Author: Mogens Larsen
Date: 01:08:16 04/16/01
Go up one level in this thread
On April 16, 2001 at 03:00:24, Chessfun wrote: >I have no problem with a simple autoplay match, as with more engines it's >simply autoplay matches. I have no preference which program plays only >my opinion that of the Deep's the commercial Fritz is strongest. IIRC you wanted to see the strongest program play against Kramnik. The current arrangement doesn't fulfill that requirement by far. It's just the ChessBase SMP Championship. An infinite amount of autoplayer games won't change that. >IMO the problem with a candidate tournament with say 10 engines are the >limitations of finding the best program, as again IMO a 9 round or so tourney >don't really do it. I don't see why not if we look at the facts. In principle there are four different ways of deciding the challenger: 1) Just use the current World Computer Chess Champion. 2) Arrange a candidate tournament with eligible programs. 3) Determine the strongest program. 4) Some kind of intricate method involving one or more of the above. The most time consuming method is finding the strongest program, so that is out of the question. That is unless various more or less creative selection requirements are added like the current situation. As in computer chess, pruning isn't necessarily a good thing. I'm not too fond of the first option, but it can't dismissed as easily. Especially since the World Champion is the strongest tournament program by definition. However, DJ and DF have performed well in human tournaments, but so could others. Arranging a candidate tournament is a reliable format even though the number of games are limited, but there are various options. Ie. two different pools and the winner from each plays a 10-20 game as playoff. Or maybe it could be a knockout format. It's not a real problem. >Plus with other results from the SSDF and other tourneys >played all over the internet there are only a few programs strong enough. You don't have the faintest idea what programs may or may not be strong enough under real match conditions. Peter Berger and Uri Blass have mentioned a few and so have I. Neither the SSDF or tourneys around the world cover most of these programs, nor their full potential. Small example: IIRC Hyatt mentioned that the Crafty code wasn't optimized, because it would make it too complicated to alter continuously. So it could become significantly faster if necessary. That probably goes for other amateur/private programs as well, eg. PConners and Ferret. Furthermore, you're still ignoring that SMP isn't devine knowledge. Others with a proven strength from the SSDF list might have tried if they had a deadline to aim for. >In order to find the best they need to play as many games as possible therefore >the fewer participants the better. It's impossible to determine the strongest program before such an event. Using only two sounds neat, but only if one of the actually is the strongest program. Your opinion about Deep Fritz, even though informed by experience, doesn't change that. >Trying to get all those participants in your candidates tourney to agree to the >basic rules would be a lot of work. Millennium themselves have been talking for >what? 3 months or so. Noone said it had to be easy :-). Just use ordinary tournament rules laid down by ICCA as a template and the reasons for complaints are minimal. Mogens.
This page took 0.01 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.