Author: Christophe Theron
Date: 10:16:28 11/10/99
Go up one level in this thread
On November 10, 1999 at 07:57:04, James T. Walker wrote:
>Hello,
>Can someone tell me if Chess Tiger will use the Nalimov tablebases?
>Jim Walker
I regret, but the first release of the REBEL-TIGER will not use tablebases.
The reason for this is that I have priorities. I want to try first the ideas
that can, in my opinion, give the best improvements in playing strength.
It's not that I deny that endgame databases can improve the program, but during
the last months I have tried several ideas that were looking much more promising
in term of playing strength, and implementation of endgame databases has been
delayed.
Tiger is not weak in the endgame, so I think the expected improvement is in the
range 10-20 elo points. Marcus Kästner has done an experiment recently, and
found that the programs that already have a strong endgame benefit very little
from EGTB.
The improvement can be spectacular on some positions, but what I want is a
program that plays well real games. In real games you have plenty of problems to
solve before you reach the late endgame. I'm addressing these problems first...
Christophe
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.