Author: George Tsavdaris
Date: 13:16:09 06/06/04
Go up one level in this thread
On June 06, 2004 at 14:19:10, Uri Blass wrote: >On June 06, 2004 at 13:29:32, eric guttenberg wrote: > >>I assume SSDF tests Chessmaster 9000 at its default settings. > >based on my memory they increased the selectivity of the default setting. > > > The >>information I get from this site and other sources is that some settings >>other than default are probably stronger. Notably the SKR settings >>which have been tested by various people fairly extensively and with >>results consistently better than the default settings achieve. > >The skr setting was tested by the ssdf with some old book of Fritz and learning >of chessbase but after some programmers like GCP complained about it >they decided not to put it in the list. > >That setting got a 2742 TPR after 165 games when chessmaster9000 has 2720. > >I believe that the learning capabilities of the chessbase interface worth more >than 22 elo points so I guess that one of the following is correct; > >1)The SKR setting is worse than the setting that was tested by the ssdf >2)The old Fritz book that SKR used was inferior book relative to the commercial >book > >Of course it is possible that neither of these claims are correct and SKR was >simply unlucky. I think both 1) and 2) are wrong and also that SKR wasn't unlucky, but the thing actually happened is that SKR played only with 4 opponents while Default with 10, so if both "versions" played the same number of games i strongly believe that SKR would have 55 or more ELO points more. Of course we have to think margin of error also, so the difference would not be 55 but will be in an interval form of [10,90] but anyway....... > >Uri
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.