Author: Bob Durrett
Date: 10:28:39 01/11/04
Go up one level in this thread
On January 11, 2004 at 12:16:36, Ed Trice wrote: >Hello Bob, > > >> >>Anyway, my thinking is that there is room for new ideas in this area although I >>do not know [and cannot know] what the commercial engines do internally. Maybe >>they already do all this? >> > >I can only tell you what The Sniper used to do back in the day of the 16 MHz >68000 microprocessor on my Mac SE. > >The material evaluation could be computed easily and quickly. The problem with >using only material evaluation is that EVERY representation of the same material >in a quiescent position would return the same exact score. > >The problem wth very accurate postional evaluations is that there are more >expensive to compute, and subject to not being 100% accurate. For example, >double pawns are "bad" in general but in the Richter-Rauzer Sicilian, they are >quite necessary. So, you cannot just have some double pawn penality whacking >every such occurance, you need some intelligent filtering, which again requires >more CPU time. > >So what The Sniper did was score the material value quickly, and if was within >some acceptable bound it would fold the more expensive postional score as well. >No sense in checking to see if you have good pawn structure if you are down a >Rook! But wouldn't you miss winning combinations and positional sacrifices that way? Many a game having a position in the game where one side is a rook down still won the game! Perhaps your last sentence is to materialistic? Bob D.
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.