[A83] Re: chess engine

To: <assembly-83@lists.ticalc.org>
Subject: [A83] Re: chess engine
From: "Shenkuang Francis Huang" <shenkuan@wam.umd.edu>
Date: Sun, 13 Apr 2003 20:43:04 -0400
References: <20030413.193048.-616187.0.matthewwmarshall@juno.com>
Reply-To: assembly-83@lists.ticalc.org


> I would absolutely love a AI chess engine that could make me think!

well here are my AI engine ideas so far:

computer makes mistakes to give beginner a chance + uses opening book
computer attempts to adjust itself to the user's playing level + uses
      opening book, but comes out of book if other side is doing poorly
computer uses pre-existing settings for AI parameters + show thinking +
      opening book
computer uses an AI developed through evolution and automatic
      code generation
Attempt to match a player's style, either mathematically, evolutionary,
      or refining engine parameters.

The thing with winboard-style engines is that each engine would need a
function to generate (pseudo)legal moves, which would add to the amount of
memory the program would require.  I think it's a good idea to have authors
be able to write their own engines though.  I'll have to work out all the
details of how to interface external engines, so maybe I'll study how
winboard does it.  If the above engines that I plan to build in are good
enough for you, maybe I won't consider external engines.

As for the possibility of a strong AI on 83,
http://www.lokasoft.nl/uk/jchess/chessgame.htm gives a demonstration of a
strong chess engine chugging away on a Java VM, which crushed me the first
few times I played it even though it was set to 2 ply.  I asked the company
for permission to use the source code (after glancing at their decompiled
code and seeing how optimized it was) but they wouldn't let me.

References:

[A83] Re: who plays chess?

From: Matthew W Marshall <matthewwmarshall@juno.com>