Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Endgame Databases

Author: Bruce Moreland

Date: 09:39:57 09/27/98

Go up one level in this thread



On September 26, 1998 at 10:41:42, Ted Hogan wrote:

>Hello group,
>
>I was reading an article in the USCF magazine by Gabriel Schwartzman GM.  He
>states that if I have a 5 piece end game cd-rom (Ken Thompson) then I could
>never beat the computer because it will play "perfect" end game chess.

This is a common misconception amongst humans.  There are some endings that
cause people trouble, for instance KRP vs KR and KRB vs KR, but there is a
tremendous amount of play in any game before you get to those.

Even endings with seven or eight pieces on the board (this is not a lot, since
two of them are kings), won't necessarily benefit by addition of databases.

One ending that I think might benefit is KRPP vs KR, where the pawns are a- and
c-pawns.  This is normally a very difficult draw.  A computer, especially one
that probes the databases at the tips, might have a good chance playing either
side of this, since it will be able to evaluate pawn sacrifices that lead to won
positions.

But think about what we are talking about here.  You can go look through your
game collections and you'll find games that were decided when material is low,
but there is an awful lot that goes on before you get to a low material
situation, and most games are decided before that.

bruce



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.