Author: Chris Hull
Date: 09:16:07 07/08/03
Go up one level in this thread
On July 08, 2003 at 06:23:55, Geoff wrote: >Hi > >... > >Q2 >Is there a method or utility program to determine the absolute rating strength >of a chess program? After a quick peruse of the ELO rating system, it seems to >have 2 drawbacks. Firstly, it is reminiscent of a Nuclear physics equation, >secondly and more importantly it is a relative rating. I had a quick tinker with >ELOSTAT, this removes the awkwardness of the calculation but the result was >totally dependent on the quess I made for the start rating. > The problem with making a test suite or method or utility program to determine the rating is that the programmers would program for the tests, kind of like teaching to the test that would be done in school.You could have an engine that does well on the tests but does not do so well in actual games. The problem boils down to "how do you test openings, middlegames, endgames, tactics, strategy, sacrifices, gambits, pins, forks, skewers, isolani, pawn structure, pawn levers, rook batteries etc, in such a way as to reveal the true strength of a chess programs." Not an easy problem to solve. Chris >I am surprised no one seems to have written a utility to quesstimate a rating >maybe from how a program performs using test suites. Ok, I know it will be a >rough estimate at best, but at least it would be an absolute number. > >Or maybe someone has done this, any pointers ? > > Regards Geoff
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.