Newsgroups: rec.arts.int-fiction
Path: news.duke.edu!newsgate.duke.edu!news-hog.berkeley.edu!ucberkeley!sunqbc.risq.qc.ca!newsfeed.online.be!zur.uu.net!ash.uu.net!world!buzzard
From: buzzard@world.std.com (Sean T Barrett)
Subject: Re: Game design article
Message-ID: <GL8E3z.GEw@world.std.com>
Date: Mon, 15 Oct 2001 05:04:46 GMT
References: <9qcn9u$d47$1@news.panix.com> <Vony7.17194$gT6.10041293@news1.rdc1.sfba.home.com>
Organization: The World Public Access UNIX, Brookline, MA
Lines: 44
Xref: news.duke.edu rec.arts.int-fiction:93588

Kevin Forchione <kevin@lysseus.com> wrote:
>some of which can stretch out for several seconds, even to
>minutes per turn. To the IF community this kind of response time is
>unacceptable.

So we need VMs and world models that are written to be
performance-conscious.

>On the other hand, limiting behaviors to locales means passing messages to
>each object within the locale to see if it wants to "react". Stimulus /
>response then becomes more complicated as each individual object determines
>its own response and the number of objects increases.

The mud system I worked on in '93 addressed this problem with
a generic "subscription list", whichis faster if most objects
don't react to any given stimulus (on average).

Every time an object moves, it registers itself with
its parent container for all the stimuli it wants
to react to. Object movements are actually very
rare in MUDs (except player movement) and in IF--typically
at most one per turn, except the odd TAKE ALL.

Note that if you move an object which contains other objects,
those other objects don't reregister, so it becomes the job
of the container to register for all stimuli that its children
are interested in.

More food for thought: the puzzles in my game "The Weapon" are
almost all based around NPC sensing, and this was directly influenced
by my experience with sense-passing in "Thief: The Dark Project"
(which is referenced several times in the aforequoted presentation
for its simulationist NPC sensing). However, I had to hack it all
since Inform's has no NPC sense model. I'm hoping that with
something like Tads3's sense-passing model these sorts of behaviors
will be more consistent and predictable and yet still be usable
for puzzles like that. If so, it points out the advantage of a
more complex world model--more space in which to create puzzles
which fit naturally with the world (by 'world' I mean the dynamic
interactive world the player perceives, rather than that descibed
by the static text)--an advantage Harvey describes but that you
might wonder whether it's applicable to IF.

SeanB
