Author: Drexel,Michael
Date: 23:47:58 01/29/04
Go up one level in this thread
On January 28, 2004 at 20:13:46, Mike S. wrote: >On January 28, 2004 at 14:40:48, Drexel,Michael wrote: > >>On January 27, 2004 at 16:52:59, Mike S. wrote: >> >>>(...) > >>You conclude from one match (specific player vs. specific program) that >>a normal GM is "most probably" weaker. >>The same person would cry out loud if someone would draw the conclusion from one >>short match Ruffian 2-Shredder 8 that ended clearly in favour of Ruffian 2: >> >>Ruffian 2 is most probably stronger than Shredder 8. > >In contrast to computerchess (where thousands of games are produced), it is >legitimate to draw conclusions from the Rogozenko match IMO - 16 games - because >we don't have much more results of that type. IIRC there was a Shuffle match >Comp vs. GM Jussupow, who lost too. I don't remember the details. 16 games does not mean anything. Especially in FRC (or even Shuffle chess). Some start positions might favour the humans, some might favour the machines. How would you know that not all 16 start positions favoured the machines for example? > >The only alternative would be, not to draw conclusions at all als long as 100, >500 or 1.000 GM vs. Machine games without opening book haven't been played >yet... > >>>(...) Most engines know >>>the common opening principles quite well (different quality of engines >>>undisputed). >> >>Man, you have obviously no clue about what you are talking here. >>Two of the three best programs according to SSDF (Fritz and Junior) know nothing >>about common opening principles. > >I didn't write "All engines know all common principles perfectly." I wrote "Most >engines know the common opening principles quite well (different quality of >engines undisputed)." > >Most, not all >quite well, not perfectly (IOW, not without exceptions) > >If I really would have no clue, my wording wouldn't have been so cautious. But I >have investigated engine's own opening abilities, understanding of developement, >etc. and even written articles about it. Some of what I saw was great, some was >incredibly ugly. So I have more than a clue about that. > >>Junior moves its Qeen around in the opening like someone who hasn't learned yet >>how to move with knights and bishops and Deep Fritz plays sometimes even worse: >> >>http://talkchess.com/forums/1/message.html?345123 >> >>I would call a human who plays these moves a Novice. Other chessplayers call >>them Patzers (or even Idiots). >> >>Even Shredder 8 is not able to play a decent opening completely without book >>moves. (...) > >You critizised that I draw conclusions from a 16 game match, but present only >one game each. Engines can look like idiots in a specific motif in one game, but >handle the same thing master-like in the next game. Engines are much less >constant than human chess player in such chess elements of a somewhat "higher" >nature, like developement, opening tempi etc. No, top engines are _in general_ relatively weak in the opening phase in classical chess, especially if they evaluate their position slightly worse. Es liegt natürlich nicht im Interesse der Top Großmeister die Engines dumm aussehen zu lassen. Das machen sie nur wenn sie im Match zurückliegen. :) Michael > >Nevetheless I think it's true, Shredder's opening play without book is better >than Fritz'. I had the same general impression in my experiments (I cannot >comment much on recent Juniors), also I remember others mentioning this. But it >can differ from game to game. > >Regards, >M.Scheidl
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.