Author: g.müller
Date: 14:04:14 09/13/98
Go up one level in this thread
On September 13, 1998 at 13:12:52, Bruce Moreland wrote: > >On September 13, 1998 at 11:52:09, blass uri wrote: > >>In this tournament the strongest version of cstal is playing(and not the latest >>or the commercial) >>Tiger11.2 is playing because Thorsten thinks that maybe it is strongest >>but the strongest version of fritz does not play >>Using this stupid powerbook is not the strongest version. > >My suggestion to this would be to factor this in to your opinion of the >tournament, then. If you think it was a bad book, then blame the result on the >bad book. > >The same thing has happened with Crafty. Someone wants to do something official >with Crafty and it becomes an effort to find the best Crafty version, get the >right book, the right hash table size, etc., and people have gotten at least one >step of this process wrong a couple of times that I have seen. > >I had the same problem with Fritz 4. I wanted to see how Fritz would handle a >position, so I ran the position, reported the results, and was flamed because I >hadn't gone to the trouble to optimize hash table size for my system, and I was >using the default that it was installed with, so Fritz solved the position some >percentage slower than it "should" have. > >I remember some argument about Rebel 8 (I think) installs, etc., regarding >X-mode and other stuff, discussion about how to set the thing up so that it >would play strongest. > >I think that since Thorsten is known to be an anti-Fan of Fritz, that he should >take especially careful steps to make sure that he gets the computer set up >right, selects the right book and gets it set up right, and documents the games >so that people can play them through and understand the opening book choices. > >It sounds like Fritz dynamically changes its opening book weights, so it will be >hard to verify the book. I test it with my power.ctg and it plays in two from eight times the same. So I think Thorsten plays the game correctly. He is no fan from fritz sure, but this is his opinion because he do not like the style Fritz5 plays and sure he knows that everybody here knows that he is no Fritz fan, so he would be an idiot if he manipulates the game by manapulating book-settings and he is no idiot. > >>The good thing in fritz5 is that you can do a book from a database of games. >>I think it is better to do a book from a database of only at least 2500 players >>and use this data for the fritz5 book. > >Thorsten should feel free to use it an any configuration as long as he has it >set up properly, describes his hardware, and describes the book. You shouldn't >have end-users messing around creating tournament opening books, same as you >shouldn't have them messing with eval-weights, you should use some configuration >that the average user is already using anyway, or the same configuration with an >add-on product installed, in my opinion. > >>I am sure that fritz5 did the moves in the tournament but I suspect it is >>because >>it learned wrong data from blitz games like in paris. > >Another issue with book learners. Probably best to run on a clean install so >you avoid this kind of thing. > >Although I'm not sure how Thorsten's machine could have learned from Frans' >machine, unless they are fraternizing while Thorsten is asleep. > >>The results of the tournament is not a reason not to buy fritz5 because the >>buyer does not have to use the same powerbook(I think that most of the buyers >>did not buy the powerbook). > >Pretty much nothing anyone says here is a reason to buy or not buy anything. >Anyone who is paying attention to anything knows that Thorsten is not an >independent, dispassionate tester. He obviously has strong feelings about all >of this, to the extent that people accuse him of influencing games. > >Personally, if I was going to buy a chess program, I would get my information >from a large variety of sources, then choose the one that seemed to be the best >*for me*. > >bruce
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.