Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Programmer vs Program strength

Author: Dan Homan

Date: 08:14:24 06/05/98

Go up one level in this thread


On June 03, 1998 at 17:05:32, odell hall wrote:

>I am curious do programmers know how to defeat there own programs. It
>would seem reasonable since they would have specific knowledge about the
>strengths and weakness of their own programs. For instance i would love
>schroeder to tell me the best way to beat rebel. Or is this question
>rediculous?? Can I please get some comments on this from various
>programmers

My program is about 18 months old.  For the first 6 it could not
beat me ever... even if I hardly thought.  For the next 3-4 it would
still lose, unless I blundered, but I needed to think to win.
Over the last 8-9 months it has improved considerably.  Now I can never
win at fast time controls... I have not tried it at slow time controls
in a very long time.

My strategies for beating my program (in the first 9-10 months of its
development) were to wait to the end game and push my pawns.  It didn't
know enough about passed pawns to defend properly.  Now it does, but
my pawn eval isn't perfect, by any stretch.

Recently, I observed that strong players (stronger than me, anyway)
attack the king to win.  I stopped a good deal of this by putting in
some evaluation for open files near the king, but some players
make excellent use of bishops and knights (and later the queen) to
strip the king's defenses and give my program serious trouble.

I will try to patch this hole in the future, probably not for some
time though, as I need to finish my research/thesis this year.

 - Dan



This page took 0.01 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.