I heard when you cheat your account gets closed, but it is kept visible.
That would add to the shame of being caught. On another note, some of these idiots are reusing the same names. 7471, 747capt, 747captain? I would bet money those were the same guy.
>>Renegade \o/
Was this for one game or several games? How long was the game out of database? This would help us get some idea of the likelyhood of "help" other than just %.
Not being privy to the communications you received from Chess.com regarding this case I obviously can't know for certain, but I have difficulty with your claim that there was an indication from them that they were able to prove that the individual wasn't using an engine. "Inconclusive" I can believe, but your using very certain language in reference to this and I worry that you're misrepresenting what was actually communicated to you. I'm frankly surprised that anything was communicated to you at all as it seems at odds with Chess.com's policies.
The Grobe: I was communicated the result of their investigation: they do not have enough evidence to close that account. I do not have any other information and I have absolutely no idea about their cheating detection methods.
Of course, nobody can PROVE with certainty if one is using a computer or not. All we can do is to build some statistical evidence and say that it is unlikely that an amateur can play above the super-gm level. Unlikely does not mean impossible however.
OK, so yes, I was wrong: I should not have used the word "prove".
All this talk of move matching is non-sense. What engine? Fruit? Rybka? TheKing? Junior? Which versions? At what ply...12? 17? 25? What about Rybka 3 Human with contempt set at 75 and a talented centaur behind the wheel? Good luck....
If you talk about identifying stylistic differences I'm with you. The most obvious difference between strong human play and engine play is the method that's used to bring the game to a close. Engines don't simplify. Humans do. Engines will choose the most complicated and convoluted tactical sequence if it ends in mate. Humans choose the simplest route, the one that wins the piece that guarantees the won endgame and encourages resignation. Detecting these stylistic differences is much more reliable than move matching but of course it requires a human with chess skill to do so it's not a solution that's going to scale well.
One last thing: I can read subtext -- it's obvious that Costelus has doubts about chess.com's detection system. Guilt by implication is crap.
I said once, in detail how I analyzed that game. You did not want to read. Fruit 2.3 was the engine, depth was 15 for all the moves and at least 17 for the non-matches. This really does not matter, the same engine, on the same computer, with the same settings, thinks for instance that Topalov, in his recent match with Kamsky, played weaker than the normal level of some of the top players here.
costelus - You've made your point and no one cares...chess.com is not going to respond to your bait...give it a rest...
So....you support the fact that he is calling you a cheater?...being rated above 2100 and no OTB credentials puts you in a very suspect category in his mind...I have an even better idea...why don't you 2 go pm each other about the whole situation...heck you can even talk about me as I don't give a rats ass...
Riiiiiiight...and while on the internet, I'm the lone heir of Bill Gates...
You "don't give a rat's ass" yet found the time to type up these thoughtful rejoinders.
richie_and_oprah> Any Non-GM game which provides engine matching (Rybka, 64 bit running on multi-core) 60% of the time (or greater) once a game gets out of known database is getting assistance. Period.
http://www.chess.com/forum/view/game-showcase/a-fun-miniature
My out-of-book Rybka 3 agreement percent is 79%.
http://www.chess.com/livechess/game.html?id=18567334
My out-of-book Rybka 3 agreement percent is 75%
Greater than 60% agreement in Blitz games, twice in one day! I'm no GM, but it's obvious these games aren't evidence of assistance. 
richie, I'm running the latest version of Rybka 32-bit on multiple-processors with maxed out memory. Note, Rybka on multiple processors is non-deterministic by design. In other words, on this run the agreement % was 79% but next run maybe it's 82% or 75%.
likeforrests: Are you running latest verion on Rybka 64-bit on multi-processors?
Even if he (or she?) is using the non-deterministic multi-core Rybka, it doesn't change the fact that the percentage of move matching depends entirely on the character of the game in question (which was the point that likesforests was trying to make, I believe). Some games are more forcing in nature or simply have positions in which there are many obvious moves. Positions where there are few "good" moves will result in a high percentage of move matching if the human player is relatively strong. It's the quiet positions where move matching becomes much more relevant....
There IS A BIG DIFFERENCE between 32 bit (not so strong) and 64 bit.
Like pre-Rennaisance to modern day science, likeforrests.
Untrue when talking about analysis mode. There's no time limitation to restrict the depth of the analysis. Single core Rybka is deterministic, multi-core is not. The 64-bit is faster but the results will be the same if 32bit has enough time to calculate to the same depth.
RaO, my understanding is that the difference between 64-bit and 32-bit is faster eval functions based on the use of bitboards. The real playing strength of the top engines comes from the quality of the eval functions more than the speed. You can overcome a speed issue by just letting it crunch longer (given sufficient RAM). In fact, the 64-bit version will require more RAM in order to keep the same hash table size. I bet the 32 bit Rybka versions still beat 64 bit crafty.
Check the computer chess rating list (http://www.computerchess.org.uk/ccrl/4040/rating_list_all.html). The 32 bit Rybka 3 comes in ahead of plenty of 64 bit engines.
richie_and_oprah> There IS A BIG DIFFERENCE between 32 bit (not so strong) and 64 bit.
According to the Computer Chess Rating List, Rybka 3, 32-bit, multi-processor is still 100 elo stronger than Hiarcs 12 and stronger than every engine except Rybka 64-bit and Naum 64-bit. I believe maxing out my RAM also makes my copy a bit stronger. Granted, when I move to 64-bit my silicon monster will get a significant upgrade.
richie_and_oprah> In the end, a human being will have to make a hudgment call.
On this, we agree.
eastendboy: I agree.
And that is why a trained eye knows what to look for.
Fair enough. Concerning chess.com's detection system, my biggest concern is that there are accounts being closed as a result of an automated detection system. I would hope that there's a system of checks and balances in place that guarantees that a human of relatively strong strength (Class A+) is the one that renders a final verdict. Using automated systems to catch those that are suspicious is a great idea but I'd feel much more comfortable knowing that no account has been closed unless a human looked at the games and concluded that cheating had taken place....
If I were running chess.com's fraud detection system, they would have more than one detection system in place.
I very much like the idea of keeping the possible cheaters on the site with a label-- it keeps the revenue up, and players know what they may be in for. I don't know any of the names the posts have been using (Capablanca, etc.) and I still can't imagine how a computer can tell someone is cheating in their games, but the true reality is that no one will get any measurable benefit by cheating and gaining a few hundred rating points on this site. It won't go on their resume, it won't be listed in their life's achievements, and it won't help them be happier and wealthier. What's the real loss here? Let them play with a label if it's so 'insignificant' of a percentage. K
A recurring “truism” is that no human player can consistently match Fritz or any other engine in many games over time and once out of database to a greater extent than averages of
Top 1 Match: 60%
Top 2 Match: 75%
Top 3 Match: 85%
and anyone who exceeds these stats is without doubt using an engine.
Really? The player I reported had a top 1 match of 90%, and top 2 match of 100% (the moves in an opening database not counted). Still, using their state of the art system, chess.com clearly proved that he is not using an engine, but his enormous native talent.