Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: GrandMaster Standards

Author: Komputer Korner

Date: 11:59:28 03/29/98

Go up one level in this thread


Until Fide institutes a so called GM written test:)))))))))   the rating
formaula and title system based on results is the only fair way. It must
play against human GMs to earn a title. I am not sure if FIDE is
interested in this however, but there will be enough exhibition matches
in the future to give us an indication if micros are strong enough.

On March 29, 1998 at 11:23:18, Mark Young wrote:

>On March 29, 1998 at 11:04:51, Fernando Villegas wrote:
>
>>
>>On March 29, 1998 at 10:32:49, Mark Young wrote:
>>
>>>What standards should be used to judge if a computer program is of
>>>grandmaster strenght? Should we use a rating standard and say if a
>>>computer reaches 2550 elo then its of grandmaster strenght, or do we
>>>need to study how it wins. Does the program need to have the positional
>>>understanding of a grandmaster to be one?  I ask this because the micros
>>>are very close if not already there to meet some of the standards. What
>>>standards should we use to say yes this program is of grandmaster
>>>strenght? I think its telling that a top GM like Anand would even
>>>consider playing Rebel, unless he had some respect for the way the
>>>micros are now playing.
>>>
>>>                                                 Mark Young
>>
>>Hi Mark:
>>I suppose a program or Godzilla, whatever, should be considered of GM
>>strenght level if gets enought elo points beating GM human players, not
>>other computers. Any system of measument only has sense inside the pool
>>of data where it was done. A program can have a 2600 elo measured
>>against other programs and it's OK, but that rating cannot be considered
>>as meaning a GMI standard in human terms. Then, if inside the human pool
>>a program gets the rating, it does not mmatter how, if thorught
>>knowledge or sheer speed. Issues about undesrtanding are confusing and
>>shadowy, elo points gotten in competion are not.
>>fernando
>________________________________________________
>
>I agree 100% it must be play with human players. When I said elo i meant
>Fide rating. I know some people who would argue that even if a computer
>got a 2550 fide rating they would not consider it a GM. I tend to agree
>that the rating is the only thing that should matter. But many people
>who I have respect for make good points in the other direction.



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.