Author: Albert Silver
Date: 05:54:08 12/26/04
Go up one level in this thread
On December 25, 2004 at 11:56:36, Gilles Garrigues wrote:
>Every year I buy that big database from ChessBase and every year I find after a
>few minutes some annoying bugs / inconsistencies.
>
>This year I found for example several thousand games played between 1952 and
>1969 (before the introduction of the ELO system) where one of the player HAS an
>ELO ranking ! Probably leftovers from another base with estimated ELO's.
>
>Also when you look at the "short games" (for example less than 5 moves) you will
>realize that many of them are game fragments (incomplete game scores)where it
>would have been better to give no move at all (just the final score in order to
>have complete tournament tables).
>
>Well next year maybe... ?
It's not really unusual, and I should add that the solution may not even be
clear. Why do I say that? Because many players obvioulsy want to use the ELO
rating as a measure of the quality of a line or move when searching through a
tree or using stats. If a line hasn't been played much lately except by 2300
players, and was at its peak in the 60s with the top GMs in the world, would you
want to see the stats reflect the line was weak because the players had no ELO
ratings then. Mind you, I'm presuming the line wasn't refuted, only not played
as the GMs couldn't find an advantage. So what to do then?
Perhaps they should make better use of a GM-IM element, who can say? Though that
would also only bring the limit from 1969 to 1950 when the FIDE titles became
reality. Perhaps add another element then such as a checkbox for "est." for
estimated ratings.
As to errors, it is often a very hard job to fix each and every one. If you find
an error, and don't report it, then do not be surprised if it is repeated in the
next edition. I sent a detailed report with over 200 rating errors to Chessbase
some 5 years ago. This included oddities such as a player with a 2903 rating
(not a blitz rating either) and no less than 30 unknown players with a rating
higher than 2700, and GM Dreev with a 1620 rating. In fact in Big 1999, a number
of GMs were rated 1620 for some strange reason. In the next edition, most of
these were corrected. Still, if it hadn't been reported, who can say how long it
would have taken to catch these.
Albert
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.