Author: Uri Blass
Date: 09:29:07 02/16/03
Go up one level in this thread
On February 16, 2003 at 11:56:20, Robert Hyatt wrote: >On February 16, 2003 at 11:01:49, Roy Brunjes wrote: > >> >>In a game between Fritz 8 and a custom personality of mine for CM9000, the >>following position arose in a 15/10 game: >> >>[D] 5Q2/8/2p4k/1p1pP3/8/6KP/8/q7 b - - 0 45 >> >>Black here should easily draw with ... Kh7. Fritz 8 thinks for almost 30 >>seconds on my Athlon 700 MHz machine and plays ... Kh5 and scores the position >>as 0.00 expecting a draw. >> >>CM9000 immediately shows evals nearly one pawn in its favor and after a few more >>moves, Fritz shows an eval of more than 4 pawns in White's favor. >> >>Curious that Fritz would err so greatly in a simple position. Is it a bug or is >>it explainable some other way? >> >>Here is the entire game for those that may think the game history and draw >>detection may have something of a role in this 15/10 game: > >Most likely it is just a lack of time. Finding that Kh5 is trouble requires >at least several seconds for Crafty, while Kg6 is scored as a draw. I think that it is simply lack of knowledge. Finding that the white pawn is unstoppable after the next moves(46.Qe8+ Kg5 47.h4+ Kf5 48.Qf7+ Ke4) is easy The problem is that Crafty does not see that the following position is better for white by static evaluation. [D]8/5Q2/2p5/1p1pP3/4k2P/6K1/8/q7 w - - 0 5 Movei also does not see it but for different reasons and it simply does not evaluate pawn race even in pawn endgames. Uri
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.