Author: Don Beal
Date: 14:02:35 10/05/04
Go up one level in this thread
On October 05, 2004 at 15:49:00, Gerd Isenberg wrote: >a really nice lever to tune on tactial positions. >New to me and interesting - thanks for sharing! > >Moving twice in a row but only archiving a small positional improvement >considers first move already bad enough to prune it - hmmm. >I guess during game play it is also worth to lose some extra games due to poor >positional play and seeing tactics against itself a bit earlier ;-) Yes, it may be going in the wrong direction. Probably most programmers would do better to strengthen positional play at the expense of tactics. But tactics is more fun! :-) And it will depend on what's in the evaluation and all the myriad other details of a particular program. Anyone who finds their program's playing strength goes up please donate a dollar per ELO to Steve's CCC fund! :-) >What about keeping WACBOOST variable, probably initialized by evaluation or some >eval related heuristics to determine some metric of tactical aspects for both >sides, like en-prised pieces, hanging pieces, pins, possible checks, forks, king >safety, weak as well as passed pawns, etc? It's something that could be tried. Certainly humans would be able to recognise certain positions as standing or falling on tactical swings alone, so perhaps a useful "high-WACBOOST" recogniser could be programmed.
This page took 0.01 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.