Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Gravy for the brain that supports a 2500+ elo standard for computer GM's

Author: Robert Hyatt

Date: 19:06:22 06/19/01

Go up one level in this thread


On June 19, 2001 at 12:15:52, Tapio Huuhka wrote:

>On June 18, 2001 at 17:52:43, Dann Corbit wrote:
>
>>On June 18, 2001 at 17:28:55, Mark Young wrote:
>>
>>>I.      Below is the 11 top countries with the most GM's
>>>
>>>II.     In the Top 11 countries the one with the highest GM average is Ukraine
>>>        at 2545.30 Elo with 37 GM's
>>>
>>>III.    In the Top 11 countries the one with the lowest GM average is
>>>        Yugoslavia at 2479.46 with 41 GM's
>>>
>>>
>>>I think it is important to understand what it means to be an average Grandmaster
>>>when discussing what we mean by GM strength for computers. Some in this forum
>>>seem to insist that computers must perform like the Elite Grandmaster?s. Before
>>>anyone considers them to be playing at Grandmaster strength.
>>>
>>>This shows a lack of understanding in what a normal Grandmaster is, and what the
>>>true strength of a normal or average grandmaster is, and their abilities, and
>>>Elo status.
>>>
>>>Titles      Country        GM  Average rating GM   IM  Average rating IM
>>>1           Russia         119  2545.28            304  2415.07
>>>2           Germany        49   2509.14            146  2401.75
>>>3           United States  48   2507.50             79  2408.19
>>>4           Yugoslavia     41   2479.46             94  2383.39
>>>5           Ukraine        37   2545.30            116  2422.35
>>>6           England        33   2523.91             42  2381.86
>>>7           Israel         32   2535.63             31  2404.19
>>>8           Hungary        28   2491.89             91  2372.02
>>>9           France         22   2539.36             43  2386.12
>>>10          Bulgaria       20   2499.35             44  2366.89
>>>11          Spain          18   2486.28             46  2406.15
>>
>>ELO is not the only feature of what it means to be a GM.  The qualifications may
>>have been relaxed recently, or maybe I'm just remembering incorrectly.  In any
>>case,
>>From:
>>http://handbook.fide.com/handbook.cgi?level=B&level=01&level=01&
>>we have:
>>
>>"1.0. Requirements for the titles designated in 0.31.
>>
>>1.1  Grandmaster: Obtained by achieving any of the following:
>>
>>1.11  Two or more GM results in events covering at least 24 games (30 games
>>without a round robin or Olympiad) and a rating of at least 2500 in the FIDE
>>Rating List current at the time the FIDE Congress considers the application, or
>>within seven years of the first title result being achieved. (See 1.7, 10.10)
>>(GA '93)
>>
>>1.12  Qualification for the Candidates Competition for the World Championship.
>>
>>1.13  One GM result in a FIDE Interzonal tournament.
>>
>>1.14  Winning the Women's World Championship match. (GA '93)
>>
>>1.15  Winner on tiebreak in the World Junior Championship. (GA '93 and EB '99)
>>
>>1.16  A tie for first place in the World Junior Championship is equivalent to
>>one 9-game GM result. (GA '93)
>>
>>1.17  Winner on tiebreak in the Continental Individual or Continental Junior
>>Championship is equivalent to one 9-game GM result. (GA '95 and EB '99) Arab and
>>Centro American - Caribbean Youth events will be treated as Continental
>>Championships. (GA '93)
>>
>>1.18  Clear first place in the Women's Candidates Tournament is equivalent to
>>one 9-game GM result. (GA '93)
>>
>>1.19  One 13-game GM result in the Olympiad will lead to the award of the full
>>title. (GA '93).
>>
>>1.20  Winner of the World Senior Championship (GA 97)"
>>
>>
>>Especially interesting is 1.13.  If a chess program has had a GM result in an
>>interzonal, then it is a GM!
>>
>>Has any such match taken place?
>
>Now that we are talking FIDE rules, I wonder why it's still conveniently
>forgotten that opening libraries and endgame tablebases do not agree with FIDE
>rules (conduct of players). Human players are prohibited from using books during
>play.


This argument is flawed _and_ pointless.  It is flawed because as a human, I
memorize openings by reading them, studying them, and playing them.  I know
players that can read a game one time, and play it back move for move 5 years
later.  "Kolty" (for those that had the pleasure of meeting him) could do this
in a heartbeat, and for anyone that saw his impossible knight's tour demos
naming each square as directed by the audience and then using those 64 names
to do the tour, was amazing.

He obviously had a great memory.

So does a computer.  The computer does not use a "book".  It uses something
stored "internal" to the machine, just like a human.

The argument is pointless because when you look at a chess program, that is
nothing but "written computer instructions", the computer "moves" pieces on
its internal electronic board, it uses stored (written) patterns in the
evaluation, etc.  Basically a computer and human are different.  The human
has some capabilities that the computer doesn't, which I could claim is "unfair"
too. :)




>
>Maybe this has something to do with the fact that computers can't compete in
>FIDE events and that many players don't want to play against computers. I think
>it's not reasonable to compare human players with computers. Look what happened
>to Kasparov: in the prefece of MCO de Firmian tells that much of the success of
>Deep Blue was due to their good opening preparation (so Kasparov was outprepared
>that time, too :)


Computers _can_ compete... but the registration fee is extremely large.  A
resolution allowing this was passed several years ago (It might well have
been rescinded but I have not heard that.)  At the time it was passed, most
of us thought "that is a ridiculous amount of money to require..."






>
>What would be the rating of top programs without opening libraries and endgame
>tablebases?
>
>Tapio



This page took 0.01 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.