Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: GM's are determined by meeting FIDE TITLE criteria not by rating

Author: Sune Larsson

Date: 05:24:17 06/16/01

Go up one level in this thread


On June 16, 2001 at 07:01:13, Chris Carson wrote:

>On June 16, 2001 at 06:14:47, Bill Gletsos wrote:
>
>>If you want to claim a computer is a GM let the computer earn the GM title just
>>like any human would have to.
>>
>>Clearly a player is a GM because they meet the necessary FIDE Title criteria not
>>because they have a 2500 rating. GM's didnt get their titles because of their
>>rating but due to meeting a set of criteria that established them as being of as
>>some of you would call it "GM strength". In general this criteria requires them
>>to get 2 or more GM "norms" in events covering at least 24 games(30 games
>>without a round robin or Olympiad) and a rating of at least 2500(within 7 years
>>of acheiving the first GM norm). These events have to be valid Title events.
>>
>>There are 548 players over 2500 on the latest FIDE list with another 10 being
>>rated 2500.
>>
>>There are many players on this list rated over 2500 who are not GM's but only
>>IM's. Some are only even FM's and some have no FIDE title at all, although it
>>should be noted that the majority of this latter group are from Myanmar.
>
>This would be great if FIDE let computers compete in FIDE rated events and would
>give titles to programs.  FIDE has banned all computers from FIDE rated events.
>Events can not be rated if a program participates, thus no program will ever get
>the GM norm.  This is why this debate rages and will continue to rage.  Even
>before the ban last year, FIDE would never give a GM norm to a program.
>
>I personally do not care if a program gets the GM title from FIDE.  I am only
>interested in "playing strength against humans and other programs".  Programs
>have given performances above 2600 (2642 over 49 games in a 3 year period,
>opponents rated 2548) and over 2500 (2525 over 232 games spaning a 3 year
>period).  All games at 40/2 and against FIDE raated players.  FIDE says that GM
>strength is 2500, the programs are playing above 2600 on a consistant bases on
>the fastes hardware.



>
>I do think that the opportunity is there is some organization (SSDF, Braingames,
>...) to grant program titles based on human game performance.  Hire a few GM's
>to play the number of games against the programs (keep the identity of the
>program a secret until after the tournament).  Charge a fee to the program
>company, if the program gets the norms and rating needed, then it gets the
>title, same a the human, just a different organization.  This would also
>generate some interest for a continued GM vs DGM (Digital GM) matches, at least
>until the computers are mostly invincible.  Match play this the best chance for
>humans to beat the machines, this way the human can learn the weakness and play
>on that.  Tournaments are needed to establish a rating.


 This sounds interesting to me. It looks possible to arrange such games
 for programs. Following the FIDE standards, IOGM (Inofficial Grand Master)
 and IOIM (Inofficial International Master)- titles could be achieved by
 chess programs. Then the results would speak for themselves.
 Another issue is that the performance of a specific program can differ
 tremendously in a sample of important games vs human title holders.
 Progs play a different kind of chess than humans do. Much weaker in some
 areas - stronger in others. Sometimes patzer - sometimes genius.
 It's up to the humans trying to exploit this. And ,as said, the results
 and the achieved rating would then speak for themselves.

 Sune







>
>Some people make a valid point that the qulaity of the games (anti-computer
>strategy) can make the computer look like a 2100 player.
>
>For me, results mean more, just my opinion.  You are ofcourse entitled to yours.
> :)
>
>Best Regards,
>Chris Carson



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.