Author: Robert Hyatt
Date: 12:05:04 05/04/01
Go up one level in this thread
On May 04, 2001 at 15:02:19, Dan Andersson wrote: >Recognizing pawn levers/breaks is a good thing, then one should have a bonus for >attacking those points. That will make the program play more purposefully in >closed positions. A common theme for programs not aware of breaks, is the >needless shuffling of rooks since the payback for the break will be outside the >search horizon. When a bonus applies, the program will with a higher probability >take the correct position with the rooks and the break will be possible to >calculate as a byproduct of the knowledge. > >Regards Dan Andersson This was one cute DB eval term "potentially open file". Which was a file with a reasonable lever on it, but one which could not be activated within the search depth window due to not seeing deep enough. Hsu (in his book if it ever comes out) points out a couple of places where it affected the move actually played in the Kasparov match.
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.