Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Fritz5 Silences All Critics-Dr. Hyatt included!!

Author: Dave Gomboc

Date: 00:28:29 05/01/99

Go up one level in this thread


On April 30, 1999 at 19:18:00, José de Jesús García Ruvalcaba wrote:

>On April 30, 1999 at 19:07:27, Dann Corbit wrote:
>
>>On April 30, 1999 at 18:48:30, Mark Young wrote:
>>
>>>On April 30, 1999 at 18:27:39, Dann Corbit wrote:
>>>
>>>>A computer will be a GM when they pass exactly the same stringent tests as a
>>>>human GM.  That means performance ratings at a contest and establishing all the
>>>>GM norms, etc.  One victory over a player does not mathematically demonstrate
>>>>anything except the outcome of a single experiment.
>>>>
>>>>However, all that having been said, I suspect that computers are a lot stronger
>>>>than I formally thought they were.  It is not inconceivable that some programs
>>>>on the right hardware are at GM strength.
>>>>
>>>>To extrapolate that caliber from a single match is not repsonsible.
>>>
>>>So when are you sending off your protest letter to FIDE for treating GM programs
>>>as second class chess players. You want the chess programs to pass exactly the
>>>same stringent tests, yet you know this will never happen. I don't see many
>>>invites for computer programs by fide to play in all the tournaments so they can
>>>pass exactly the same stringent tests.
>>Any mathematically equivalent basis would be equally acceptable.  Note: "A GM
>>lost a match to a computer." is not a mathematically equivalent basis.
>>;-)
>
>Hi Dann,
>	I do not see what math has to do with these problems of finding out if the
>current top programs are GM strength. What do you mean by "mathematically
>equivalent" in this context?
>José.

I think he's saying that if a software program plays 25 GM and IM players of
varying rating, and performs at a 2600 performance level, he'll grant that it is
GM strength.  A 25-game match against one opponent isn't good, though: there's
not enough variety in the opposition.   25 games at G/90 aren't good, because
the time control is not proper for FIDE play.  Et cetera.

Dave



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.