Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Crafty 17.9

Author: Igor Gorelikov

Date: 07:54:54 06/02/03

Go up one level in this thread


>No
>Crafty17.9 has only slightly better rating in one of your list but Ruffian has
>big advantage relative to that version in another list.
>Uri

You skipped the third rating list (long).

Yes, it is based on a small number of games and still is not reliable.

For reference:
--------------
Time control: 80m+3s (at faster hardware)

1 SOS 11.99 	2721  (24 games)
2 Crafty 17.9   2708  (28 games)
3 Ruffian 1.0.1 2706  (28 games)

And the first rating list (rapid) mentioned by you is not precise due to a
small number of games played by Ruffian (45 games while for precise rating
you need at least 70 games).

What I want to say in sum is that prior to declare that an engine is number
one you need to test it more hardly under the following conditions:

1. All the neighbours (by rating list) must play at least 70 games.
2. Most of its games should be played vs closest neighbours!
3. You should have also a big pool of engines for precise ratings (at least
200) to measure in "millimetres" but not in "meters".

The more precise of my rating lists is the second rating list (Blitz) mentioned
by you because all top engines here played a decent number of games.

And I must repeat the same:

"By my tests, Crafty 17.9 is equal or even better than Ruffian 1.0.1."

If you want I may add the following:

"By my tests, Ruffian 1.0.1 is equal or even better than Crafty 17.9."
;-)

Both sentences are right.
But what of them is "righter" only the further tests may shows but not
words, words, words...


Best regards,
Igor Gorelikov



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.