Author: Robert Pawlak
Date: 18:48:30 08/22/01
Go up one level in this thread
On August 22, 2001 at 21:00:25, Charles Young wrote: >The ICCA tournament is not statistically significant as not enough games are >played. But there are eighteen programs participating using swiss pairing, so >we "can" gauge the strength of programs. > Ok, using this same logic, let's take 18 people off the street and throw them in a nine round swiss tournament. Then let's use that ONE tournament to infer that player A is better than player B, where both players place about in the middle of the pack. Also, let's draw these conclusions even before the tournament is complete :-) See the problem here? There are not enough samples to infer program X is better than program Y (especially in this case, where not much separates these two programs). Furthermore, look at the relatively small number of different openings played. Maybe program X plays the Ruy better than program Y, but the situation might be entirely different for some other opening variation. >If we use Crafty as a program strength baseline, in my opinion, Ferret is not as >strong as Crafty. I believe that Ferret is a relatively weak compared to the >professional computer chess programs. Ok, everyone is entitled to an opinion. My opinion is different. Go look at the computer accounts on ICC, finger ferret and crafty and still tell me that your assertion holds. And besides, this argument has been rehashed an incredible number of times. I'm sure others have made these points before. Bob
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.