Newsgroups: rec.arts.int-fiction
Path: news.duke.edu!newsgate.duke.edu!nntp-out.monmouth.com!newspeer.monmouth.com!news.maxwell.syr.edu!newsfeed.skycache.com!Cidera!news-reader.ntrnet.net!uunet!ash.uu.net!world!buzzard
From: buzzard@world.std.com (Sean T Barrett)
Subject: Re: Sycamora Tree: "Inform is outdated"
Message-ID: <GD33wv.2Mr@world.std.com>
Date: Wed, 9 May 2001 20:07:43 GMT
References: <lve8etcbql4ccn6uj5flmeo0i52bdb0con@4ax.com> <GCtqry.KBA@world.std.com> <emshort-0405012000320001@user-2inik44.dialup.mindspring.com> <QY_J6.1$V24.105@inet16.us.oracle.com>
Organization: The World Public Access UNIX, Brookline, MA
Lines: 61
Xref: news.duke.edu rec.arts.int-fiction:86576

Mike Roberts <mjr-SEENOTE@hotmail.com> wrote:
>The author would never
>actually write any of the prose rendered by the game, but would somehow
>imbue the rendering engine with the author's prose style and narrative goals
>for the story in such a way that the text-rendering engine would write the
>prose that the author would have written to describe the evolution of the
>model state had the author been sitting next to the player narrating the
>gaming session personally.
>
>Clearly I'm talking strong AI pipe dream AD 2350 here.  Clearly your point
>is spot on for the foreseeable future and then some - the most writing any
>currently conceivable "prose engine" is going to do is a little parameter
>substitution here and there, and even that is going to be painstakingly
>prearranged by the author.

While I'm in sympathy with the overall sentiment here, I think you
underestimate the current system.  There is one (and only one) deep
simulation in current IF: object containment.  (A few other things
are simulated, such as light, but it's a very surfacey simulation.)

Correspondingly, there is also one significant piece of text that is
generated algorithmically in all games: object lists, especially
nested (recursive) object lists.  These can be overriden entirely
by producing a sentence per object; Inform allows applying some
controls by grouping objects in the list.  But there are no hooks
for controlling the "style" of the text (although there are hooks for
doing things like changing it from an English phrase to an indented
list); and, indeed, as soon as you get many objects in the list (say five),
the text becomes awfully mechanical sounding--written in a way no
human author would ever write.  But it is clearly not the kind of text
brought to mind by "parameter substitution here and there, ...
painstakingly prearranged by the author".  (If it was painstakingly
prearranged, it was by the libary author.)

And yet nobody seems to be complaining about this aspect of IF [*].
Certainly some of that may be because the "high-quality prose"
games tend to have fewer objects in lists because they have fewer
objects lying around--e.g. Hunter, In Darkness--but the big puzzle
games may have many objects--even in the case of something like
Spider & Web, which has a huge inventory list.

So I think there is a degree to which creating a deeper simulation,
and in return giving up control and quality of prose for the things
involved in that simulation, is not such a big deal.  This is more
from a game designer standpoint--deeper simulation allows more
interesting gameplay--than from a writer's standpoint--the prose is
likely to become more mechanical-sounding--but it in no way affects
your ability to tell a particular story.  (Simulationism itself may
affect your ability to tell a story, but turning over parts of the
game to generated prose does not; and anyway, how many people feel like
there is so much detail in the object containment model that their
ability to tell a story is restricted?)  And one could hope that
the mechanical sounding text would become unnoticed, or rather
transparent, in the same way object lists are today.

SeanB
[*] Except me. I've spent a huge amount of time[**] on an object-list
generator in TADS3, trying to come up with a technology that can
produce more natural, contextual output.  So far, it is a total failure.

[**] Well, a couple of days.
