Newsgroups: rec.arts.int-fiction
Path: news.duke.edu!newsgate.duke.edu!solaris.cc.vt.edu!news.vt.edu!feed2.onemain.com!feed1.onemain.com!uunet!dca.uu.net!ash.uu.net!world!buzzard
From: buzzard@world.std.com (Sean T Barrett)
Subject: Re: Locational Puzzle Theory
Message-ID: <GKLzK0.3qz@world.std.com>
Date: Wed, 3 Oct 2001 02:43:10 GMT
References: <82675075.0110020840.12d1c469@posting.google.com> <9pcs6l$fan$1@news.panix.com> <9pcsrv$lji$1@news.fsf.net>
Organization: The World Public Access UNIX, Brookline, MA
Lines: 53
Xref: news.duke.edu rec.arts.int-fiction:93119

In article <9pcsrv$lji$1@news.fsf.net>, Adam Thornton <adam@fsf.net> wrote:
>Well, yeah.  That's my basic problem with the model.  The categories are
>so broad that there's no interaction--whether game-related or not--that
>can't be put into it, and therefore it loses most of its practical
>application.

I think there's something to be said for the claim "most adventure
game puzzles involve locations" being a non-trivial claim; I would
(and have) argued that object location is the only deep simulation
in most adventure games; all objects have it, any number of "core"
commands with highly predictable non-vacuous behaviors are defined
for it (go nw, get, drop, put in, empty, empty into), and a fairly
complex text-generation system (for contents lists) is used (but I
suppose the complexity is more due to the difficulty of generating
English output than the complexity of the simulation).

Most other systems have a quite simple simulation: open or closed,
locked or unlocked, worn or not worn, lit or not lit; and the more
complex of those (e.g. both open/closed & locked/unlocked) in fact
do tend to be rather commonly used in puzzles (or at least were in
the puzzle-game heydays, perhaps less so now).

A figure-out-able puzzle needs sufficient context for participants
to find the solution. Systems with deep simulation generally offer
many more opportunities for providing exceptions to the simulation
and for finding ways in which to get emergent behaviors out of the
simulation than something with less simulation which sets up fewer
expectations that can be built on.

Hmmm. A figure-out-able puzzle needs the player to be able to gain
knowledge of the expected behaviors of actions--if you can't predict
that your solution will work, you're just guessing [to a first
approximation, with lots of exceptions]). This suggests to me
three broad classes of ways to expect behavior: based on game-external
knowledge (real-world knowledge), based on hard-coded text output
by the game, and based on behaviors of simulated actions. The
distinction I'm making with the last two is that the hard-coded
stuff tends to be exceptional--each thing with a different behavior.
If things behave "consistently", that tends to be a sim (or it
*could* be a sim, anyway).

So something like Photopia makes its actions draw a lot from
real-world knowledge; your expectation that a given command is
going to work comes from your knowledge of the real world and
some guidance from the author pointing you in the right direction.
Something like Spider & Web's gadgets are a new custom simulation.
An example of the middle (game-text-driven) escapes me right now--
perhaps the pool scene in Photopia goes in that direction, actually--
but I'm pretty sure it's common in the more linear story games.

All of this said from the strongly-biased POV of a simulationist.

SeanB
