Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Q: Extrapolate the data

Author: Dann Corbit

Date: 16:06:22 02/09/00

Go up one level in this thread


On February 09, 2000 at 18:27:11, Mike CastaƱuela wrote:
>
>Next sunday is the minimatch (2 games, 1 hr per game)
>of Deep Junior vs. spansih GM Illescas.
>
>he has said that "It has a rating of 2700, a priori,
>I'm must be the victim,, but I trust in the experience gained
>working with Deep Blue-II, it must serve me well now".
>
>ClubKasparov assigns 2676 to DeepJunior based
>in games against humans (rapid and standard games).
>
>Q: If rating of PC programs (DJ as representative here)
>   is aprox. 3000 Elo in 5 mins. Blitz, and 2700 in 1 hr,
>   which is the rating to 2hr/40 moves?
>   (is GM strength, above of 2600, contradicting the
>    point of view of Prof. Hyatt?)
>
>Note: Its question not pretends start steril polemics,
>the question is by curiosity, experts, answer must be
>based in scientific data plus realistic perception.

I think your last phrase is the gist of this.

We already know computers are much better than humans at speed chess.
For longer time controls, there is very little data.  It appears that souped up
computers are approximately 2500 ELO (with a large margin for error.)

The GM quote you supply is just someone's gut feeling.  A very educated person
in that arena, to be sure, but still has no scientific weight at all.  If he had
said instead that "Computers are nowhere near human strength" it would probably
not have been used.  Yet GM's have said things like this from time to time too.
So which is it?  Can we have our cake and eat it too?

I don't know why people want to believe that computers are GM's.  It is clear
that they *WISH* it to be so.  It may or may not be the case.  We simply don't
have enough evidence either way right now.

The best available data is that provided by Rebel and Junior.

I don't think anything gained from FICS or the like has any real validity.
We don't know if the opponent was seriously playing or just trying out new lines
or whatever.  But tournaments where money and reputation are on the line are
definitely excellent indicators.




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.