Author: Dann Corbit
Date: 17:38:36 12/13/00
Go up one level in this thread
On December 13, 2000 at 20:00:03, stuart taylor wrote: >On December 13, 2000 at 18:56:44, Dann Corbit wrote: > >>On December 13, 2000 at 13:59:15, ERIQ wrote: >>>how much memory do you think it would take to get such results ?? >> >>Memory is not the issue. The number of experiments against opponents of known >>strength is. >> >>To answer your original question: >>"Does anyone Know how strong is chess tiger on 1ghz mashine elo ??" >> >>The answer is 'No.' >> >>However, without trying to quantify it precisely; CT on 1GHz is darned strong. >> >>But then, I think you probably already knew that. > >Maybe near to 3000, according to non-adulterated (-100) ssdf elo testing. >Certainly according to selective search ratings. Nowhere near 3000. It's roughly exponential. A super GM with an ELO of 2600 is astonishingly good. A super GM with an ELO of 2700 is embarrassingly good. A super GM with an ELO of 2800 is... Kasparov. Win expectency for a difference of 1000 points is 0.00315231 Win expectency for a difference of 500 points is 0.0532402 So against an opponent of ELO 2000, it would lose 3 points out of 1000. For an opponent of ELO 2500, it would lose 5 points out of 100. No chess program will approach ELO 3000 any time soon, against either real GM's or other computer programs.
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.