Author: Marc van Hal
Date: 10:10:46 01/10/03
Go up one level in this thread
On January 09, 2003 at 08:44:30, Uri Blass wrote: >On January 09, 2003 at 08:14:09, Russell Reagan wrote: > >>On January 09, 2003 at 06:00:48, Ed Schröder wrote: >> >>>On January 09, 2003 at 04:58:53, David Rasmussen wrote: >>> >>>>Will you? >>> >>>No, I am not. The last years I moved to a new way of testing, the idea comes >>>from the brain from Christophe, thus protected. >>> >>>Ed >> >>Without revealing anything, just tell us, how much have you learned from him >>over the years? Would you say it is a significant amount? It sounds like the >>relationships between the authors of the top commercial engines is interesting >>to say the least. >> >>Russell > >I do not know how much but only in the last year Ed used games to find that a >Rebel personality of Thorsten performed better. > >There are 2 possibilities: >1)Ed did not trust the testing methodology that he has and also used games. >2)The testing methodolgy that Ed is using did not tell him that Thorsten >personality is better when games told him that(it is possible if the way of >testing give wrong results in minority of the cases or is not relevant for every >change). > >Uri Ok this would be my methode for testing Well I guess the first testing methode was the epd diagrams Later on the Knaak positions. And later on games Rebel or Tiger versus the Chessbase products . I think the last is not a real good testing methode for real chess playing strenght . But more for computer computer results. I would sugest Chessmaster and it's strongest personeletys instead I would put Schredder as the next in the row. then Human computer games (preferably advanced) And only then the other programs. (Not that the last are worse but they probely will play more moves which Rebel or Tiger expect. So not much will be added to learning.) An other intresting methode could be the positions just before a stunning combo comes on the board played by humans. Will the program play to that combo? Marc
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.