Author: stuart taylor
Date: 03:10:02 05/20/02
Go up one level in this thread
On May 20, 2002 at 01:27:15, Uri Blass wrote: >On May 19, 2002 at 19:47:00, stuart taylor wrote: > >>On May 19, 2002 at 13:15:44, Uri Blass wrote: >> >>>On May 19, 2002 at 11:31:27, stuart taylor wrote: >>> >>>>That's one of the most important questions, for me. >>>>Would Genius 3 ever a) be beaten uniformly in every single game? b)Or atleast >>>>would it never win? >>>>If that has not yet been achieved, is it ALMOST achieved by now? >>> >>>I do not think that it almost achieved. >>> >>>A program may even beat another program that is 400 elo better in a single game. >>> >>>Genius3 is clearly better than being 400 elo weaker than the top programs. >>> >>>I guess that Genius3 is going to have chances to win against the top programs at >>>least in the next few years. >>> >>>A candidate master always has chances to beat grandmaster in a single game when >>>the opponents are humans. >>> >>>I see no reason to expect different things from chess programs. >>> >>>Uri >>I think it is not the same with chess programs, and as an example, Chess genius >>2 beat turbo King 2 at every single game (CG2 was on a 386/33). And that turbo >>King had been beating many other programs most of the time, e.g. Psion by R. >>Lang, CM2100, Zarkov, Novag scorpio! But come Chess Genius 2 on a 386 33mhz, and >>it lost every single game even at time odds, and even when started in a bad >>position to CG2 against TK in a good position, in middle of a game. >>S.Taylor > >I know that an old piece square table of my program that should probably not >have more than elo rating of 2000 could beat Nejmet3.05 that should probably >have rating of more than 2400. > >Nejmet3.05 also could beat Fritz7. > >I believe that there are programs at the same level of turbo king 2 that can >beat genius3 in a single game. > >It is possible that there are 2 programs when one can always beat the other >program but I do not believe that you can decide about it based on the >difference in rating. > >Uri The interesting thing is that Novag Scorpio did NOT lose every game to Genius 2, and maybe not even to Genius 3, on 386/33. But Turbo King which had a plus score against the Novag was absolutely crushed in every way every time by the Genius, even Genius 2 on a 386/33. Every new thing Turbo King tried was immeadiately exposed by Genius to be inferior. So I feel that it takes less than 700 in comp. comp games to get total domination, perhaps, even if not 400. I mean, comp. comp. ratings have always been constantly re-evaluated. S.Taylor
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.