Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: A logical look at how Dr. Hyatt and others want to cherry pick data

Author: Robert Hyatt

Date: 18:52:35 06/19/01

Go up one level in this thread


On June 19, 2001 at 11:01:30, Mark Young wrote:

>On June 19, 2001 at 10:49:07, Gian-Carlo Pascutto wrote:
>
>>On June 19, 2001 at 10:30:36, Mark Young wrote:
>>
>>>The argument that Dr Hyatt and others are using in the data is skewed because
>>>some of the grandmasters are too old in the ratings list, and these old
>>>grandmasters are no longer playing at a GM Level. This is suppressing the
>>>average GM rating.
>>
>>I think you missed the point of his argument, which is that
>>a GM title is based on the strength a player at a certain time.
>>
>>The strength of that player can change, even worsen, but the
>>GM title will stay.
>>
>>The age of the player is just one cause, but the most obvious one.
>>
>>The GM's below 2500 can be having a bad period, or already be
>>over their top. If they were prodigys they can be well over their
>>top at their 20's already.
>>
>>The point Robert is making is that the GM title is awarded for
>>a peak performance, relative to the entire life of a GM.
>
>This is not correct, and has already been addressed in the prev. posts.
>
>The Title is not awarded for peak performance relative to the entire life of the
>GM. All it take to became a GM is 2500+ ELO and 2 norms. That is it, Why do you
>guys insist that at computer play as an Elite Grandmaster before it is
>considered playing at a normal Grandmaster level, This makes no sense.
>

Because to earn the GM title, a human _must_ play at a 2600+ level for 24-30
games.

If you don't understand the Elo system and the "normal curve" I suppose we can
go into that.  FIDE apparently does... and hence the 2600+ "spike" in the
rating curve to show that a player _really_ is a GM, and not just "lucky" in
a couple of tournaments.  If I were re-doing the FIDE rules for computers I
would put in a "bound" so that the norms have to be earned within N events,
not just over a "lifetime" since a program can play so many events that
statistics _demand_ that even a 2400 program will produce a "norm" here and
there just because of good luck.



>
>>
>>This is why your saying that there are GM's below 2500, hence
>>2500 is a good mark to be a GM, is flawed.
>
>No it is not flawed by cause your statment is not correct.
>
>
>
>>
>>--
>>GCP



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.