Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Margin of error in ratings?

Author: Dann Corbit

Date: 14:24:11 06/04/99

Go up one level in this thread


On June 04, 1999 at 11:35:41, walter irvin wrote:
[snip]
>well to be honest with you no matter how many games they play, there will be a
>margin of error,the real margin of error might actually increase after a certain
>number of games on swedish list .the main use on the swedish list is how strong
>a computer is against another computer .
>
>     one of the things that i always wonder about is how all the ratings of
>older chess computers keep dropping until they eventually drop off the list. now
>i know the old computers are not getting old age or bad health.the computers
>play the exact strenth as the day they were created,the swedish system is flawed
>in this respect . but it is the best thing out there for figure out which NEW
>program is best.
Well, it may be flawed, but if so, it is flawed in exactly the same way that
FIDE, USCF, etc. are flawed.  Imagine if you and 20 friends played chess games
all the time and kept score.  Imagine also that one player was head and
shoulders above the other (for the sake of argument we'll say it is you).  Your
ELO rises until it is around 2700 or so because you are a much better player
than the others are.  Now, Kasparov, Karpov, Anand, Kramnik and Tupalov join the
fun of your group and start adding data to the pool.  Will you ELO stay at 2700?



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.