Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: What ELO is perfect chess?

Author: Dann Corbit

Date: 13:48:58 02/22/02

Go up one level in this thread


On February 22, 2002 at 15:22:00, José Carlos wrote:

>On February 22, 2002 at 15:13:25, William H Rogers wrote:
>
>>Now I am currious. Is Crafty really rated that high or was its opponents not
>>rated low enough? This is not a slur to Dr Hyatts Crafty, we all know that it
>>plays great chess, but maybe instead of raising a players rating, they should
>>consider lowering an opponents rating as to not exceed the possible max(i.e.
>>3300). I think that would bring a much more resonable response to all players,
>>except of course, those whose ratings were forced to be lower.
>>Food for thought anyway.
>>Bill
>
>  The answer is that ELO is only a _relative_ measure of results. ELO tells you
>"how player A performs compared to a given pool of players". There's no up or
>low limit, it's just a comparison. For example, in my private tests, I always
>set Averno 0.32 rating to zero, because I'm interested in know the advance in my
>newer versions compared to that. So, in my list, Crafty is 600-700 ELO.
>  ICC is _a different pool_ and so ratings are not comparable in absolute terms.
>But they are in relative terms. For example, if Averno 0.32 and Crafty play on
>ICC, Crafty will have a lot more ELO than Averno. Let's say 3300 for Crafty and
>2500 for Averno.

Your zero line is actually optimal, since the precision of calculations will be
higher with that baseline than with any other.
;-)

Not a property of the mathematics itself, but (rather) a product of floating
point math on a computer which is really preformed on a subset of the rationals.
[Much to the surprise of many]




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.