Newsgroups: rec.arts.int-fiction
Path: gmd.de!ira.uka.de!yale.edu!yale!gumby!wupost!udel!rochester!cantaloupe.srv.cs.cmu.edu!pww
From: pww+@A.GP.CS.CMU.EDU (Peter Weyhrauch)
Subject: Consistency in Worlds
Message-ID: <BxBDM0.6oG.2@cs.cmu.edu>
Originator: pww@A.GP.CS.CMU.EDU
Sender: news@cs.cmu.edu (Usenet News System)
Nntp-Posting-Host: a.gp.cs.cmu.edu
Organization: School of Computer Science, Carnegie Mellon
Date: Fri, 6 Nov 1992 21:41:57 GMT
Lines: 30

Phil, you seem to have good ideas for the physical world consistency.

The Oz system also uses a link-based system to represent the physical
world.  And you are right -- maintaining consistency is tough.  In
addition to IN and ON, we had PART-OF as well, but that was not as
tough as IN and ON.  The advantage is that once you have the system
set up, extending the set of actions is easier.

Unfortunatly, that was just the beginning, since this didn't seem to
capture notions like doors.  I thought long and hard about how to
implement doors.  If a room (a container object) connects to another
room (another container object) through a door, how are all these
objects related?  We ended up implementing a general notion of
"connection" between any two objects.  These include both action and
sensing connections.  So, doors, phone-lines, telescopes, radios, and
the mail slot in the front door, which I never could get exactly
right.  (Here's a teaser: what can you see and hear if Andre the Giant
has just hung you on a coat-hook on one side of an open door?)

This leads to a general question of how realistic you want your world
to be.  An absolute simulation of the world is impossible, if you
believe the physicists, and presumable anything close is too
computationally expensive.  On the other hand, worlds with only ROOMs,
SOLID_OBJECTs, and PEOPLE might seem a bit limited.  The answer is
clear but unhelpful: do as much as you need to tell you story.
However, if you are building a general system, you have more than one
story to tell.

	Peter Weyhrauch
	
