Author: Francesco Di Tolla
Date: 03:28:00 10/31/03
Go up one level in this thread
On October 30, 2003 at 14:32:08, Gian-Carlo Pascutto wrote: >Hi, > >I don't understand why ChessMaster is allowed to use a Kure book (general.ctg) >in the SSDF testing. > >Last time this happened (with Gromit) it was specifically said by Alex Kure >that it was _not_ allowed to use the Fritz books to test anything other >than Fritz. So why is ChessMaster using it now? Did Kure give permission >(which I seriously doubt)? > >IMHO using the ChessBase learner for ChessMaster is also very questionable. > If the original book cannot be converted, my solution to this problem is very simple: 1 - prepare a book for the engine (not letting other companies to have access to it) by building a book with a selection of GM games 2 - disable book learning (if not supported by the engine, which would require own book use also, so always the case if the CB GUI book is used) 3 - use a setting to induce a high spread of the book moves usage (for this engine only), so that the effect of learning of the opponents will not induce a (meaningless) sequence of games in the same line, still opponents can take occasional advantage of book learning notes: - step one could be done by someone independent from any company or SSDF and the step should be repreated for each engine needing a book, building a somewhat different book (to avoid entering in the learned patterns of another previous engine with no book). In this step there should be no (deep) book-cooking selecting moves to be played or avoided, besides may be up to an establised move, which could be say 10 half-moves. - step 3 might look questionable, but to cope with the problem, it looks as the minimal impact I can figure out to avoid giving anybod any advantage. regards Franz
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.