From: Mr Noe Dubrovsky (s328940@student.uq.edu.au)
Date: Mon Oct 21 2002 - 11:41:14 MDT
----- Original Message -----
From: Charles Hixson <charleshixsn@earthlink.net>
Date: Tuesday, October 22, 2002 2:47 am
Subject: Re: the brains in bahrain think mainly of hussein
> I seem to remember from back in the 1960's that there was a
> program that
> played an unbeateable game of checkers. Nobody could beat it, not
> the
> world champion. (Was it designed by "Samuels"?) But nobody was
> interested in playing something they had no chance of beating. So
> all
> commercial programs use a less perfect algorithm.
I don't think your story is quite correct there. The Samuel's checkers
program of the '50s was impressive because it learnt its evaluation
function while playing itself. It did play some US state champion and
did quite well, but is not "generally regarded" to have been a brilliant
checkers program. Chinook, otoh, is a giant man-thrashing beast.
>
> Chess is more difficult to program, but even so most commercial
> chess
> programs are stronger than most players.
>Or they can be. They
> come
> with adjustable strengths. Becasue nobody likes to loose all the
> time.
> Go is more difficult than chess, but the programs are starting to
> get
> pretty good there, too.
Chess is not really much harder to program. The algorithms involved are
almost identical. It might be more tedious due to chess's quirky rules
and trickier evaluation, but pretty much identical tricks are used. The
difference is that chess has a much bigger state space so search
algorithms don't get as deep in the time allocated. Go has a much
bigger state space still.
alejandro
This archive was generated by hypermail 2.1.5 : Sat Nov 02 2002 - 09:17:41 MST