Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: More testing of CM_Utzinger

Author: Kurt Utzinger

Date: 11:43:37 05/03/01

Go up one level in this thread


On May 03, 2001 at 13:49:04, John Merlino wrote:

>On May 03, 2001 at 06:26:19, Kurt Utzinger wrote:
>
>>You are right: The CM8_Utzinger is much better at longer time controls.
>>
>>But please again note my already at CCC published objection against playing
>>Chessmaster.CMP's against each other:
>>
>>You will have noticed that I do never play CM8_default vs CM8_Utzinger or
>>similar matches between the various CM8_personalities. It is my deepest
>>conviction that such trials are more or less a waste of time. At first I have
>>made this too but soon recognized that whatever settings are used vs other top
>>programs, the supposed best CM8_setting at a given time control had [like
>>CM8_default] not much chances against Junior6a or most other top programs.
>>Furthermore I have investigated about 150 test positions with six CM8_settings
>>[CM8 default, and 5 others, no Chessmaster.CMP already defined in CM8] only to
>>see that they all were close together. In any case: with these tests it was
>>simply not possible to define the best CM8_setting. And funny, Chessmaster_8000
>>default took second/third place in these tests, the same CM8 which does not
>>succeed in the expected manner against other top programs. All these tests and
>>trials confirmed to me that it is somewhat useless to find the top CM8_setting
>>by let different CM8_personalities playing each other. And frankly spoken, who
>>in the chess world is interested in knowing that [let's say as an example] the
>>CM8_Titan [nothing against you Graham!] wins most of the time against any other
>>Chessmaster 8000 personality. Such a statement does not bring anything if the
>>same CM8_personality does not obtain fairly good results [at least 50 % is
>>required] against other top programs like ChessTiger14, Crafty, Fritz6,
>>GambitTiger2, Gandalf432h, Hiarcs732, Nimzo732, Shredder5 and so on.
>>
>>Kurt
>
>I completely understand your point, but there are two kinds of Chessmaster users
>to consider here: the kind the represent the users of this board, and the other
>95% (or more).
>
>The users here are (typically) owners of multiple chess programs (for various
>reasons). And many of them are interested in finding "the best" one of them all.
>This, of course, leads to people who keep upgrading and testing new versions
>against old versions, etc. etc. etc....
>
>The OTHER 95+% of Chessmaster owners have ONLY Chessmaster (although they may
>have more than one version), and my testing is intended for them. All I'm
>interested in is finding the best OVERALL settings CONFINED to the Chessmaster
>world. If some settings can be found that are clearly equal or stronger over ALL
>time controls and on ALL hardware (which is what the default settings are
>intended to be), then I will definitely sit up and take notice (as I have done
>with CM_Utzinger). But, if someone creates a personality that is fantastic at
>blitz but weak at longer time controls, then I am not interested.
>
>jm

At first glance a logical explanation. As far a blitz is concerned I would not
care much about playing strength. Even a "bad" CM8_personality beats 98 % of all
chess players. And if the same CM8.cmp is good at longer time controls this
might even convince the OTHER 5+% that still have not bought a Chessmaster and
all those who are more intersted in a good program for anylising purposes. Am I
wrong in assuming that the OTHER 95+% of Chessmaster owners are more or less
only interested in using the nice features of the Chessmaster than in playing
strength? On the other hand I do also understand your point of view and maybe
the selling policy of CM proves that I am on the wrong way. If so however, all
the work done to improve the CM8_settings are for me and similar persons rather
useless.

Kurt



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.