Author: Dann Corbit
Date: 01:30:50 01/13/00
Go up one level in this thread
On January 13, 2000 at 02:37:53, Bert Seifriz wrote: >On January 13, 2000 at 01:29:56, Jouni Uski wrote: > >>As You see from Enriques tournament Fritz is superior to Rebel - so >>Anand is simply correct. >> >>Jouni > >20 games have no statistical relevance, even if you do not believe it. A better statement would be "very little statistical significance." But you are correct in your basic assertion. Imagine the following experiment: "Which is stronger, heads or tails?" We flip a coin. The side that comes up most is stronger. Go ahead and try it -- everybody. Quite surprisingly, most people will *not* get 10/10, even though they really are almost exactly even. Here is my sequence: T - T - H - H - T - H - H - H - H - T - H - H - T - H - T - T - H - H - T - T 9 tails and 11 heads (I used one of those fancy new quarters with a Georgia peach.) Seems like heads is a little stronger. How about some other results? I will not be at all surprised if someone gets 15 of one kind. Yet we know they are about even. It is *exactly* the same with chess programs that are about even. Like, for instance, Fritz and Rebel. Not an opinion. Math, baby. Pure math. Now, one may be stronger than the other. But to prove conclusively would take a staggering number of trials.
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.