Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Fruit 2.2.1 in CEGT 40/40

Author: Uri Blass

Date: 06:16:54 11/04/05

Go up one level in this thread


On November 04, 2005 at 08:46:07, Heinz van Kempen wrote:

>>
>>Heinz, yes, looking at the ELO performance against the various opponents is
>>fascinating. Against some of the weaker opponents, the ELO performances are
>>huge. So once we get more matches against weaker opponents, and bring the
>>average opponent ELO down to a similar level to Fruit 2.2, then I reckon we may
>>see maybe a 5 point advantage to Fruit 2.2.1 over Fruit 2.2
>
>Hi Ray,
>
>yes,  would be better than nothing and about what most CEGT testers expected
>beforehand.

The difference is smaller than the statistical error so we can say nothing
even if the difference in rating is 5 elo.

I think that it could be expected because possible gain from the 5 piece
tablebases is very small.

potential gain from tablebases is higher for weak program but unfortunately
fruit is not a weak program.

Uri



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.