Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Further testing of CM_Utzinger

Author: Albert Silver

Date: 18:16:57 04/30/01

Go up one level in this thread


On April 30, 2001 at 19:44:39, John Merlino wrote:

>I continued my testing of the CM_Utzinger personality. If you remember my
>earlier testing, I merely played 100-game matches between two copies of the
>default Chessmaster personality, one using the CM8000 book, and the other using
>the Kostick book (which is the book used by CM_Utzinger). The 200-games tended
>to show that there was little to no difference between these two books (although
>200 games is in no way a large enough sample).
>
>This time, I went in the other direction, testing the default personality
>SETTINGS against the CM_Utzinger SETTINGS, but with both of them using the
>CM8000 book. My theory was that, perhaps, the success of the personality might
>not be due to the combination of settings+book, but rather to only one of those
>two factors.
>
>The results are less then encouraging:
>
>100 G/10 games:
>Chessmaster 55.5, CM_Utzinger 44.5 (+35 =41 -24)
>
>100 G/60 games:
>Chessmaster 52.0, CM_Utzinger 48.0 (+24 =56 -20)

I'd be careful about this. A case in point is this: when testing the two twin
brother engines Tiger and Tiger Gambit, it was quickly noted that although Tiger
Gambit's results against other programs were very encouraging (not to mention
entertaining) as it downed them one after the other, its arch-nemesis was its
very own tamer twin: Tiger. My suggestion would be to do a similar test against
two (at least) other top programs of different playing styles to also compare
the results. They may simply corroborate the results you just presented, but
they could just as easily shed new light. In any case, I honestly believe this
would be necessary to get to the truth on the strength of the settings.

                                       Albert


>
>So, it appears that somehow there may be a magical combination of the
>settings+book that gives CM_Utzinger its power.
>
>Meaning that I am now going to test the full package against the default
>Chessmaster personality. Results by the end of the week....
>
>jm



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.