Author: Moritz Berger
Date: 11:58:09 02/23/98
Go up one level in this thread
On February 23, 1998 at 14:45:49, Thorsten Czub wrote: >The problem is : WHEN fritz5 is that strong, why didn't we found out >about ? >Why didn't you tell me before. Or Peter Schreiner. Or any other guy >HAVING FRITZ at home ? I told you a long time ago that Fritz 4 was strong and belonged among the top programs. I also told you that Fritz 5 was stronger, book learning etc. should give it a "significant" (i.e. measurable) better performance than its predecessor. Please don't call me NOBODY. >Anybody of us HAS fritz5. >Are we all blind ? >No - there is a mistake somewhere. >Lets wait until we see the games. I told you 6 months ago about my results. You refused to believe them. I even played more than 50 games against your pet favourite CST (and sent you the games!!!). Fritz scored >80% against most versions. >My accusations are still serious. As long as they don't throw out the >doublettes the whole idea of testing the playing strength is cheated. I tested without doubles (even without books in a C19 position of my own choice) and Fritz won about 60% of the games against the likes of Genius 5, Rebel 8&9, Hiarcs 6, M-Chess 6, ... >There is no 170 ELO jump ! SSDF systematically "downtested" Fritz 4. You yourself (!!!!) stated that Fritz 4 with big hash tables might indeed be 100 points stronger than they rated it. So there you go: P90->P200 + book learning and Powerbooks + engine improvements = 70 ELO points? Would be sad if it didn't figure this way! >We would have recognized this directly after the program appeared. But >we did not. Didn't YOU predict Fritz *wouldn't* get to the top 10 at the SSDF? I always argued that their results were somewhat fuzzy. But you insisted they had a relevant meaning. Now suddenly, you "don't believe" their results, even when the exact hardware and book settings they used for Fritz 5 are the best documented for any program the SSDF has ever tested? Always your humble servant Moritz
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.