USCL Computer Cheating

Running Chess Tournaments
1

“** Official Warning ** The use of illegal Computer Assistance on US Chess Live will not be tolerated. The Computer Abuse Team believes that there is sufficient evidence that you’ve been computer cheating. This is your first and only warning; If we suspect Computer Abuse again we will take more severe action against you.”

Hilarious. I think the only reason I got this message is because I have like a 50-match win streak on USCL vs. 1800s and below, no one higher rated is accepting my seek ads and the fact that I switch programs a lot. My standard rating is 400 points higher than my blitz, but I suck at blitz, and its 200 points higher than my USCF.

I note that switching programs is the criteria they say they will use, so I guess I can’t blame whatever TO for sending me the warning. However, a USCF expert doesn’t have to switch programs to beat 1600 avg competition consistently. (For the record, mostly it’s to answer trade requests from my online stores in my magic the gathering online game) that I have up concurrently, and I’ve never played in a USCL tournament). A look at my games would show how both I and especially my opponents play really badly. I could easily prove beyond a reasonable doubt that I’m not cheating beyond their rather absurb definition of “switching programs=cheating”.

When I post a seek ad, I get a message akin to “three people have seen your request”. So, it’s not like I am going to worry in the slightest if they do indeed kick me. The server is barely worth my time to begin with.

At any rate, in the interest of fresh topics, if you were an online administrator, what criteria would you use to determine computer cheating? Would you stick to master-level game analysis? conversations with players? computer information-gathering programs? all of the above?
I have to think USCL has stuck only with the latter, though the first two seem much more revealing to me. Though there are millions of unique motives for everything in the world, it seems like people would have lots better things to do than computer cheat on casual-only games on USCL…would motivation be taken into account in your final judgment of evidence?

Looking to hear your thoughts.
Mostly curious because I’ve never ever been accused of cheating before in my life.
Ben Bentrup,
Senior TD

2

Ben, Grant Perks has a post discussing this very topic on the USCF Issues Forum on the thread “The man with the hat beats GM Smirin at World Open”.

3

Everyone considers USCL a joke, even those who play there regularly (like me). So let’s abstract the question from the specific server (USCL) about which it was posed, and consider it in terms of a “real” chess server – ICC.

From what I can tell, ICC is considered state-of-the-art at detecting and punishing online computer cheating. Their method relies entirely on computer analysis – that is, matching moves of suspect games against the first or second choices of one or more computer engines.

The extensive discussions of this past week’s World Open (alleged) cheating incidents within various forums, strongly suggests that the “ICC method” of engine-move-matching, is quickly gaining steam as a preferred way to vet suspicious over-the-board games, as well.

I have problems with this whole concept (i.e., if 80% or whatever of your moves correspond with Shredder’s choice, then you are presumed to be cheating). But I’m trying to keep an open mind.

Both ICC and USCL are clearly aware that if they ever had to “prove” in court that someone cheated, using their methods, they would get laughed out of the room. So they stick with penalties, such as revoking someone’s account, that could never subject them to any legal liability if applied arbitrarily. They don’t even publicly identify people they’ve booted for (suspected) cheating (doing so might conceivably leave them open to a libel claim). They just boot you, and don’t tell anyone but you.

This ties in with a side issue, of “rule-breaking” versus “cheating.” I believe it says in most chess servers’ EULAs that you agree not to program-switch. Therefore if they see you doing that, they can boot you for violating the EULA; they needn’t prove, or even claim, that you actually cheated.

A similar approach could be used in OTB events that had published rules against making phone calls during play, or having electronic devices within the playing hall. Being caught could then subject a player to forfeit or other penalties simply for violating the rule, with no need to prove he was cheating.

I had heard that ICC ejected numerous GMs from numerous tournaments based on these methods. So when I joined, I was expecting to get accused of cheating the first time I beat a higher-rated player. But it never happened.

Now I’m reduced to hoping I may someday do something over-the-board that might impress someone enough that they’d accuse me of cheating. :smiley:

4

I believe both ICC and USCL have had this ability for 4 or 5 years, so I wouldn’t call it state-of-art technology.

Take any game of at least 30 moves out of an Informant and see how many of the moves Shredder, Fritz, or Crafty pick as their first choice. I wouldn’t start comparing until move 11. I would also count the first choice as being the move the program list at the same interval for each move, e.g. 15 seconds, 1 minute, etc. I doubt you will find two games by the same player from a single tournament that match either one of these programs more than 67%. Then do the same thing for each of the games from the World Open played by the two accused cheaters. I won’t be surprised if your opinion didn’t change.

Ben mentioned program switching as being the possible means by which he was wrongly accused. The concept of program switch detection is to review for certain software files running when the user switches out of the chess server’s client during a game. If during a game you switched to Chessbase, Bookup, Chessmaster, Shredder, Crafty etc and back to the client on each move they will label you a computer cheater. Switching to a wizard card site wouldn’t trigger the program switching software.

I suspect that one of Ben’s opponents simply accused him of cheating and based on the discrepancy between his blitz, regular, and USCF ratings he got the warning. There is an art and a science to catching computer cheaters. When someone is wrongly accused it is more than likely the art that went wrong and not the science. Which appears to be the case here.

5

But there is something I don’t understand about this computer matching technique. That is, when someone claims that a computer matches an opponents moves they need to define how they used program. For example, did they set it to a certain depth level. That is, the person’s moves match Rybka to a depth level of 12. Or for a five minute game, do you set the program so it has to complete the analysis of the game in five minutes?

Now suppose you find a 80% match rate. You need to state what are the chances that the match rate is the result of chance. How could you arrive at the probability.

Second suppose the moves match 100% but on examination of the game we find that the opponent dropped a piece on every move and the player simply took the piece on every move. In this case, would we be justified in referrng to the player as a cheater. Therefore, an approach would have to also consider the level of play of the opponent.

I’ve heard many internet servers (e.g., playchess, ICC) claim they can detect cheaters. My question to them is what is the sensitivity and specificity of your methods. Sensitivity refers to the percentage of cheaters that you catch and specificity refers to the percentage of noncheaters that you correctly label as noncheaters. If the internet people cannot answer this question, then I would suggest that their confidence in their methods is not entirely justified.

I too have a funny story from chesslive I beat a computer program three games in a row very quickly by sacrificing pieces which it didn’t seem to be able to handle. That is, it kept taking them and then I would mate it. I got a message stating the program was reporting me for computer abuse and it had put me on it’s noplay list. When a computer beats me, can I report it for human abuse? I have a feeling a lot of this human abuse is going around. Maybe we should file a human abuse/discrimination suit.