Author: Rolf Tueschen
Date: 04:17:04 01/09/06
Go up one level in this thread
On January 09, 2006 at 06:06:59, Uri Blass wrote: >On January 09, 2006 at 05:28:33, Rolf Tueschen wrote: > >>On January 09, 2006 at 05:23:21, Uri Blass wrote: >> >>>On January 09, 2006 at 05:13:42, Rolf Tueschen wrote: >>> >>>>On January 09, 2006 at 04:47:44, Uri Blass wrote: >>>> >>>>>Wrong knowledge may be sometimes worse than no knowledge. >>>>>Kasparov believed that he was playing against Genius2 at that time but >>>>>Genius3(p90) probably plays above 2600 level in active chess and above 2400 >>>>>level in tournament time control. >>>>> >>>>>Note that I agree with you that programs are not overrated in the ssdf list and >>>>>genius3 on p90 can perform better than 2312 against humans at tournament time >>>>>control. >>>>> >>>>>Uri >>>> >>>>Of course SSDF is overrated. For the simple reason that they pretend to have Elo >>>>taken from human chess. But this is just wrong. >>>> >>>>They get test results but these results are a product of imbreeding. Machine >>>>play machines. You just cant conclude anything from that for human chess. The >>>>point is simply that no human super GM with over 2600 Elo would have such >>>>weaknesses we all know from CC. You know that, Uri. >>> >>>This proves nothing. >>> >>>Top GM's also have weaknesses that no computer with ssdf rating over 2600 have. >>>It does not prove that they are weaker than 2600. >>> >>>Uri >> >> >>It's astonishing me! I didnt want to say that machines are "weaker" than 2600 >>but that the SSDF Elo has NOTHING=ZERO to do with human chess. Will you agree? > > >I agree that it is better not to compare the ssdf rating with human chess. >We cannot get conclusions about results against humans from the ssdf list >and it is even possible that a machine with 2600 ssdf rating can get better >rating against humans relative to machine with 2800 ssdf rating. Exactly! But this is NOT what usually people might understand if they read the seemingly so scientifical results of SSDF. And take Fritz 9, where the claim is that it now is better suited for human chess. Of course then it is slightly worse against programs-ONLY-tweaked for computers. - We completely agree and therefore I was a bit surprised above, but now it's all ok again. I knew that you knew. :) > >Uri >Uri
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.