This "effort" in one specific game might be enough to get cought. Every game with a very high performance compared to a strong engine can cause you to get banned. So it will be risky for the cheater. There will be no safe road.
cheating on chess.com

Ralex>This "effort" in one specific game might be enough to get cought... If your opponent has scored 85 % or more, compared to the engine, he might be cheating... Every game with a very high performance compared to a strong engine can cause you to get banned.
I'm glad Erik's using a more advanced system, which seems to be working well.
When I play a much weaker player, how to exploit their errors is sometimes obvious, and my moves and those of an engine match in nearly 100% of out-of-book moves. I'm hardly unique. If you're coming up with your own formula, you need to account for that.
likeforests: not two games are alike. Naturally you score high against a much weaker player, when the moves are obvious! So you know about this more advanced system and that it`s working well? I only know of one site where they have a "task force" working thouroughly with this problem. And that will be RedHotPawn.

JG27Pyth: Everybody blunders. Never blundering over many hundreds or thousands of moves is a sure sign of cheating (via blunder checks). I trust that the methods used by chess.com are in fact sophisticated enough to take the kinds of scenarios you've identified into account and I think you'd be surprised what is and is not detectable. Even nuanced, subtle cheating designed to beat detection will leave a signature.
Erik, if you say you detect some cheaters, why don`t you make their names public? Or put a (C) next to their names?
TheGrobe: When you go over a game with an engine, you will soon find out whether cheating can be an issue or not. High percentages will lead you the way. Some moves will not be the very best. But they will still represent a high percentage. Try it out for your self! Post 383

Yes, my point was more in reference to some of the alleged cheating that is done with this in mind and is designed to avoid this type of check (per JG27Pyth's post #395). What I was trying to convey, is that this behaviour too will leave a detectable signature as it is forced and deliberate when compared to a game that is played naturally. It may be much harder to detect, but the signature will be there and it will be detectable.
A good example of this same idea lies within the field of forensic accounting where an investigation is often looking to identify very similar behaviour (i.e. falsified accounts designed to look as legitimate as possible). As an example, one methodology that is used in that field is the application of Benford's Law which governs the typical distribution of the numbers one through ten as the first digit of a set of numbers that is being investigated. Deviance from this distribution in a set of accounting records is a red flag.
As I've said, I'm sure that these types of signatures (though obviously not Benford's Law specifically) will also exist in the games played by cheaters and I trust that the detection methodologies employed by chess.com are very much in tune with them.

Ralex, now you know a second site that takes cheating seriously. chess.com relies on a mix of personal honesty, secret detection algorithms, human reporting, and admin verification to detect and limit engine abuse. The details of the algorithms are, of course, secret. But even before Erik added this stuff, below a certain rating... it's rare.
An example of what an account closed due to cheating looks like.
A thread about Erik & staff banning quite a few cheaters.
likeforests: There`s no such thing like secret detection algorithms. There will probably be a policy concerning suspicious elevating and supernatural high scores. It will come down to reporting and engine aided verification. It takes manpower to handle this issue. I doubt that Chess.com got this manpower. This is the reason why most sites let it slip through their fingers. Anybody can have an engine look in top 100! That will be a healty exercise! Look at post 383.
Thank you, likeforests! I have read those posts. And they are all fine! I know, efforts have been put into these matters. But I believe that there`s still a long way to go. Though this is a young and vibrant site, so there will still be time!
In all honesty i had no clue people could cheat till my friend showed me how they do it, which was this past weekend. Ive thought about it and it makes no sense to me, you get nothing out of it. Sure your score goes up but you wont be able to develope your game. The people who cheat can only play online because if they play at a club or at any other place they wont have their trusty program to help them out. So i say, let the cheaters cheat. Scores mean nothing if you dont earn them.
One thing that is kind of interesting is that if you were in fact a cheater, and then you read a bit of stuff about how they detect, and then you got scared and went back to being your old self, this could actually get you caught. They might be able to say that over a certain stretch of time that you were getting assistance, and outside of that you were not.
Or maybe the potential offender only uses assistance in tournaments. Or only in one specific tournament.
I've said it before, but all the talk we do here is peanuts compared to what they actually do. We think for 10 seconds, and write the first thing that comes to our minds. They think about the problem and discuss it. The difference in applied IQ must be quite significant.
Oh, and for the above suggestion, you'd have to correct for the fact that the player might actually put "more effort" into the tournament games than some side games they have going at that time. Or, maybe once they realize that they've already qualified to get to the next round of the tournament, that they "don't care" about the games as much. Lots of interesting angles to the problem.