Author: Vincent Diepeveen
Date: 14:33:51 12/25/00
Go up one level in this thread
On December 25, 2000 at 11:17:57, Uri Blass wrote: >On December 25, 2000 at 09:11:03, Vincent Diepeveen wrote: > >>On December 25, 2000 at 02:20:46, Christophe Theron wrote: >> >>>On December 24, 2000 at 17:15:47, Vincent Diepeveen wrote: >>> >>>>On December 24, 2000 at 13:11:49, Christophe Theron wrote: >>>> >>>>>On December 24, 2000 at 09:09:27, Jeroen Noomen wrote: >>>>> >>>>>>My congratulations to Vishy Anand, for winning the >>>>>>FIDE World Championship 2000! >>>>>> >>>>>>3,5-0,5 in the final against Shirov, that leaves no >>>>>>discussion whatsoever. Anand was the best, remained >>>>>>unbeaten and scored a clear victory in the final. >>>>>>Well done! >>>>>> >>>>>>Jeroen >>>>> >>>>> >>>>>I am still absolutely amazed that a World Championship can be decided this way. >>>>> >>>>>A score of 3.5-0.5 is not statistically significant, not even with a low >>>>>confidence. >>>>> >>>> >>>> >>>>What a crap. >>> >>> >>>Thanks. Critisism from you tells me there must be some truth in what I say. >> >>in contradiction we didn't open the big stinking hole yet that is >>called home preparement. i didn't want to touch it yet, but >>in this post it's time to do so. >> >>> >>> >>> >>>>First of all all insignificant statistical elements that cause confusion >>>>were not allowed to join in this worldchamp in the first place. >>>> >>>>Secondly 3.5-0.5 in a final is very impressive. >>>> >>>>third did you forget how TOUGH anand had it to get into finals and how >>>>easily Shirov came there when looking who they played? >>>> >>>>Note that Shirov showed to be worlds best blitz player by >>>>knocking out people in blitz in a very convincing way! >>>> >>>>>It is now clear, at least amongst the experienced computers chess operators, >>>>>that such a result means NOTHING. >>>> >>>>You clearly don't see difference between what a computer is where any >>>>detail can distract scores or give weird results and a >>>>world championship mankind where no detail is getting done by random >>>>generator... >>> >>> >>>The kind of chess that computers play is not produced by a random generator. >> >>>BTW statistics about chess matches are independant of playing style or WHO or >>>WHAT is playing the match. >> >>>Why would human player escape from basic statistics? Because they have soul and >>>computers don't??? >> >>Ok let's adress SSDF. >> >>Though it's changing, tournaments book are very small in programs. >>Basically old versions of programs *always* play the same. >> >>The openingsbook of the best book is made by a player having 2163, >>no FIDE rating yet but he'll have that next year. >> >>0403 6495632 Noomen J. M 2163 2* >> >>He can't even invent his own new theory, he just copies some lines as played >>by GMs. >> >>But the big trick is that at schroeder BV, and the same happens with other >>companies, they play more games as the tournament books are wide from >>other programs. >> >>Now you can play Anand, Kramnik or Kasparov, but they don't play like they >>played several years ago, and you SURE do not know in advance whether >>a certain line wins against them. >> >>Not to mention that it's highly unlikely that you are 100% sure you kill >>them, whereas in computerchess if you have tested a few thousands of >>games at home, you have tested basically your book against the *entire* >>tournament book of the opponent so you KNOW what happens at SSDF. >> >>This says something of course about how well tested your product is, >>but it also says it has nothing in common with mankind. >> >>You can surprise Shirov only if your novelty is *real* good, but you sure >>can't surprise him with a release of a new book. >> >>This where for example the book from Fritz6 and Nimzo 7.32 is simply >>more or less a copy of the NCO book, with exception a few lines >>called 'unclear' lines they have been played at autoplayer and after >>it was clear who won the line it is added for that side, without >>figuring out whether it *objectively* is best. >> >>Computerchess has nothing to do with objectivism, but everything with >>testing and killing old programs at home. >> >>This in contrast to mankind. If i play only NCO lines with white (i play >>1.e4) then even against opponents of 22xx i'm not even sure whether >>those lines work. It can work 1 time against someone but then he >>either buys a few books, or he analysis till he finds a better move. >> >>It has happened only 1 time in my life that i fell for the same trick. >>It was against Eric de Haan. Still the game didn't go 100% similar as >>i thought i could improve somewhere, but i was quite sick when playing >>the same line. It was dead lost! >> >>Now you can't test in advance whether you are 100% sure going to >>win against Anand with a certain line. >> >>You can do however against fritz5. > >Thoretically it is possible but practically I do not see it. > >I do not see the new programs get 100% or even only 90% or 80% against old >programs like Shredder2 inspite of the fact that Shredder2 is using an old >hardware and a known opening book and inspite of the fact that Shredder2 has no >learning function so the opponents can try to repeat the same line that they >used to win against it. > >Uri You are using the wrong example, shredder2 is the worst example for this! Shredder is not getting used much anymore! In fact it would be cool to know how many people ever heart of shredder2 and know how it looks like! Interesting for schroeder BV is to beat chessbase & MCS, for chessbase to beat schroeder & MCS. So it's obvious which program against which program you want to look at. Take into account the date that things get shipped to SSDF however! Anyway that's an entire different contest *obviously* as playing in big tournaments where everyone gets there prepared at *that* time, instead of preparing against older products. Also learning of programs is not on the same level. I remember recently a note from Karlsson about learning of a program which is killing learning of other programs completely. How can that be translated to human chess? Further most auto232 players do not know who their opponent is! Humans nearly always know their opponent! Especially the highest level you know *exactly* what kind of opponent to expect! Top GMs know playing 1.e4 against Karpov is not smart, they know playing Gruenfeld against Kramnik is not smart. But against other opponents playing Gruenfeld can be very smart! Now in computerchess things are completely different! If you start auto232 playing in SSDF, then you are completely ignorant about this! Doesn't take away that it is *a contest* with its own advantages and disadvantages. But it is simply a different contest as playing in a worldchamp with programs, which in itself is also completely different from a human worldchamp again! Greetings, Vincent
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.