Author: Nelson Hernandez
Date: 06:00:31 11/20/05
Go up one level in this thread
What you propose would only work if every participant had identical hardware, too. And if the opening books were all given the same internal settings. And if book customization was banned. If you did all that then the tournament would be "fair", assuming of course all these rules could be enforced. But then luck would play a very large role because a program might be unlucky and be put in a disadvantageous position repeatedly after exiting book in a series of games. It seems to me that depriving engine developers the ability to outwit their opponents in the opening is a lot like asking two boxers to fight with one hand tied behind their backs. It may be a fair fight, but it would be unnatural. These tournaments are inherently too short to prove anything in statistical terms. If you want to know which engine is strongest, follow the CEGT rankings. Tournaments are a battle of computer software (that's both engine AND book) wherein hardware plays a sometimes important role. It is a test of cunning and wits and, I imagine, a lot of fun for the participants. You want to take away the fun and create a sterile, scientific playing environment when 9, 11 or any practical number of tournament games could not possibly deliver a definitive result as to which program is strongest. Now, why engine developers hype their wins in these tournaments is a separate topic. If you were them, what would you rather promote: a strong tournament result which everyone can understand, or an arcane statistical statement, such as "we currently have the highest ELO with an xx% degree of confidence based on yyyy test games"? The latter simply doesn't sell, except to statisticians.
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.