Author: Mike Hood
Date: 06:00:36 07/01/03
Go up one level in this thread
On July 01, 2003 at 05:41:31, Graham Laight wrote: >When a GM is contemplating a move, he doesn't say to himself, "Hmmmmm. I would >give the resulting position a score of 1.723". > >Such an evaluation is nonsense anyway. There should properly be only 3 >evaluations: > >1. Winning position > >2. Drawing position > >3. Losing position > >It would be nice if a program could work as follows: > >"nb5. This position contains a possible bishop trap". > >"nd5. This puts more pressure on the opponent's king" > >"Opponent classification: bishop trap success rate = 25%" > >"Opponent classification: king attack success rate = 15%" > >"Choice = nb5". > >-g Well, I'm certainly not a grandmaster, but I don't quite agree with you. I don't calculate fractions of pawns in my head, but I do calculate whole pawns. For instance, I might think to myself: "If I make that series of moves I'll gain a pawn advantage, because I can exchange a Knight for a Knight and Pawn". I also think about positions, which is where the fractions in evaluations come from, ie "If I exchange Bishops my opponent's my opponent's King will be left wide open". Or I might even do a combination, like: "I can exchange my Bishop for two of my opponent's Pawns, which leaves me a Pawn down, but I think that the weakening of my opponent's position compensates for my loss". The problem is quite simply that computers don't think the same as humans. Computers don't understand chess, they can only simulate understanding by reducing chess to a series of numbers. Computers have AI instead of RI. "Articifical Intelligence" instead of "Real Intelligence". Positional fractional evaluation is necessitated by the need to shorten the search path. A computer that is able to perform a brute search to an infinite search depth would not need to evaluate positions; material evaluation would be sufficient. So don't be too hard on today's programmers... they're doing a darn good job of creating good algorithms that give reasonable results despite today's hardware limitations.
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.