Author: Fernando Villegas
Date: 10:54:01 04/17/00
Go up one level in this thread
Hi Edf: Sorry if I am obsesive with this issue, about which I have posted -and been refuted- many times, but I still think that the answer to an improvement from the high level already reached by programs cannot be the usual trick of pondering the mix of criteria, how much of this o how much of that should be counted in this or that position. I know that there programas -certainly yours- trhat makes big chanmges in his evaluation system according to some general rules about the stage of the game, but I think that is still somewhat coarse. I insist in my idea: the engine should have as his main module an evaluator to determinate what the position is, to which category -beetweeen many- the game has got and only then to arrange the kit of algorythms to evaluate wuth the search. Sure ly this waht an strong player do: he just not say to himself, "hey, we are in an ending, I will take care of passed pawns", but instead he look very deep and see if he is really in an ending, what kind is if it is one, etc. In other words, he try to catch what is happening in a dynamic sense. "Opening", "middle game", etc are just descriptive categories, not fitted to real playing. Could you drive really well a car just remembering you must be cautious? Ok, that's all, forgive my amateur parade... Fernando
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.