Newsgroups: comp.ai.games
Path: cantaloupe.srv.cs.cmu.edu!rochester!udel!gatech!newsfeed.internetmci.com!howland.reston.ans.net!torn!nott!cunews!freenet.carleton.ca!FreeNet.Carleton.CA!bt422
From: bt422@FreeNet.Carleton.CA (Michael Neame)
Subject: Re: Town Simulation
Message-ID: <DGCuIC.645@freenet.carleton.ca>
Sender: bt422@freenet3.carleton.ca (Michael Neame)
Reply-To: bt422@FreeNet.Carleton.CA (Michael Neame)
Organization: The National Capital FreeNet
References: <44rvgm$q2p@rc1.vub.ac.be> <DFwGtw.2wI@freenet.carleton.ca> <44v14u$itv@zeppelin.convex.com> <DFynJH.74p@freenet.carleton.ca> <45jod4$1flq@watnews1.watson.ibm.com>
Date: Thu, 12 Oct 1995 21:32:36 GMT
Lines: 58

Andrew Southwick (asouthwick@vnet.ibm.com) writes:
> Since big plans will more than likely just put off those afraid of four-
> year net projects, how about a *small* simulation to start, to give those
> interested a feel for what needs to be done, what can be put in, how complex
> the programming job will be, etc.
> 
> RARS doesn't have an extremely large amount of contributors, even though
> the interface is fairly easy to get into.  Something else I've learned from
> RARS: do you want your bots to learn to do everything themselves, or will
> you give them enough heuristics to get by in most situations?
> 
> Will the environment the sims work in be small or large?  Will there be
> a great variety of biomes, or just a few?  How many sims do we want to
> start with?
> 
> I suggest starting with:
> a 10x10 map, 4 different land types, 1 sim.
> 
> Define the interface to the sim, what he can know.  *THEN* worry abount
> implementing something thoroughly complicated.  I think it would be
> tremendously valuable to start with a simple introduction into the problem,
> which encourages new coders as well as providing heavy-hitters with an
> easy playground.
> 
> I'm thinking here of proposed RARS additions, including variable-width
> tracks, multiple levels, banked curves, and the like.  It takes quite a
> bit of time to get a car just working on a simple *flat* track--why
> frustrate newbies (and ourselves) with unneeded complexity?
> 
>>willing to follow this up, and I decide I will, do you want me to send you
>>some more detailed thoughts on this? What platform would you prefer? I do
>>IBM's, pretty much, and I'd thought of doing it in C++ (a natural subject
>>for OOP) What platform and language would you prefer?
> 
> The sim should be cross-platform: a generic engine, a non-graphical API
> for sims, and a platform-dependent display module.  Anyone can plug in
> new APIs; sim improvements go into the engine; and ports are done by
> simply replacing the platform-dependent stuff.
> 
> This is a project I know *I* can do.  And we can learn a lot from other
> successful attempts at creating a group testbed for AI.
> 
> BTW, I use a Mac at home and OS/2 at work, and C++ everywhere I can.
> In fact, I think I've decided what I'm going to be working on after
> ER is over and before I go to bed tonight.

Well, it sounds like you have more experience in this than I do. 
Therefore, although I'm willing to work on just about everything, 
If there's one part you'd like me to work on, let me know what, and
I'll start plugging away. I'd like to start on the basics for the 
sim first, design it so it can be easily complexified later. Let
me know!
 TTFN
-Mike




