Newsgroups: comp.ai.games
Path: cantaloupe.srv.cs.cmu.edu!das-news2.harvard.edu!news2.near.net!howland.reston.ans.net!news.sprintlink.net!psgrain!fizban.solace.mh.se!news.ifm.liu.se!liuida!evalu
From: evalu@ida.liu.se (Eva L-Ragnemalm)
Subject: Re: Some thoughts on game AI: Doom, Descent, Heretic
Message-ID: <1995Mar27.183956.2089@ida.liu.se>
Sender: news@ida.liu.se
Organization: CIS Dept, Univ of Linkoping, Sweden
References: <3k2166$712@netnews.upenn.edu> <DavidKraay.10.2F649770@mgmt.purdue.edu> <jjchoi-1303952056120001@jjchoi.student.harvard.edu> <3k3jg0$64d@news.rwth-aachen.de> <Bryan.Walls-1403951051020001@colin.msfc.nasa.gov>
Date: Mon, 27 Mar 1995 18:39:56 GMT
Lines: 71

Bryan.Walls@msfc.nasa.gov (Bryan Walls) writes:
>In article <3k3jg0$64d@news.rwth-aachen.de>,
>dak@messua.informatik.rwth-aachen.de (David Kastrup) wrote:
>>Those games of the "take thousands of opponents of not much less
>>strength out and win" types *rely* on the fact that the opponents
>>are pretty stupid. If they were not, if they behaved as intelligent
>>as the programmer could design them, you'd lose so fast it would not
>>be any fun.
>>
>One thing to consider is that if you were in a situation where the other
>creatures had some intelligence, they would have to question whether it
>was wise (not to mention moral) to attack you in the first place. A game
>based on such a situation would require negotiation, dealing, limited
>pacts -- well, the sort of things you end up doing in multiplayer games,
>at least.

>-- 
>Bryan Walls                                    My Words are not NASA policy.
>Bryan.Walls@msfc.nasa.gov

Now, this is beginning to sound interesting! Games where you can
negotiate with your opponents - where there may not even be any
designated *opponents* - maybe they're all potential allies?
Now, that's a game I'd like to see!

And, I missed the original post, byt why is all the discussion about
shoot-em-up-type games? (Except for one who mentioned NetHack)

How about adding REAL AI to *adventure-type* games? I'd *love* to
play an adventure like King's Quest and *not* have to repeat every
statement ten times in the search for the one synonym that the
game makers implemented the test for! That'd just require a larger
lexicon and a bit more advanced parser. That's the NLP-branch of REAL AI.

And making game entities react "intelligently" towards the player
(and each other?), like having basic needs and abilities similar
to Brook's robots? The basic needs being the preservation of its
own life, feeding itself, sleeping, etc. Abilities would be movement,
possibly fighting, picking up stuff, putting stuff down - probably
the same abilities you'd give the player... Most importantly it
would have to perceive its environment...

Entities could have memories - if the player is hostile, it could
remember that and either attack if it sees you again, or flee. If
you're kind to it, it might follow you around...

Now, these are things I'd *really* call game AI.

(By the way, somebody mentioned the diffifulty of finding the shortest
path in games like Civilization. If the "like" signifies a grid-based
world, check out the research on Distance Transforms, where algorithms
for Constrained (Euclidean) Distance Transforms have been designed with
decent complexity. MacSokoban (Macintosh version of Sokoban, implemented
by Ingemar Ragnemalm) does use a distance transformation algorithm from
the author's research... Talk about uses for research :-)


	/Eva L. Ragnemalm


<<<<<<<<<<<<<<<<<<<<--------------------------------------->>>>>>>>>>>>>>>>>>>>
Eva L. Ragnemalm
Dept of Computer and Information Science                    Phone: +46-13281947
S-581 83 Linkoping University, Sweden                  Internet: elu@ida.liu.se
<<<<<<<<<<<<<<<<<<<<--------------------------------------->>>>>>>>>>>>>>>>>>>>
--
<<<<<<<<<<<<<<<<<<<<--------------------------------------->>>>>>>>>>>>>>>>>>>>
Eva L. Ragnemalm
Dept of Computer and Information Science                    Phone: +46-13281947
S-581 83 Linkoping University, Sweden                  Internet: elu@ida.liu.se
<<<<<<<<<<<<<<<<<<<<--------------------------------------->>>>>>>>>>>>>>>>>>>>
