Author: Scott Gasch
Date: 09:59:13 12/20/01
Go up one level in this thread
Commercial chess engines are, I think, much more heavily tested than typical amateur engines. That contributes to their strength. The stronger amateur engines like crafty and ferret are, because of the calibre of the authors, very rigorously debugged. This is one of the reasons they compete on the same level as the pros. Commercial engines are not using any "unknown" board representations or search techniques. Perhaps some are using forward pruning techniques that are not published anywhere. The degree to which this affects their playing strength is debatable. I'd be surprised if there was another technique like nullmove out there but it wouldn't be the first time I was wrong... I also think commercial engines tend to focus their evals on important positional features better than most amateurs. This is one result of the superior formal testing these engines enjoy. I, for one, _know_ my eval is lacking in a lot of areas. It's hard to write a good eval. It's even harder to write a good eval that is still fast. Note, I'm probably not the best one to answer this question because I don't pay very much attention to the pro engines. I've thought about buying one or two to test my engine with and try to get ideas from in the past, but so far I haven't done it. The only experience with pros I have is watching my engine fight for its life against them on ICC. Scott On December 20, 2001 at 12:28:45, Russell Reagan wrote: >I held a discussion recently in IRC with one of the founders of the website >www.gamedev.net. We talked about several topics, but I recall him saying that >the techniques used in commercial software will always be ahead of what is >common knowledge and freely available to the general amateur programmer. > >Gamedev.net is mostly concerned with video games, and mostly 3D video games. In >the 3D video game market, this is probably true. Is the same true in the chess >programming market? I know that there are quite a few amateur programs that are >capable of giving today's top commercial programs a good game. Crafty is always >near the top of the pack in tournaments it competes in, and Ferret won the last >CCT ahead of Fritz if I recall correctly. > >My question is twofold. 1) Are commercial programs significantly stronger than >amateur programs today, and 2) are the techniques used in commercial chess >programs vastly different from the techniques used in top amateur programs? In >other words, is there likely to be any alternative (better) board representation >or alternative to alpha-beta that a commercial chess program uses that the >general computer chess programming public isn't aware of? > >Russell
This page took 0.03 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.