Chess.com FAQs and Discussion on Cheating

I pitty the fool who wrote this article.

I like the accumulation of hysteria

No, we DO take cheating seriously. Unfortunately, I can't really give you a good solid answer to your questions because to do so would be to give away information that I'm just not able to give out.

What do we want you to report? We want  you to report anyone you think is cheating. 

Is our cheating detection system perfect? No, there is human judgement involved, and that always leaves room for error, but we believe it to be the best and most reliable system available. 

Irresistible

I didn't do nothin'.  It was like that when I got here.  All the other guys were doin' it.  I only did what ANYONE would have done.  I had no real choice.  Was that wrong?  Because I gotta tell ya....  God/my dog told me to do it.

I noticed a major contradiction in the ite's FAQs and info about cheating - on the one hand they say the number of cheats is minimal - but on the other hand they say that they close down many accounts every week for cheating!!!! Which is the truth , guys?

Another point is - what's to stop these people reregistering under another pseudonym?

Baseballfan, here is what you can give out: the percentage of matching with the top 3 moves of an engine for the game I reported you. The game is public, anyone can run any analysis on it and make any such results public. Please specify the engine used and the depth. Seriously, many people here (included myself) would love to see if their matching analysis is wrong.

I agree that something around 80% would be inconclusive evidence. I also agree that human judgement must be involved. For instance, in a sharp game, one can achieve very high matchings.

Do you know how I picked the person I reported? AT RANDOM from those with high ratings (>2500). 

Congratulations costelus -- If the intent of your incessant overstating of the problem was to strike paranoia into other users, then you have succeeded.

Inverness, I would urge you not to leave.  Cheating is not nearly as widespread here as costelus would have you believe.

TheGrobe, that is ridiculous. Nobody needs anybody to tell them that X is cheating. People look at the games, especially those of the top players. It's quite common to examine such a game in order to find how to play a certain opening line. It's also not a big deal to let an engine run while doing this examination.

Nobody is stupid. When somebody sees that a certain player agrees with the engine much more than a GM, isn't it natural to have doubts? The fact that I publicly expressed my doubts doesn't mean I am the only one :)

Perhaps much more harmful than the cheating is to think that your oppoenent might be cheating (even if he might play fair).

The staff I think can make clear many doubts. For instace, they can say that "X has a rating of 2700+ on chess.com, but his ELO rating is 2500+. However, he doesn't want to reveal his true identity". That would be great!

I don't think it's ridiculous at all -- Inverness didn't cite games from the top players, he quoted you.  You've done more than just express your doubts, you've beat everyone over the head with them -- you sound like a broken record, and not just in this thread.

I think it's perfectly fine to express your concern, but the way in which you're doing it does a real disservice to Chess.com who now not only have to deal with the few cheaters that are here, but also a perception problem that you insist on perpetuating.

i agree

 I agree. Cheaters can cheat all they want. What, if you win a game do you get a cash prize? Do you instantly win a 7 day vacation in Hawaii? Cheating will not benefit them in any way, other than the "pride" that they "beat" a player. They might get away with it now (chess.com, please take no offense. I am not doubting your cheater catching strategies) but every single thing can't go your way. Eventially in life they WILL get caught, even if they don't get caught here, and the punishments will be more severe.

costelus, in My System, Nimzowitch famously wrote: "the threat is worse than the execution" -- you are demonstrating his point by analogy...  the threat of cheating has you so unhinged that you surely have already wasted hours more time discussing possible cheating than actual cheating will ever waste for you. 

Also, one of the points you tirelessly flog is this one:

I personally have no doubts that 90% agreement with the first 3 choices of any reasonable engine, after the opening, not counting forced moves, is cheating. Why no doubts? Because no player in the history has achieved that!

Um, I personally have a lot of doubts about that... I don't think you got the full story on this...  I believe your point holds true only when looking at master vs master games, when each player is setting tough problems for the other to solve.   When a master plays a patzer, I think the master is often quite capable of finding and playing the strongest move on the board, time after time... to understand what I'm saying, consider: -- if I hang my queen, do you really need Rybka to tell you to take it?

At any rate: Hardcore engine cheating exists and will be caught... and I've known and said for a while now: Cheating-lite surely goes on too, and is IMO, uncatchable.  Play chess according to your own conscience (or don't play chess if you can't find a way to be comfortable)... but don't let this stuff driver you crazy, and please, stop driving the rest of us crazy, too...

Jg27pyth, if you don't read carefully ... I replied once to the same objection you raised.

I said 90% agreement for UNFORCED moves. So, if there is a clearly winning move (like taking a full queen for free) and some other moves not so clearly winning, well, in such a case nobody should count taking the queen as a match!

Also, I never said that cheating is rampant here. I never said that I suspect everybody or generalizing statements like this.

I write most of this replies while waiting for my opponents to move in live chess :) Still, I spent an enormous amount of time playing with MirceaH. And maybe I have also another motivation...

And by the way ... the patzer who lost in the game I talked about has an ELO above 2200. Why don't you analize the game yourself?

No, but Rybka may be needed to tell someone not to take the queen.

Jg27pyth, if you don't read carefully ... I said 90% agreement for FORCED moves. So, if there is a clearly winning move (like taking a full queen for free) and some other moves not so clearly winning, well, in such a case nobody should count taking the queen as a match!

I read very carefully... there is nothing FORCED (caps, yours) about taking a hung Queen. It might be clearly winning, but it isn't FORCED. It's just easy to find.

... Secondly, a careful reader might have noticed that I wasn't giving that as a typical example, but rather as an exaggerated and obvious example of the principle I was trying to encourage you to understand (you didn't) which is that when a master plays a patzer, the master has a much easier time finding that "best" move -- the patzer doesn't set the master difficult problems.  

I have casually analyzed several games on this site in which both players are relatively weak players but one has a long string of move with very high agreement with an engine... (and you would declare him a cheaters with your whack, unscientific, completely self-cooked up non-methodology...) for the very reason I've mentioned... the best move simply isn't hard to find. The losing player is gasping along failing to find the magic Rybka defense, and the winning player is just playing the easy-to-find continuation that is still available.Let's be extra clear now -- not a FORCING continuation... just easy to find.

Please, before you spout again your -- I personally have no doubt that when moves agree over 90% of the time which is a higher percentage than any player ever in the history of recorded time blah blah blah... -- Please cite where you got this information, and what the methodology was used to analyze the games. Then please tell us if your testing is as rigorous and uses the same methodology? Are you sure your results are comparable to the earlier study?  I hear that Capablanca had the highest engine correspondence averaging around 80% (if memory serves, it probably doesn't)... any idea what Capa's best single game was, correspondence wise? I wonder if it wasn't rather higher than 80%, maybe high enough to cause you, personally, to have doubt.

source, please

Everything was said here, again, you prefer not to read:

http://www.chess.com/forum/view/game-analysis/computer-game-analysis

In particular the methodoly is discussed in detail here:

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

(again, said this a hundred times). 

What is your scientific methodology? Why don't you make it public?

Why don't you analyze that game? All those who posted in that topic and messaged me about that seem to confirm my results.

The correspondence game seems to be not so different from OTB (I analyzed a couple of games of the top corr. players from 1970-1980). 

Master? Haha, the player I'm talking about does not even have an ELO. Nobody knows him, nobody has ever seen him in a chess club in his area.

I should add that I would be very happy if you prove that I was wrong about that player and that game! No kidding.

Oh, now I see. My mistake, Jg27Pyth: 90% agreement for UNFORCED moves. At many points in the game I'm talking there is more than one good move. All evaluated by the computer within +-0.10 of the optimal one. Yet, the player almost always picks the best move out of let's say 5 reasonable choices (and no, it is not the most natural move).

Anyway, the bottom line: with the same engine I followed the recent match between Anand and Kramnik. It indicated to me that both Anand and Kramnik are quite weak players, compared to the level of the player I reported.

These are my conclusions. Please prove that I am wrong with clear evidence! Show your results and your methodology!