The New York Times has an article on a better approach to dealing with claims of computer cheating in chess. IM Kenneth Regan has developed a system that compares the game played with how a computer would play it. But unlike the simplistic, and often wrong, x% of moves match Fritz/Shredder metric, Regan has built up a model that also relies on the relative strength of the player. He has done this by getting computers to analyse thousands and thousand of games, and then matching the 'correct' choices to the strength of the players.
Apart from looking at a couple of interesting cases (including the Sebastien Feller case), he has also come up with another conclusion. According to his model, the 40th ranked player today plays at the same strength as Karpov did in the 1970's!
Hopefully Regan will be able to continue his work in this area, as a system for dealing with claims of cheating in chess would be a boon for organisers and officials, especially where the claim is without foundation.
Subscribe to:
Post Comments (Atom)
3 comments:
40th ranked player as good as Karpov? So there is no Elo upwards rating creep after all!
AO
I think we need to know more about the methodology before concluding that rating creep has been disproved, or that Karpov is number 40. If the program is looking at opening moves it will mark Karpov-1980 down relative to modern players who have better theory gained from computers, and in the ending may be marking Karpov down for actually being better than the computer. Also if the program is looking at thousands of games it may not be giving each move too much time.
Lets assume he's correct about the Karpov analogy notice how hes done well to use Karpov as a comparison point so people stroke their chins and say "oh golly thats interesting".
Whereas if you had used Fischer or Kasparov you would have people instantly frothing at the mouth.
I'm guessing its not as simple as opening theory. Opening theory is probably corrected for.
Mark Patterson, Esq.
Post a Comment