Newsgroups: rec.arts.int-fiction
Path: news.duke.edu!newsgate.duke.edu!zombie.ncsc.mil!alnews.ncsc.mil!feed.news.qwest.net!cyclone-sjo1.usenetserver.com!usenetserver.com!hub1.nntpserver.com!news-in.nibble.net!headwall.stanford.edu!newsfeed.stanford.edu!paloalto-snf1.gtei.net!crtntx1-snh1.gtei.net!denver-snf1.gtei.net!news.gtei.net!coop.net!world!buzzard
From: buzzard@world.std.com (Sean T Barrett)
Subject: Re: [Inform] Nice simple straightforward question
Message-ID: <GICo2M.MG1@world.std.com>
Date: Mon, 20 Aug 2001 04:49:34 GMT
References: <9kua39$1ei$1@news.panix.com> <20010819150224.01442.00000863@mb-ci.aol.com> <GIC6zC.B9M@world.std.com> <9lq26q$1bi$1@cascadia.drizzle.com>
Organization: The World Public Access UNIX, Brookline, MA
Lines: 42
Xref: news.duke.edu rec.arts.int-fiction:91328

Dan Shiovitz <dbs@cs.wisc.edu> wrote:
>I assume this has been suggested before, but I think you'd probably
>simplify things by admitting that taking an object is non-atomic, and
>breaking it down into actions which are (ie, free a hand, reach for
>the glass, touch the glass, grasp the glass, lift the glass, put the
>glass in your inventory, release the glass).

I suspect that when we read about Implementors talking about how
modern IF hasn't really moved beyond what they were doing, this is
the sort of thing they're thinking of.

And as a simulationist, I'm all for it. But "taking" is one of the
things we simulate decently already; the fact that people might be
more comfortable with a more simulationist implementation of TAKE
doesn't really clear up to me how to handle less simulationy-verbs
(SING, BLOW ON COW), and especially verbs that authors create
themselves (which shouldn't require them to make such detailed
breakdowns)--for which I think we'd still want a convenient-override
scheme.

Admittedly, I have sketched out designs for implementing both
a general concept of "nearness" (useful for a breakdown of large
scale actions, for implementing puzzles like "MOVE CHAIR NEAR
BOOKSHELF", and a few other things), a notion of "connectedness"
(useful for ropes and a few other things), and then a detailed
model of hands (using nearness to move hands near things, and
then connectedness to attach hands to things when they are gripped);
but I felt like the system would be overall too unwieldy, too
complex to generate text for (need to print the details if something
interesting happens, but not otherwise), and too baroque for
authors to understand when they just want to make a simple
override; and they would require a brand new library anyway,
so it would be far too likely to be an experiment that would
founder in the end. My apologies for the length of that sentence.

Perhaps because such techniques would also provide sufficiently
benefits for sane overridability they would still be worth it, but
I don't think I'm willing to invest the time in it. I guess it
would be interesting to try a small-scale implementation of it,
but I'm not sure there's any such thing.

SeanB
