Author: Peter Schreiner
Date: 09:28:00 02/28/98
Go up one level in this thread
Subject: New SSDF-List - especially for Thorsten From: Karsten Bauermeister E-mail: caratan@uni-muenster.de Message Number: 15244 Date: February 24, 1998 at 20:31:44 >Everytime the same thing...! When a program comes out and plays not the >style Thorsten (Czub) likes, he asumed it in a very awful and unfair >style. >Of course there were people, they had results like in Sweden. Of course >there were people, they told you, that Fritz 4 is stronger than the >rating in Sweden. There were very interesting and thrustfuly opinions, >which supposed, there were faults in playing thr first games of Fritz 4. >The fact, that later the results were much better, than the first ones >is a strong "Advice" to that. >But the fact is, Thorsten always "do not believe" anyone, who has a >different oppinion from yours. >I never played tournament games with Fritz 5, because I have only one >fast computer, but the games aginst my own person and other that I have >seen, were always strong played - what Thorsten don't like is the style >of playing. Perhaps some day the machines will show us, how to play >chess. Perhaps the nice looking style of grandmaster chess is really bad >compared to the (nearly) perfect style. Interesting. But why did you tell us your impressions NOW, after the new SSDF-List is published ? In the past you never told us something about the strength of Fritz 5. To make it clear : I have nothing against Fritz 5 and I think, that it is a strong program. But naturally I was absolutly surprised like many others about the performance in the SSDF. Some people, for example Moritz Berger, told us it again and again in the past, that Fritz will receive a lot of benefit from hashtables for it `s strength. Seems that he`s right. But now I see to my surprise a lot of people, which now said : For me it`s clear, I always had this knowledge. Why didn`t you tell us your impressions, before the new list was present ? >What everyone should know about Thorstens way of testing, is to break up >games, he doesn't like or he _*thinks*_ they are decided. Dear Karsten, is this okay ? I dislike this presentation very much. Nobody is without errors. I remember very well the WM in Paderborn 1995 ....Should we eventually discuss here your work in operating Hiarcs there ? >What he doesn't like is not Fritz, but there are two "fast and dump" >searchers leading the swedish list and Fritz is the kommercial product >of ChessBase, the firm with the "high-price-politic". >My opinion was and is: Fritz 5 is much stronger, than Fritz 4: More >knowledge in every part of the game, faster, really deep looking etc. >I ever believed that Fritz would have >2500 on the Pentium 200 MMX and I >think, that it will fall a little bit. Most of the former leading >programs (The King 2.0; Rebel 8; M-Chess Pro 5; Hiarcs 6) rise down some >points some months later. >I gratulate the guys (especially Frans Morsch) of ChessBase for there >work and of course I use it for my own analysis. So were Matias Feist >very surprised, that I used Fritz 5 in Paris to analyse a game of Comet! >More, I don't believe the result in Sweden is the result of the >advantage Fritz 5 is using bigger hash tables and power books. Thorsten believes something, you don`t believe something ....? Again : for me it`s a little bit suspect, that you `re claiming this, after the new list is published. People like Moritz claimed this for a long time, this has for me more credibility. To make it clear : I`m surprised about the performance of Fritz 5, but it`s not a great problem for me. You `re a well known expert, for example cowriter in CSS. Didn`t you recognize the real basic problem ? Let `s take a look to some facts : The published list of the SSDF is a trusted source of information for many years. The basic principle of the SSDF was, that chess programs and computers were tested ONLY in EXACTLY THE SAME FORM, as they were AVAILABLE in the public. This basic principle is the argument, which made the SSDF ratings VALUABLE for all interested people. One additional aspect is the fact, that the SSDF is non-commercial. Okay until now ? Now, in my knowledge for the first time, the SSDF allowed a producer of chess software, to participate with a SPECIAL hardware/software configuration, which is in difference to MChess, Rebel,Hiarcs, Genius etc. NOT COMMERCIALLY available !! So for me it`s absolutly uninteresting, what you or anybody else believe. I see there some danger, that in the future the SSDF rating list will become WORTHLESS , unless they will return to their basic principles. Looking to the SSDF-Ratinglist from the 22.02.98 I want to discuss the following points with my friends here in CCC : 1.)In difference to ALL other programs in the list Fritz 5 have been tested with a special Powerbook, which is normal NOT delivered with Fritz 5 ! Some guys told us here -- I think it was Moritz -- that the book wasn`t tuned against other programs. But this is not my point : this special book wasn`t supplied with F5, why did the SSDF accept this ? 2.) Fritz 5 wasn`t tested with the standard auto232 interface. It only works with a SPECIAL hard/software combination supplied by the producer. We don`t know, what `s possible with this special setup. But fact is :in difference to MChess, Rebel, Hiarcs, Genius etc. nobody of us has the possibility, to test and work with this special combination. Why ? Because it wasn`t delievered with the normal Fritz 5 package.... 3.) If you read the magazine CSS = ComputerSchach & Spiele then you `ll find in earlier editions a lot of results between Fritz 5 against some other programs like Rebel, Genius or MChess. So in consequence this autoplayer was available for ChessBase since summer. The SSDF received this special autoplayer some months later. Perhaps you can agree, that this procedure makes me a little bit sceptical...... 4.) The SSDF - Version of Fritz 5 requires 64 MB RAM size for minimum. This request is absolutly unusual and a privilege, which no other has demanded. Perhaps we see in the future the programm "kalashinkov X.Y.", which requires a minimum of Pentium II 400 MHz, 256 MB RAm ....etc. 5.) Fritz 5.0 is tested with endgame databases. Is this also the case with other programs such as MChess 7.1 or Shredder ??? These are the facts for my doubts. To make it very clear again : I have NOTHING against Fritz 5 or against the work of the SSDF. If somebody suspect, that here is a campaign against one these two parties, then this impression is wrong ! But I see only one way to stop all negative rumors : The SSDF shouldn`t allow these special conditions and continue the testwork with their well known basic principles --> testing only programs in a commercially available form, to which every computer chess fan in the world has access. If not, then we `ll have two versions of programs in the future : 1.) one for the public 2.) one for the SSDF In my view not a very good development. Let me know what you think ! -Peter
This page took 0.01 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.