Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Sorry Rolf - the winner is the winner.

Author: David Dory

Date: 19:44:16 09/13/02

Go up one level in this thread


On September 13, 2002 at 18:39:40, martin fierz wrote:

>On September 13, 2002 at 15:45:05, Rolf Tueschen wrote:
>
>>I'm sorry but I cannot agree! Would you think taht Elo is a universal utility
>>for calculating the chess performance of humans alone or humans and apes and
>>flowers? Remind you of the apples and beans theorem. It doesn't make sense to
>>make a ranking with the Elo for totally different species. MEPHISTO ROMA has
>>nothing to do with JUNIOR 7. Besides that they both play chess. But where is the
>>point for comparisons?
>
>i've played a tournament games against a GMs rated 2697.i've often played
>first-round games in opens against beginners with a 1600 rating. it all counts
>for my rating... why do you need to make up special rules for computer ratings
>if we don't need it for human ratings?
>
>aloha
>  martin

Hi martin,
  To establish an elo, or a winner at a tournament, equal hardware is certainly
not required. Only wins, losses and draws, of course.

What Rolf and I object to is testing/competing with a chess program on vastly
different hardware, and then saying - "Yep that MyChess program has an elo of
only 1242 - just lousy." (or whatever results are thus claimed, good or bad.)

When in fact the program was run on a 486 at 50Mhz against competitors on
Athlons at 1.6Ghz, or some such. I'm sure I'm exaggerating, but it is atrocious.

Such a "tournament" lacks meaning to me. It's like having a foot race but some
racers have deep sand in their lane while others have the finest track material
made. Now who's really the fastest runner? Running the race repeatedly in the
same lane's doesn't improve the disparity of the race conditions. If anything it
makes matters worse.

Nothing wrong with some legacy competitors using their original hardware indeed,
this may be necessary in the case of dedicated chess playing computers.

All software should be migrated up to an equal (or as close to equal as
possible) hardware footing when competing with the new releases, however.

My pet peeve #128 :)

David






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.