IF Theory Article
Jorn Barger
Representing Stories: History, and a New Approach
by Jorn Barger
Abstract: To pack the maximum story content into the smallest memory, you
need a concise vocabulary that can summarize the full range of human
behavior. If person, place, thing, and motive are taken to be the simple
elements of such a vocabulary, the primitive compounds should be the
standard *relationships* that any two (or more) of these normally display.
While the ancestry of story representation includes the I Ching, Roget's
Thesaurus, and Polti's 36 Dramatic Situations, the modern ambassador of
story representation in the world of artificial intelligence is surely
Roger Schank. During the 70s at Yale, Schank pioneered an approach to AI
that involved building storybases in memory, and achieved what still stand
as the only successful natural language translation systems, albeit within
extremely limited domains.
Schank's student James Meehan's Talespin was an excellent prototype for
interactive fiction, stumblingly building aesop-like fables. It was
impossible to scale up, though, and was absurdly *brittle*, lacking the
sort of common-sense background knowledge that Doug Lenat's Cyc project is
hoping to provide, through its slow, painful, memory-extravagant brute-
force attack.
The prospect of a videogame appealing to a Cyc server for every little
nuance of projected reality, is horrifying to contemplate: realtime is no
time for *inferencing*! I want to propose a neat, Schankian end run, in
the form of a database of standard stories that, by simple recombinations,
give birth to a rich microworld.
The cornerstone of this system lies in that hoary old trio: person place
and thing. But we add to that a little psychology: a typology of human
*motives*: food comfort sex respect family self-expression. And then we
ask, for any two of these "initials"-- person place thing motive-- what
are their normal relationships?
Persons may go to places, or leave them, or traverse them, etc. Persons
may make things, use things, acquire things. Persons may suffer motives,
pursue satisfaction of them, satisfy them. And persons should have
standard relationships to other persons, too. They may be kin, or sexual
partners, they may communicate, or cooperate, or conflict with each other.
A story may involve one person weighing two motives, or one person
achieving gratification of a motive by an exchange with another person
whose motive may be different.
A story will be a sequence of relationships, changing in time. By
exploring these relationships combinatorically, and the simple stories
they generate... well, it might give a deeper weight to the concept of
'virtual reality'.
Table (abbreviated):
person-thing relationships: make acquire use break destroy
person-place relationships: go-to leave stay possess defend
person-motive rel's: gratify suffer
motive-motive rel's: tradeoff substitute
person-person-thing rel's: give take
etc.
jorn barger
jo...@chinet.chi.il.us
ps: coming soon: the relevance of joyce's ulysses to IF!
David Whitten
>Here's an article I wrote for the JCGD a couple of issues ago:
>
>Representing Stories: History, and a New Approach
>by Jorn Barger
>
>Abstract: To pack the maximum story content into the smallest memory, you
>need a concise vocabulary that can summarize the full range of human
>behavior. If person, place, thing, and motive are taken to be the simple
>elements of such a vocabulary, the primitive compounds should be the
>standard *relationships* that any two (or more) of these normally display.
>
>and thing. But we add to that a little psychology: a typology of human
>*motives*: food comfort sex respect family self-expression. And then we
>ask, for any two of these "initials"-- person place thing motive-- what
>are their normal relationships?
>
>jo...@chinet.chi.il.us
>
Let me start by saying I'd love to read the entire article!! the abstract
is fascinating!
In your typology of human motives, I would suggest that you include
survival instead of food. Since you have already thought of subclassing
your motive list, you can include food under survival, and I think a lot
of interesting behaviours of people relate to death, life after death,
survival before death, etc.
I know that if you are going to include person place and thing you should
also include ideas as well. (Since nouns can also be ideas...)
I think a sample typology would be a neat basis for story fragments as well.
I think you can get a pretty broad coverage of a lot of story fragments pretty
easily, but you do get a fast growth in number of fragmentss.
IE: if you have only four things taken two at a time == six combinations
only five things two == ten combinations
four things three at time == four combinations
The general formula (I believe--math wizzes please double check)
is n things taken m at a time thus==factorial(m)/(factorial(m)*factorial(m-n))
As an upper bound to the number of fragments you would have to store, we
are interested in the number of possible fragments you can get from:
n
----
\ ( n )
/ ( m )
----
m=1
Using the binomial expansion formula and (n m) to mean n things taken m at a
time: (ugly and hard put them on top of each other in ascii)
n 0 n 1 n n 0
(p+q) = (n 0) p q + (n 1) p q + ... + (n m) p q
and then substituting p=1/2 and q=1/2 means:
n
n ----
(.5+.5) = \ ( n )
/ ( m )
----
m=0
---------------
n
2
n n
moving 2 to the left hand side we get the sum equal to 2
but the sum starts at 0 so we have to subtract (n 0) == 1 to get the answer.
n
Thus the number of fragments would be 2 - 1.
Since the plots would actually only be interesting when you talk about
expansions to at least the next hierarchy level, you have a number of concepts
(person place thing idea survival comfort sex respect family self-expression)
I have expanded motive since I think interesting stories involve more than
one motive, but someone else may disagree with me.
These ten concepts would still only be 1024 basic story fragments. I know
that this seems big, but some of these possible fragments probably would take
a bit of thinking to even expect to make sense at all. What does family mean
without a person? or what does it mean to have self-expression for an idea or
a place?
Perhaps you have already done some of the hard work, but I think this would
be an interesting thing to discuss on rec.arts.int-fiction in the future.
comments anyone?
Dave (whi...@fwva.saic.com) US:(619)535-7764 [I don't speak as a company rep.]
Jorn Barger
a bit here.)
The structure I'm proposing starts with a simple abstraction hierarchy
that has person-place-thing-motive as its top-level discriminations,
but each of these are of course further subdivided (and one great
challenge is to find other concepts at this level of generality).
The topological trick is to do a cross-product of the whole hierarchy
*with itself*, recognizing that the stories suggested by the cross-
product of, eg, *survival* (a motive) with *tool* (a thing), *will
inherit structure* from the simpler crossproduct of motive and
thing. So we postpone the combinatorial explosion by carefully
exhausting the toplevel relationships before exploring
1) their specializations (the crossproduct of more-specialized
elements)
2) their *complexifications* (3- to N- dimensional crossproducts
of top level nodes)
There's a profound difference between these two sorts of ramification,
in that the menus for type 2 *will always offer the whole abst. hier.
over again*, while type 1's menus will be peculiar to the point
you're moving from.
An example:
Imagine you've navigated from the zero-point of this 'content
network' to the relationship "person uses tool". Now, you may
choose to specialize person or tool ( --> person uses crowbar),
or you may prefer to add an element to the "person-tool repertory"
like motive-survival, suggesting the obvious story "person uses
tool to satisfy survival motive".
More later....
jo...@chinet.chi.il.us (Jorn Barger)
Alejandro Rivero
In article 1...@chinet.chi.il.us, jo...@chinet.chi.il.us (Jorn Barger) writes:
> ps: coming soon: the relevance of joyce's ulysses to IF!
Which chapter of Ulysses would do you like to port to IF? Surely the errant rocks
could be a good example for programing, but not a interesting one.
I doubt Joyce being relevant to IF, at most he is in the same sense that Homerus.
(Ex: Judy Malloy book?)
My opinion, the first interactive book could be Rayuela, from Julio Cortazar,
in the 30's. You can read across chapters in any order. Cortazar suggest two
the short one (first half of the book) and the long one, mixing the rest of
chapters into the first group.
-Alejandro Rivero
Zaragoza Univ, Spain
Actually I havent read the book, I tryed it years ago, when I was younger, and it
bored me.
Ernest Adams
Journal of Computer Game Design? It is always looking for new
articles, especially serious theoretical ones.
Write to
The Journal of Computer Game Design
5251 Sierra Road
San Jose, California 95132
Phil Goetz
"The Beast that Shouted Love at the Heart of the World" was written
in about a dozen sections that could be read in any order.
There didn't seem to be much point to this ability, however;
it didn't make the reading interactive.