Author: José Carlos
Date: 08:57:04 10/16/01
Go up one level in this thread
On October 16, 2001 at 08:57:39, Uri Blass wrote: >On October 16, 2001 at 08:38:48, Steve Maughan wrote: > >>Steffen, >> >>Goliath 1.5 instantly says it's a draw. Goliath is good at these type of >>positions as it has some sort of speculative draw heuristic which sometimes >>works and sometimes doesn't - in this case it works well. I assume, but don't >>know for sure, that it works something like - "if the last six moves (12 ply) >>were all checks and no captures then assume that it's a draw" > >It means that goliath guess draw score and does not ee it. >Goliath can be wrong in other cases so goliath results is totally irrelevant. If a program/human evaluates a pawn in seventh as very dangerous, can be right sometimes and wrong others. If a program/human evaluates a enemy queen around the king as a danger can be right sometimes and wrong others. If a program/human evaluates two pawns in e4 and d4 as a strong center, can be right sometimes and wrong others. So we conclude that anything different from checkmate or draw is irrelevant. I believe this can't be correct. Any such assumption can be good or bad, depending on _how often_ it is correct or not. There're programmers that like speculative attacks. Some of them are succesful, some not. We don't say specutative attacks are irrelevant because of that. The important thing is try to be right as often as possible, even if the program looks stupid sometimes. Human players are not much different. There's Tal and there's Petrossian. There's Alekhine and there's Capablanca. There's Kasparov and there's Karpov. Different approaches to chess. Different criteria. Different decisions. The one that is right more often is the strongest. José C. >I knew that goliath see the draw in 0 seconds but did not think that it is >important to post it. > >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.