Author: Robert Hyatt
Date: 17:57:02 02/14/02
Go up one level in this thread
On February 13, 2002 at 16:36:39, James T. Walker wrote: >On February 13, 2002 at 15:38:58, Kurt Utzinger wrote: > >>Black to move. The position is a draw, even if White could manage to win both >>Black pawns. But quite a lot of [top] programs do not at all understand this and >>show completely wrong evaluations. >> >>A good example where a 1500-ELO-player does better than the so called >>2700-ELO-silicon-monsters!! >>Kurt >> >>[D] 8/8/5k1p/6pP/6K1/8/8/3B4 b - - 0 1 > >I believe that playing the position properly (saving the draw) is more important >than the eval shown by the program. I think some programmers don't care if the >score shows that the program thinks it's ahead because it has a bishop for a >pawn. I think it's more important that it gets a draw when the game is >technically a draw. I also see many cases where one program will show a score >of +56.25 where others will show only +10.15. Does it really matter? I also >believe that chess programs do not understand _a_n_y_ positions. They simply do >what they are told and hopefully in most cases it is the right thing to do. The >score is simply a means to arrive at what is hopefully the best move. Every day >I see two progams playing on auto232 where they both think they are ahead and >even when they both think they are behind. They still play better than any 1500 >player I know. If the eval is wrong, the programs will make the wrong move. IE they might go for this position when they find a way to win a pawn, when the original position was actually winnable. And they say "haha, I can win another pawn which is even better..." And they draw a winnable ending as a result. The number _is_ important...
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.