Author: Jeffrey Stout
Date: 09:39:00 07/16/00
Go up one level in this thread
On July 16, 2000 at 02:00:25, kurt wrote: >On July 16, 2000 at 01:07:58, Martin Grabriel wrote: > >>Against the Stonewall Attack.... >>why not just get the program to castle Queenside once the Stonewall pattern >>(white pawns at d4, e3 and f4) is detected? i.e. tweak the evaluation function >>to give more points for a Queenside castling once pattern is detected. >>This is probably how a human opponent would play when faced with the Stonewall >>Attack. >>Just a humble opinion. >> >>Yes, this would make a lot more sence than short castling, >> at least delay castling with Nc3 as Crafty,s book likes to play. I play the Stonewall against humans and computers. Actually, humans only castle queenside about 25 to 30% of the time. The most important way to attack the stonewall is something that computers don't do well...plan!!! Computers should focus on keeping the White's King's Knight off of e5, attacking the center, and place the major pieces on the queenside strategically. Humans win about 50% and draw 33% of the time by playing the pawn to d5, the knight to f6, the pawn to c5, the knight to c6, and then the bishop to g4. I find computers take any white stonewall opening sequence with this order will play well and I will gladly test to make sure. Jeffrey Stout
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.