Author: Albert Silver
Date: 11:55:23 02/22/02
Go up one level in this thread
On February 22, 2002 at 14:33:07, Paul Doire wrote:
>If Bobby Fisher at his peak was about 2800 elo and now Kasparov
>is about 2820 elo (+/- 20-30 elo)....then GM's haven't improved
>nearly as much as programs have over the last 20-30 years.
2800 Elo isn't some absolute number, it is a score that says how well that
player is doing against others.
Suppose Fischer were running the 100 meters against several other runners, and
suppose he were faster than all of them. His Elo wouldn't say his time in the
100 meters, it would only say by how much he was beating the others. Now just
suppose Spassky is running at 10.5 seconds (though his rating would also only
say how well he was doing compared to others), well Fischer's rating would say
that he was beating Spassky and others as fast as Spassky by about 0.3 seconds,
without ever saying how fast anyone went.
Now suppose time passes and Kasparov's rating says he too is about 0.3 seconds
faster at the 100 meters than the second best: Kramnik. Does this mean that both
Kasparov and Fischer are running at the same speed? Suppose I tell you that
although the rating can't show it, Kramnik is actually doing the 100 meters in
9.5 seconds.
The above scenario is exactly why you cannot possibly compare ratings of
different periods in time.
Albert
>
>Isn't it quite likely that at the current rate of improvement for
>computer chess (just looking at the last 10 years)...that inevitably
>between advances in hardware and software a computer will be world champion
>caliber and/or World Champion in less than another 10 years?
>
>Even if GM's improve they will not improve at nearly the same pace as computer
>chess .
>
>Paul
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.