Author: Roy Eassa
Date: 08:06:16 10/09/02
Go up one level in this thread
On October 09, 2002 at 11:00:39, Thorsten Czub wrote: >i think your conclusion is wrong. >your mistake is that you don't ask yourself WHY fritz leads the ssdf-list. >Fritz is NOT leading it, because it knows much about chess. > >And it is not WINNING against humans (low ELO) because it knows much about >chess, but because it searches that deep. > >WE judge that Fritz is the strongest, or stronger than others, >because we see the ssdf-results. > >and we believe that a chess programm beating all others is also the strongest. >IMO this is a senseless measuring. > >Because in the case of the match between kramnik and fritz, the lack of fritz >knowledge is confronted with the big chess-knowledge of kramnik. > >its clear that when kramnik can direct the game this way the search depth >of fritz has nothing to do, it can break through with his superior knowledge. > >IMO other chess programs like Rebel, Hiarcs8, Shredder, even oldie Mchess or >others would have played different and better. > > >A human player is always impressed when a computer is good in the field >he himself thinks he is very good. > >so for a chessplayer good in tactics, one should choose a chess program >that is good in tactics too. this will impress the human beeing and make >him problems. > >for a chessplayer good in positional, one must choose a chess program that >plays good in positional. > >We make the mistake to believe that the ssdf delivers us with ABSOLUTE results. >The mistake is to believe that the ELO and the performance fritz gets against >computer-programs is 100% the same as the ELO of Kramnik. > >But this is not the case. > >strength is not strength. > >Elo number is not = Elo number. > >Although shredder and hiarcs8 and others might lose against Fritz in a match >on our autoplayers, they must not be weaker against kramnik than fritz is. > >Because their strength is in the same area than kramniks strength is. >while fritz has a lack here. > >of course shredder and hiarcs do not know as much about positional and chess >than kramnik does, but they run on a fast hardware. > >and would reach IMO a better result than fritz does. > >if you have a sports car, that is made for fast speed on the motorway, >you don't drive with it where no roads are, in the forrest or on the beach. > >if you drive a jeep, you will be better than the sports car on the beach and >where no roads are and in the forrest, but the sports car will be better >on the motorway. > >now WHO is better ?? sports car or jeep ? > >You cannot say this. > >the sports car drives faster, it is fritz. >but the jeep has bigger tires, it has 4-wheels-drive and >is better in the forrest when it is very muddy. > > >the same is with chess programs. > >ssdf measures the motorway strength. there a sports car like fritz >(slim core, less knowledge = small weight of a sports car, much PS) >wins. > >but drive with it into a muddy forret with trees lying in front of you >and the sports car has no chance. > > >Maybe you get the point. >don't be too much disapointed. computerchess is not dying. >its only that chessbase chose the wrong program. The only problem with all that is this: those other programs may be 20% better or even 50% better positionally than Fritz, while Kramnik is 10000% better positionally. Switching programs wouldn't make a significant difference in the face of such a chasm.
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.