility to specify those motions in all cases.
We are all proud of the agents we create, and our chests
swell with parental delight when one of our creations
performs its stuff for our users. The sad truth, however, 14. Conclusion
is that our agents are still pitiably stupid, and often
create disappointment when our users treat them with Interactive storytelling demands creative partnership
expectations of greater intelligence than they can deliver. between engineer and artist. We have possessed hard-
This embarrassing problem is exacerbated by smugly ware capable of handling interactive storytelling for at
calling those agents wizards; a wizard is even smarter least a decade; the reason why we have not yet produced
than the average person. I therefore chose to use the good interactive storytelling technology lies in our
term lizards instead, and each of my lizards appears on failure to forge effective creative partnerships between
the screen in pretty little photo making it quite clear that engineers and artists. I fear that such partnerships
this agent is just a stupid reptile (compare Fig. 2). It is remain out of reach of most engineers and artists.
not good enough to deliver maximum power to your Engineers must confine their efforts to the creation of
20 C. Crawford / Computers & Graphics 26 (2002) 13–20

Fig. 2. Example screen: user interface for the storyteller with appropriate terminology.

tools for artists, making a good-faith effort to create the found at http://home.netcom.com/Bapstern/interacti-
tools that best meet the true needs of the artists; and vestory.net/links.html [5].
they must learn at least some of the artists’ terminology.
Artists in their turn must trouble themselves to learn
algebra; they must endeavor to express their needs to the
engineers as clearly and precisely as possible; and they
must accept the limitations of technology without References
blaming the engineers. Most of all, both sides must
[1] Snow CP. The two cultures, reissue of snow’s lecture of
overcome their mutual antipathy and approach each
1959. Cambridge: Cambridge University Press, 1993.
other with genuine respect. Only then will the guidelines
[2] http://academics.vmi.edu/gen ed/Two Cultures.html
presented here help them cooperate productively. [3] http://www.erasmatazz.com
Readers may find more information on the Erasma- [4] Miller GA. The psychology of communication. London:
tron interactive storytelling technology at http:// Penguin Books, 1967. p. 82.
www.erasmatazz.com [3]. A thorough index of websites [5] http://home.netcom.com/Bapstern/interactivestory.net/links.
on interactive storytelling and related technology can be html

Potrebbero piacerti anche