Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: beyond 3000+

Author: Jeroen van Dorp

Date: 10:39:10 10/06/03

Go up one level in this thread


On October 06, 2003 at 08:57:10, Uri Blass wrote:

>It already happened.
>Look at the blitz rating on ICC.
>These numbers have no meaning

These numbers _do_ have meaning, as they do the job they're developed for:
indicating _differences_ in strenght. It even tells you something about
winning/losing chances.

As long as people stay fixed at ratings as indications of absolute strenght they
will tell you that these numbers have no meaning.

> and it is not clear that 2700 fide of today is
> going to be equivalent to 2700 of the future.

It is perfectly clear that the 2700 Fide rating is _not_ going to be equivalent
to Elo 2700 in the future, because ratings mean something different in a
different pool of players.

In _actual_ life these ratings will not fluctuate with hundreds of points
because of a gradual overlap of player pools, but the system is not developed to
compare 2700 players from 1975 with 2700 players from 2015.


>The only way to give meaning to the numbers is to allow a fixed
>software+hardware(it should be not derministic so players will be unable to
>learn a fixed way to win against it) to play when it get a fixed rating and
>other players get rating relative to it but thanks to fide computers are not
>allowed to play so this option is not going to happen.

You mean the only way to give them meaning to _you_, or maybe as soon as is
decided that the Elo numbers indicate an absolute strenght. In that case the Elo
calcualtion algorithm should be different; the current calculation is _not_ able
to calculate absolute numbers, and _not_ designed for it.

J.




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.