Newsgroups: comp.ai
Path: cantaloupe.srv.cs.cmu.edu!bb3.andrew.cmu.edu!news.sei.cmu.edu!cis.ohio-state.edu!pacific.mps.ohio-state.edu!math.ohio-state.edu!howland.reston.ans.net!swrinde!ihnp4.ucsd.edu!newshub.nosc.mil!news!news
From: mitch@manta.nosc.mil (Ray Mitchell)
Subject: Checkers Board Evaluation Algorithm?
Message-ID: <1994Oct3.204704.5555@nosc.mil>
Sender: news@nosc.mil
Organization: NCCOSC RDT&E Division, San Diego, CA
Date: Mon, 3 Oct 1994 20:47:04 GMT
Lines: 29

Greetings All,

	Please redirect this if it is inappropriate for this newsgroup.
"comp.ai" has been suggested for at least getting steered in the right
direction, however.  I have posted concerning this issue to some of the
various "game" newsgroups from time to time to no avail.

	I have written a checkers game using the standard minimax algorithm
with AB pruning with no problem.  My problem, as is typically everyone's
problem, is once I get to a certain search level and need to evaluate
the board, I have no idea of what is a good board position and what is a
bad position.  I have merely counted pieces and tried a few other arbitrary
things but this has proven totally inadequate.  If I were programming a chess
game I'm sure I could find a lot of information but checkers is another story.

	Although not trivial, it seems that it should be much easier to come
up with a good checkers program compared to developing a good chess program.
I have read many of the standard game references such as Levy's book and the
computer checker work done by Samuel at IBM (in the 50s and 60s I think).  But
most of it seems to vague (or I'm too dense) to actually implement.

	What I am really after, if possible, is simply a good board evaluation
algorithm which I can implement in C.  Esoteric theory and the like are
secondary.  If anyone has any suggestions or can pass this on to anyone who
may I would appreciate it.

Thanks a lot,
Ray Mitchell
mitch@nosc.mil
