Author: KarinsDad
Date: 09:28:07 07/12/00
Go up one level in this thread
On July 12, 2000 at 10:44:11, Daniel Chancey wrote: >The main concern is the # of times the comptuer lets itself into getting >hammered by a raging attack. So many times the computer ignores it, eats a pawn >and dies. (Like the trap used to kill termites) > >Here are my suggestions that hopefully could make a much better computer and put >them in near Super GM status: > >1.Count # of squares near own king, that the opponent is attacking into the >evaluation. My evaluation routine gives a score for given squares. At the moment, it does not use this to take into account important squares based on the root position, but I hope to get that in one day. For now, it just gives a slight bonus to the 4 center squares and every square next to each king. It also gives a little lower bonus to the 12 squares around the center squares and all squares 2 squares away from each king. Most other programs do other things for king safety, but I think that these are probably different than mine. I think that king safety is almost a misnomer. Either the opponent has a threat on your king that will gain him something in a given position, or he does not. But, for the most part, that will be determined in the search engine and not in the evaluation routine. The most the evaluation routine can do is attempt to steer the game away from positions where there is potential danger to the king. It cannot necessarily prevent the program from waltzing down a path where danger looms beyond the event horizon. It can only attempt to do so. KarinsDad :)
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.