Chess.com FAQs and Discussion on Cheating

I strongly disagree. Look, if I have two moves for my king out of check, one leads to mate in 2, the other one is OK for me, I don't have to be a GM to find the correct move. You count this as a top 1 matching, while this fact is irrelevant. What is relevant is the following: in positions where 5-6 moves are playable, the human choses one in the top 3 choices of a computer. 

Also, the automatic evaluation provided by Fritz is not that good. The program fails sometimes to evaluate correctly a good move. It will eventually see that the move is good only after 2-3 moves. 30s/move is also likely to label many of the moves of an experienced ICCF cyborg as non-matches. You can read more about cheating detection here, where you can see that producing a reliable analysis is far from easy. 

http://www.cse.buffalo.edu/~regan/chess/fidelity/

In the game Kramnik-Topalov above what is your conclusion? Kramnik definitely cheated, as he achieved much higher matching percentage than an engine? In fact,Topalov did accuse Kramnik of cheating during that particular game (the toilet cable story). 

Well it is a high matchup rate, but as you know by now (you naughty person! ) you need much more than one game as evidence.  However, on the basis of this game, I would look at others by the same player from the same tourney.

I would add that I don't have time to though...

Spassky achieved a 100% top 3 matchup in round 11 of the '72 WC.  His overall results were a top 3 matchup of 79.9%

I would tender that truly random is not the essential requirement; truly representative is - although in a lot of statistical situations, these would coincide. Truly representative here would simply be a large enough sample of games in which the test is believed to be a relevant one - all items pointed out by Steve. 30 games seems reasonably sufficient, provided one puts the margins-above-SuperGM at appropriate spots. One, as you say, would likely never be able to apply a codified statistical test to the results from this kind of study. But, fortunately, that isn't really needed.

That game was a final game for the World title. A victory in such a game might prove decisive for the rest of the match. There is no possibility of saying "one needs more games for a conclusion." 

Based on the statistical matchup IN THAT SINGLE GAME, Topalov accused Kramnik of cheating. What do you think (or your engine)? Did Kramnik cheat in that game? I don't care about the other games, the whole accusation of Topalov pointed THIS game and an imaginary toilet cable. After this game, the organizers made sure there are no cables in the toilet :))

By the way, in the game you just analyzed (second game Topalov-Kramnik, World Championship 2006), both Topalov and Kramnik missed a mate in 3. Has the engine flagged this as a matchup? :))

I think the point is that it's inconclusive due to insufficient data.  It may be suspicious, indicative even, but one has to err on the side of caution and say that fair play can't be ruled out, particularly given some of the other similar anomalous results for single games that predate engines which have been cited here.  These anomalies are precisely why more than one game is required before a conclusion can be drawn.

This may seem like a really dumb question but how is it even possible to cheat at chess. the game prevents illigal moves. although i one had someone take my pawn by movin past it with theirs. on their second move of that pawn.

I'm convinced that they do, which I believe is one of the main points of contention here:  The lack of a shoot-from-the-hip first ask questions later approach and damn the consequences for innocent players who might get banned along the way.

what i want to know is if an acccount is closed for cheating whats stopping that person from making a new account and doing the exact same thing again and again and again plus again to the power of 54?

littlehotpot: There are all sorts of tools available to help chess.com detect repeat cheaters. They can analyze IP addresses, they can analyze login times, forum post frequency, all sorts of stuff. It's a lot of work to get the rating up from a very low level that it starts at, perhaps they decide that it's not worth it?

Ozzie...when you just join chess.com winning a few games means your rating goes up hundreds of points! trust me I KNOW!

how cool would it be if you beat pelger on time for your first game. someone rubbish with a rating of 3000!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!