Author: Moritz Berger
Date: 16:02:55 06/30/98
Go up one level in this thread
On June 30, 1998 at 18:36:48, Will Singleton wrote: > >Don et al, > >I think Ed has the right to market his program via this server. I enjoy his >marketing creativity, it enhances interest in chess programming. If his anti-GM >is part-hype, part real, that's fine. > >I'm sure we'll all be more interested in the exact implementation if he has a >good result against Anand. If there's a bad result, we won't care. > >I expect a loss for Rebel against Junior tomorrow, if indeed anti-GM involves >speculative (less than optimum) play. > >Will There's (for me) a good way to test any "anti-GM" engines: Just take a big annotated database and look for moves that e.g. Anand himself found remarkable (! or !! or even more interesting: !?). Some of these moves will be found by all chess programs within seconds (being just difficult for humans or forced by the plan that lead to that position). But others will be just hard to find, taking everything from a couple of minutes to several hours for the programs you test. And maybe you have also a feeling which positions might be hard to evaluate for computers... The strategy then is to compare the results with Anti-GM enabled vs. the results without Anti-GM (Rebel allows to turn that part of its engine on and off and also allows comparison with the Rebel 7,8,9 and Decade engines). By trying several positions that way, you can compare times and get a feeling of the typical kind of position that is solved more quickly by the Anti-GM engine (and maybe also spot some positions where it performs worse, but that would also mean to look for ?! and ? blunder moves etc.). Moritz
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.