Author: Christophe Theron
Date: 13:05:04 04/09/00
Go up one level in this thread
On April 08, 2000 at 19:53:41, stuart taylor wrote:
> So does junior 6 have the greatest knowledge or not? How did it fall so
>much? If you would take its recent results without mixing them with the old,
>then it would be even worse! how do the programs fluctuate so much from one
>2-month period to the next? maybe some technical reasons? or maybe fritz 6
>had technical problems which are now corrected? I mean fritz 6 was behind junior
>on last ss list also.
Read the SSDF list again. Fritz6 has never been on the SSDF list before.
Further, it's natural that ratings of programs fluctuate. If you read the list
carefully, you notice that ratings are published with an error margin (the "+"
and "-" columns). It's because of the nature of the ELO calculations. Unless you
play an infinite number of games, you cannot be 100% sure of the rating you get.
Instead, you get a rating AND an error margin.
The only thing you can say about Junior6 by reading the current list, is that we
are 95% sure that it is rated between 2659 and 2720 ELO. The value 2689 in
itself is MEANINGLESS.
The playing strength of the programs do not fluctuate. But the measure of this
playing strength DOES fluctuate, because it becomes more and more accurate as
more games are played.
It is very important to read the error margin when you read the SSDF list. Or
else, you'll miss a very important part of the information.
Christophe
> I wish I knew better where programs are holding.
>S.Taylor
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.