[MUD-Dev] A footnote to Procedural Storytelling

[Ola Fosheim Grøstad <o.f.grostad@notam.uio]

J C Lawrence wrote:
> So if it is acceptable for the stories to be interesting only in
> retrospective, that's a WHOLE other ballgame.  Janet Murray comments
> on the live improv show only being entertaining for the
> improvisationalists, and not the audience.  If our definition of
> "acceptable" stories changes to enclude, "really good fun to tell
> others about but boring at the time", what the heck ARE the metrics
> of a "good story"?

In essence: FLOW ("intensity"), which of course is what the improvisers
experienced. What happened mattered to what they could do to attempt to
control the situation, thus they invested the necessary energy.

One could argue that "a good story" has little to do with the text
itself and a lot to do with the expectations of the reader and emotional
factors (largely dependant on presentation). Context matters, the same
text would probably be received differently if it was presented as
written by God, Einstein, The President, Hitler or a computer. If you
are interested in computers then the computer generated story tells you
more than if you don't care about computers, because it tells you
something about what can be done with computers (or that particular
algorithm).  A story written by a six years old can be interesting
because it tells you something about the child or children (and our own
past, thus you identify yourself with what is presented), the same story
written by a computer is less interesting to most people... The story
text is absolutely not "self-contained"...

A "story" in a quest in a MUD can be much more interesting if you need
something (outcome, information, thing) that it provides. And even more
so if it can only be attempted done once every month. This is rather
analogous to the improviser example.

A note: It is rather obvious that a computer in theory can generate a
good story (just generate all kinds of stories, at least one are going
to be interesting if interesting stories exist). I don't think a
computer generator can guarantee that all stories generated will be
interesting to everybody, but that goes for human authors as well. So,
what you are left with is: can a computer generate plausible stories
sufficiently often? As most admins would like to maintain some control
you don't need the computer to generate the entire story either, in
particular what the reader is likely to find interesting, it is left
with "spelling things out" and sequencing. Generating is a less
difficult problem than recognition. So I believe this is possible if it
is possible for a computer to estimate whether a given text is likely to
be acceptable to a reader. I don't view that as a far-fetched idea.

--
Ola  -  http://www.notam.uio.no/~olagr/





_______________________________________________
MUD-Dev mailing list
MUD-Dev at kanga.nu
http://www.kanga.nu/lists/listinfo/mud-dev