Author: José de Jesús García Ruvalcaba
Date: 08:45:00 04/18/99
Go up one level in this thread
On April 17, 1999 at 15:44:42, Rajen Gupta wrote: >I understand that N 99 has a feature that allows it to modify its >evaluation/opening book characteristics depending on the comp opponent it is >playing. does this not mean that a different nimzo(however slightly) is playing >each time the opponent changes. This IMHO an unfair test of Nimzo's capabilities >(ie for the opponent). I would have thought that the fairest way of testing >would be a default setting(which in the programmers opinion is the best, both >for the engine as well as for the opening book) that would remain uniform >irrespective of the comp opponent. > >I wonder whether the current method of testing is considered fair by the others? > >rajen Gupta I see nothing wrong in testing Nimzo 99 with knowledge of the opponent; but I do not like its solution (slightly different engines for three specific opponents). I would prefer to see an option to give the program a pgn file with games of a player, and then let the engine analyze this file and adapt itself automatically for that opponent. That opponent can be computer or human (perhaps the user), even an old time great player (like Capablanca). This is the kind of feature I want to see in future programs. Programmers, are you reading this? José.
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.