Author: Mike S.
Date: 20:13:18 04/05/03
Go up one level in this thread
On April 05, 2003 at 17:08:56, Claude Shannon wrote:
>Has anyone ever simply set a computer to play against itself,
>choosing moves alternately for white and black from the same
>search information? At what point during the game would the
>computer make a mistake leading to victory for the other "side"?
Maybe never, in most of the cases (draws).
>Such a game would be useful in evaluating the performance of
>the computer.
I'm not sure how such a game would lead to an evaluation of the computer's
performance... If you think about judgement of the quality of the moves (i.e. by
an GM, or by a very stgrong program), I guess it won't work: There can be great
computer games played by mediocre engines and/or at medium time controls/speeds,
and very bad or boring one's played by top progs at long thinking times. (There
are various other comp benchmark methods available.)
>Also it would be free of bias that might be
>found in a game constructed by a human. Every chess program
>sold ought to come with such a demo game that would be repeatable.
I suggest to take a look into chess program GUIs, studying the usual range of
features they have anyway. Autoplay mode ("comp-comp" mode etc. or what it's
called each) has been common ever since, even in the board chess computer's era.
For example,
http://www.playwitharena.com/
http://www.reubold.onlinehome.de/
http://www.tim-mann.org/chess.html
Note, that different games will be produced by the same engine, depending on the
settings, computer speed etc.
Regards,
M.Scheidl
P.S. You should consider to change your nick IMO. (But I apologize if that's
your real name :o)).
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.