Author: Uri Blass
Date: 04:48:18 07/22/04
Go up one level in this thread
On July 22, 2004 at 07:43:53, Jorge Pichard wrote: >On July 22, 2004 at 06:16:57, Uri Blass wrote: > >>On July 22, 2004 at 03:47:42, George Sobala wrote: >> >>>On July 22, 2004 at 01:22:51, Sandro Necchi wrote: >>> >>>>On July 21, 2004 at 16:04:02, George Sobala wrote: >>>> >>>>>... I asked, "what would Kasparov do against the same opposition", not "what >>>>>would Shredder 8 do against Kasparov". >>>> >>>>Yes, but that was not the real Shredder, but just the commercial version (one of >>>>them as there is the UCI one which is stronger) and running on a slow computer. >>>> >>>>Sandro >>> >>> >>>Sandro >>> >>>Let me start by explaining that I have *enormous* respect for Shredder as an >>>engine, and for its programming team. >>> >>>As such a friend, I would caution you to be more careful with your choice of >>>words, and you may wish to retract your last comment. Are you really claiming >>>that the engine which you have sold to the world is "not the REAL Shredder"?? >>> >>>You have advertised it on www.chessbase.com as "Shredder 8 – Double World >>>Champion 2003. For most computer chess experts Shredder is the number one >>>choice. Nobody can ignore its amazing five computer chess world championship >>>titles, won at Jakarta 1996, Paderborn 1999, London 2000, Maastricht 2001 and >>>Graz 2003. So in the Shredder 8 box you have the reigning double world champion >>>2003, ready to play and analyse with you." >>> >>>And now you tell us that actually it is not the REAL Shredder in the box! What >>>is that noise I hear? Oh - it is the sound of lawyers hyperventilating with >>>excitement. >>> >>>Furthermore, the above is completely irrelevant to my point. I am sure you can >>>coax extra performance from Shredder with extra tweaks, processors and killer >>>books. So what? It is the commercial version running on a lowly AMD 1200 that >>>has an SSDF score of 2818. The same commercial version (on FASTER hardware) has >>>a TPR less than 2600 in a tournament. >>> >>>SSDF!=ELO >>> >>>Some of us already knew that, but many didn't. >> >>It does not prove nothing. >> >>I think that for a real rating the players should not be allowed to buy the >>program in the first place like they cannot buy human opponents. >> > >Commercial programs are meant to be sold :-) Yes and the point is that you cannot both sell the program and get a correct rating for it. > >>If somebody is interested in the real rating of some chess program then the >>first step is to agree that the the programmer is not allowed to sell the >>program or to release a free version of it and opponents can prepare against the >>program only by watching previous games of the program and not by using the >>program to play or analyze. > >If a programmer wants to find out the real strength of his program in comparison >to strong humans GM, he or the team should test his program against a team of >players like the Argentinian before it is released, not after, and in that case >the weakeness of the program nor its opening responses would not ne anticipated >in advanced. I agree but commercial programmers prefer to sell their program and not to get correct rating for it. 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.