Basically he is saying that if the winner's moves match the computer analysis (that is computers can't find any variation or better moves for the champ at all during analysis) then cheating is pretty darn likely. I dunno how I feel about this --- it is strong proof yet not positive proof. Its subjective to humans guessing the truth, in the end. But over a tournament, across 10+ games, it is proof too strong to ignore....
It's not guessing. You leave out certain moves, like book opening moves and certain obviously forced moves like recaptures. Then you control against the games of great players before computers were readily available. They generally find that even the best players can't do better than about a 70% matchup with the engine's first choice. (And there's another metric for matchup with the first three engine choices).
You also make sure and collect a large enough sample size against the strongest opponents they play. If all these controls are in place, and the player's matchups are still too high, then it is strong proof of cheating.
Agreed 100% BUT, while it is strong evidence, it is still not absolute. Without the device or catching the person assisting the player (if any) or something concrete, it remains a guess/judgement call ruling. I feel it is sufficient to make the claim. But I worry that someone who had an astoundingly good game may be hurt by this type of ruling. Odds of it happening? Extremely slim --- but possible.
Well, everyone's crying 'medical excuse' to get away from metal detectors and it's costly and impractical and bad for the game's image to make them play in a grounded metal box. And you have guys like Ivanov that are quite skilled at hiding things. So what else are we going to do?
BigDogg, part of the problem is that tournament organizers have created a powerful incentive to cheat - ridiculously high prizes (funded by ridiculously high entry fees) for class players. So, what do we do? I refuse to go to any tournament with a $300 entry fee. Of course, there are other incentives to cheat as well, e.g. a player wants to "earn" a title. But why go high tech when the old methods are still available (bribery).
Personally, I would agree to a quick electronic scan as a partial effort to combat cheating, but this would be very cumbersome because players are constantly entering and leaving the playing area. As far as medical certificates, it is a second-order effect - relatively small population and only a small fraction of those cheating. Also, you know everyone's rating history and there would be post analysis of the games - just as in Ivanov's case.
I'm not going to tell people what EF they should pay or what prize they should get in a tourney. I'd rather focus on the abusers, who are with us online as well, where there are no cash prizes whatsoever.
Statistical analysis of matchup rates can catch the most obvious of cheaters like Ivanov, but an occasional cheat could elude detection.
Still a big improvement.