Author: Francis Monkman
Date: 14:47:40 06/21/99
Go up one level in this thread
On June 21, 1999 at 13:40:25, Fernando Villegas wrote: >Hi programmers: >How about a table piece where the values of pieces is not generic but specific >and changing? >I know that in some programs piece tables changes according the stage of the >game, but I wonder if they discriminate between the pieces inside each type. I >suppose they changes values for all the category, that is, for the four pieces >each category of figures has or can have, for the eight pawns and for the two >Queens. >What if each specimen of category has its own value according to his position? >Why a knight in a1 must have the same specific valuation that a knight in e5? >Why enemy pieces that are aiming at our King should have same value that one >that is in the original square? It is so dfifficult to give a name to each piece > and evaluate it according his capacities, his position, his mobility and so and >so? To have a particular histry of each one? If that is made, does it not let >evaluate exchanges according to more sound positional criteria? I am tempted to >post some examples from this tournament where bad exchanges were made probably >according some abstract table pieces criteria that said "you are going to win >0,3 points if that exchange is made because bishops are 0,3 point better than >knights" and howerer from a positional point of view was sheer madness because >THAT bisho was awful and THAT knight was better. >Hoy wrong am I? Bob? Bruce? Dan? ?Frank? >Fernando For what it's worth, I've been meaning to suggest this -- based on the fact that GMs analyzing a game often talk about pieces becoming 'bigger' or 'smaller'. We also see that the evaluation of pawn advance is still a weak point for programs. That said, the advances in the last five years have obviously been considerable... Francis
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.