Author: Bertil Eklund
Date: 00:24:04 01/25/00
Go up one level in this thread
On January 24, 2000 at 18:33:28, Mike S. wrote: >On January 23, 2000 at 20:00:19, Christophe Theron wrote: >> >>I don't agree. The SSDF does a great job and their testing methodology is OK. >> > >Yes, great job - but what is the *testing methodology* ? I didn't find much on >their website. 40/120 and tournament book, that's all. Is there a detailed set >of rules, i.e. how to deal with doubles, egtb's etc.? I'm afraid it is not. >The question of doubles is of course closely related with the various learning >functions which some programs have. Although those are a strength factor in the >practical use (if working properly), I'm not sure if they should be enforced >when testing. When learning, the program tends to become more and more different >from the initial form in which it is sold. Thus, something is tested at last >which doesn't exist anywhere else. Of course, the influence may be small - but >the elo gaps are small also. >I would like it better, if all permanent learning was disabled for such tests, >so that a program would always remain in it's "just openend the box"-status. >Doubles should be removed then of course. >I'm sure the effectiveness of learning functions could be tested in other ways, >if desired. > >Regards, >M.Scheidl Hi! If I am going to play 20 or 40 different games with McPro for instance, I have to play 500 or 1000 games, if it is possible at all. The meaning with counting doubles is to expand the books and implement learning in the programs. Several years ago mr Irazoqui played a tournament, 20 games for each program but it wasn´t possible to do it with one program because it always played the same openings. We in SSDF have always hated this narrow books, especially before when we hand-played every game. It was terribly boring play programs that repeated the same opening over and over. Bertil SSDF
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.