Author: Ed Trice
Date: 08:33:39 01/11/04
Go up one level in this thread
Hello Reinhard >I think, the average piece exchange values have less importance as the game >develops and the positional detail evaluation becomes more important. The more >precise the latter would be, the more the influence of average piece values will >vanish. Or - the other side - the simpler a written chess program is, the more >those exchange values will have importance. Therefore it would be not bad to >know, which values will match best, and that could be figures different from >yours and mine. > As a result of watching this game develope, I agree that there is probably a need to change the values slightly for each of us. Vortex "knows" how to minimize the effect of a Chancellor and maximize the effect of the Archbishop in its positional evaluational, which was turned off in the version using just your weights. It would seem the Reinhardt values overestimated the chances with the Chancellor vs. Archbishop, and Vortex overestimated its chances with the Archishop + 2 pawns vs. Rook + Knight. I think the Vortex values are good versus a fallible human player, since it is very difficult to predict where an Archbishop will be just a few plies ahead in such a complex ending. I think the importance of this contest is that it shows there is room for a great deal of development in terms of coming up with accurate piece values. Thanks for sharing your research, this was a good experiement.
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.