If there is a formula by which we can ascertain with high confidence that someone is using computer assistance, correspondence chess can become a human game again!

Chess.com has stated that their method will only catch the most egregious cheats.

I find chess.com's methodology very interesting. I don't think I agree with their threshholds based on the scant evidence they've produced. Anand, Kramnik and Topalov all played at or even slightly above chess.com's highest limits of human play in their recent World championship matches.

It's reasonable to expect that a grandmaster of correspondence chess will perform at least as well as the best human in a timed game, and probably better. So instead of setting the benchmark at the level of 1980s CC players, it's clear that the benchmark needs to be higher.

Rating:
2633

Record:
386 wins
16 losses
27 draws

http://www.chess.com/echess/profile...

Accordingly, jaisol would have had 5 or 6 losses at the 140 game mark and 10 draws. So Yelena Dembo's record is not unusual, especially for the poor-quality of opponents/fans she often faced at that website.

dpruess boasts that Chess.com has never falsely banned an innocent player. Maybe that's because banning "bluesluice90" doesn't destroy the anonymous user's reputation, so rather than fight the false accusation, "bluesluice90" just plays chess at another website.

But accusing a titled player who's identity is known? This is something different. Ms. Dembo now has the choice to take legal action - the question will be whether Chess.com's methods determine with 100% efficiency that a player is cheating. I can guarantee you that Chess.com won't be able to defend their methods in court. Professionally-trained statisticians are too good with numbers to accept a single interpretation for any random sampling within a population.

No-one can say 100% that WGM Yelena Dembo used an engine in her chess.com games. What you can say is that she played far more engine-like chess when out of database than any pre-computer era Correspondence Chess World Champion finalist or Super GM otb player who has so far been analysed under the same conditions for benchmark purposes.

Dembo's account has been closed on chess.com
They would not take that decision based on scant evidence.

Your repeated point that it's not in chess.com's commercial interest to accuse a titled player of cheating is wrong on two counts.

First, chess.com's credibility and profile will sky-rocket if they correctly boot a proven cheat who is also a titled player.

But secondly, and germaine to the topic, this has absolutely no bearing on the correctness of chess.com's method. Your argument is just a red herring.

Well I should imagine any method which looks for possible engine users would involve noting the frequency at which a suspect's played moves match those of a powerful engine in any given position when the game goes out of a multi-million game database. Either that or it may involve average error/blundercheck methodology.

It would make an interesting test case, having an expert witness (ie Super GM) go over the Dembo chess.com games looking for signature engine moves & so on. Also the match up methodology would come under scrutiny & so would all the benchmarks.

If no one can say 100% that Yelena Dembo used an engine, then Chess.com is legally lost. Because this is what it said on Yelena Dembo's profile at Chess.com:

----

<WGM Yelena Dembo

ACCOUNT CLOSED FOR CHEATING

This member's account was closed because they were found to be cheating.>

----

Do you see the problem yet, Zygalski? Chess.com's methods can only suggest she was playing engine-like games. Chess.com has to interpret whether the results from the methods are suspicious enough to ban a player from their site. In a court of law, Yelena can take a team of professionally-trained statisticians and they will create a more favorable interpretation of the same data using methods supported by peer-review.

Therefore, Chess.com is lost. What they needed to have posted was:

"WGM Yelena Dembo

ACCOUNT CLOSED - This player was suspected by Chess.com of playing engine-like games, so we've close their account."

Here's another example of Chess.com's overzealous approach to banning: <List of caught cheaters>.

http://www.chess.com/forum/view/com...

This as we both know is a lie. No one can be called a cheater by Chess.com's methods or its interpretation. The thread should read,

"List of banned users suspected of playing engine-like games"

Zygalski, you cannot win this argument. Neither will Chess.com win this in a court of law. Because of their over-interpretation of the results their methods produce, Chess.com will need to defend such methods and their interpretation ABSOLUTELY. Since their methods cannot be defended ABSOLUTELY, Yelena Dembo will win. That is, if she decides to fight this.

With your methods, then Chess.com or any other site will not be able to ban anyone because of cheating.

Chess.com seems to have a well-established system to detect computer cheating; it is sad to read so many hundreds of names (some of which are probably repeats and duplicates). Yet it is quite bold to brand someone a CHEATER based on retrograde analysis of their games alone. Other playing sites might say "violation of site policies" or "account terminated" with no explanation rather than labeling them. I can see how two family members would use the same computer, only to have both accounts be CLOSED FOR CHEATING which cannot be undone. Many correspondence players seem to think that everybody is using computers during play, therefore they will too (so-called anti-cheating cheaters).

chess.com can accept or ban whomever they please for whatever reason they want. publicly calling someone a cheater is defamation, and that's a whole different story.