Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Vincent

Author: Tord Romstad

Date: 00:40:40 05/07/04

Go up one level in this thread


This discussion is rapidly becoming really silly, but I'll add a few words.

On May 06, 2004 at 23:22:49, Arturo Ochoa wrote:
>>>
>>>Move shoud be play against en engine of its level. PLaying a Match against
>>>Crafty, it would not be interesting becase we already know the final expected
>>>result.

Only if by "result" you mean nothing more than the winner of the match.
I don't see why you consider it pointless to match an engine against a
clearly stronger opponent.  I do that all the time (in fact, most of my
test matches are against stronger engines).  Even when I know that my
engine is going to lose, it is still interesting to see if it loses by
a smaller margin than last time, and to study the games in an attempt to
find out *why* it loses.

I think it is best to match an engine against a variety of opponents, some
stronger, some weaker, and some close to the same level.

>>The difference in level is not that big, I think it is less than 100 Elo.
>>
>>Maybe a Movei-Diep match would be more interesting.
>
>
>You will have to forgive me here but... To say that the difference in Elo of
>Crafty and Movei is less than 100 is a nonsense.
>
>Some days ago I had to adopt a nasty behavior because somebody posted a similar
>nonsense in another topic.
>
>Well, if you say that the difference between Crafty and Movei is less than 100,
>I suggest you that you check the Leo D. Tournaments for example.

That's a pretty bad example to bring up, because the number of games is too
small to make any statistically valid conclusions.  It's better to have a
look at Heinz van Kempen's rating list (based on Nunn matches between the
engines):

http://www.husvankempen.de/nunn/ratings/ratingsnunn.htm

The rating difference is bigger than 100, but if we take the error margins
into account it is not entirely impossible that the true difference in
playing strength is less than 100 Elo points.  Dismissing Sune's guess of
a 100-point difference as "nonsense" is going too far, IMHO.

Tord



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.