Author: Uri Blass
Date: 06:59:49 05/03/01
Go up one level in this thread
On May 03, 2001 at 05:45:14, Graham Laight wrote: >On May 03, 2001 at 01:28:11, Uri Blass wrote: > >>On May 02, 2001 at 23:20:55, Chris Kantack wrote: >><snipped> >>>To me it won't be a big deal. Not that I don't love computer chess. But, >>>(last I heard anyway), Kramnik will have full access to the program months >>>before the match. >> >>I understood that it is not going to be truth and the programmers are allowed to >>change the program before the match when the condition is that they send regular >>updates to kramnik and if they send changes some minutes before every game he >>has no time to learn the changes. >> >>Unfortunately I am afraid that it is impossible to do productive changes in the >>last moment so the only chance of the programmers to do something without using >>some tricks that mean not giving the program to kramnik is to do the program >>weaker. >> >>I do not think that something that is 50-100 elo weaker than the top programs of >>today has chance against kramnik so my conclusion is that the match is not >>interesting and the result (a win for kramnik) is known. >> >>Uri > >Maybe you're right - but don't forget that in RGCC (and here) most people were >predicting an easy win for GK against Deeper Blue in 1997. I was not connected to the internet at that time but I guessed a win for Deeper blue before the second match. I believed that after losing 4-2 against kasparov and a year of work Deeper blue should be clearly better than the previous version. I guessed 3.5-2.5 or 4-2 for deeper blue. > >If the Fritz team can get their program running on a system which is much faster >than the one Kramnik will be sparring against, the match could well be more >difficult than he expects. I disagree. Kramnik can test the program at slower time control(12 minutes per move instead of 3 minutes per move in order to see if the weaknesses that he finds still exist at 12 minutes per move). He can start with blitz games and after finding positional errors he can give the program 12 minutes per move in the relevant positions in order to learn which weaknesses are relevant for the match. 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.