Computer Chess Club Archives


Search

Terms

Messages

Subject: What Rules Should be Used to Award a CGM (Computer GM) Title?

Author: Stephen A. Boak

Date: 10:51:23 06/16/01


This is not a debate.  This is a thread to contribute ideas.  :)

Assuming we form an ideal agency for the express purpose of awarding Computer GM
(CGM) Titles, what should the title criteria be?

The rules should ultimately be:
1) Very specific (as free from subjective judgement as possible).
2) Measurable (whether the rules have been fulfilled).
3) Practical to implement (FIDE rules for humans/computers may be considered).
4) Not be tied to FIDE rules regarding Computers (the CGM rules may take into
account the existing FIDE rules, under the Practical to Implement theme, but
should not be officially regulated/limited by any FIDE rule).

[Once we determine a reasonable set of such rules, all we need is a sponsoring
agency to track comp performances and award the title when it is earned!  Maybe
SSDF, ICCM (if that is the proper acronym)....  Doesn't matter--we can address
that later!]

Food for thought:

Should the hardware platform be a factor in the rules?  In the award?

Should the CGM title be 'comparable' to a FIDE GM title?  If so, in what way?

Should a CGM be considered to be of a strength that could hold its own (on the
average) in a competition with FIDE GMs (humans)?

What test(s) must a candidate pass to be considered a CGM?  A Turing test?
Norms-based achievements?  Rating test?  How many?  Against comps or humans or
both?

All ideas on new rating systems are welcome.  Rules that utilize existing human
or comp rating systems are welcome.

Should a World Championship cycle for CGMs produce a comp-comp hero, who could
take on a human World Champion for an interesting match?

Should a World Championship cycle for CGMs involve human GM participants
(because the CGM award should in some manner be relative to human GM strength)?

I think a CGM title system would be very interesting, especially if it meant
that--

"an average CGM (or group of CGMs) could, in general, hold its own in a
competition against an average FIDE GM (or average group of FIDE GMs)"

The philosophical interest is in what rules would lead most computer chess
experts to agree the CGM title really means what it is meant to mean (whatever
that is!).

This concept could be expanded to CIM (Computer International Master) titles,
but how about suggesting CGM rules first.

--Steve









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.