Author: Sandro Necchi
Date: 11:10:51 04/04/02
Go up one level in this thread
On April 04, 2002 at 06:53:15, Sandro Necchi wrote: >On April 03, 2002 at 14:26:11, Bertil Eklund wrote: > >>On April 03, 2002 at 01:13:31, Sandro Necchi wrote: >> >>>On April 01, 2002 at 02:03:15, Chessfun wrote: >>> >>>>On April 01, 2002 at 01:55:29, Sandro Necchi wrote: >>>> >>>>>On March 31, 2002 at 16:46:42, liam hearns wrote: >>>>> >>>>>>thanks in advance! >>>>> >>>>>I did not have time enough to test it very much >>>> >>>>>, but I got the impression it is about 15 to 20 points stronger than 6.0 >>>> >>>> >>>>April fool ?. Lack of fantasy? >>>> >>>>Sarah. >>> >>>Well, believe it or not after seeing/analysing something like 20000 computer >>>games and even more human games in more than 20 years and knowing most of the >>>programs which have been available and very many experimenthal verions that have >>>been made since 1976 I usually able to see if a program is stronger and how much >>>in a few games. >>>This is because I know what they do not know and what they and when their >>>evaluation is wrong and why. >>>However before stating something I prefer to have test games results also. >>> >>>Maybe I will not be able to do it in the future, but so far I was able to. >>> >>>Sandro >> >>Hi! >> >>I am sorry but it is absolutely impossible to notice 15-20 elo points >>differences between two programs without several houndred games, maybee if one >>version is x% faster than the previuous (without any other changess) >> >>We see this all the time here when GMs and other "experts" judge a program after >> one or two games. Of course I do not mean that I check some auto232 games and by looking the played moves I can do what I ment. I look the new program version while is playing, the evaluations, the moves which are considered and how long it takes to play the correct move. How is trying to solve the problems of the various positions and how it is able to take the advantage of the opportunity that are available in the games. By comparing this with the previous version it is possible to know if some improvements have been made and how much this should be worth. Also if it is not able to find a solution to some problems etc... Of course all this based also on more than 20 years experience in this field. I hope now people can understand (if they are willing to). Sandro > >Hi Bertil, > >I am sorry too, but it is possible. > >Sandro >> >>Bertil
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.