Author: Antonio Dieguez
Date: 16:47:52 09/22/02
Go up one level in this thread
On September 22, 2002 at 17:53:27, Mike S. wrote: >On September 22, 2002 at 17:14:46, Fernando Villegas wrote: > >>Absolutely right. In fact those jumps happens precisely after a field has grown >>over-mature with some ideas that until that moment seems to be definitive, >>lasting, uniques and absolute. > >Yes... I wonder, if in computerchess, such an idea could be to **remove the >concept of material values (piece values) completely from chess programs**, and >do the evaluations only by positional/dynamical factors alone. Remove the concept of material values completely? everytime I read that I put a weird face. Well, imho material values, ie good wonderfull and exceptional indicators of piece potential, are very important... if a band is doing a deadly king attack but is a whole piece down there can be a problem, but then the attack must be rewarded more value, etc. > Currently these >two are mixed, with a growing importance of the latter. Also, I think that the >material values are sometimes handled flexible depending on the position (? I'm >not sure about the last point). well for example pawns are very common that its value perse changes depending on the stage of the game. That thingy one could say that is a material value change but it is like yet another eval bonus too. >Things like board coverage or maneuverability >(and other things probably) already represent the material indirectly, in >addition to the piece values themselves. but how about a bishop that is on a1 with bad mobility?, instead of +3.1, how do you regard? you better don't change that +3.1 very much, as after all, the positional bonus in chess programs are just "human" thumb rules (very rough approximation) to evaluate a position. >After all, the average piece values are just a "human" thumb rule (very rough >approximation) to evaluate a position. > >But I don't know if that would be practicable... it certainly sounds very >revolutionary. > >Regards, >M.Scheidl > > >P.S. This has nothing to do with Ruffian which I don't know.
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.