How To Catch Cheaters? - OTB

Hi. I have been reading about how cheating in OTB tournaments is becoming, in the view of many, a threat to the future of chess.

In this topic, I would like to defend 3 ideas:

1- Cheating is still far from being something that can damage chess if the right methods are taken to prevent it.

2- Statistical analysis is not an efficient method to prevent cheating.

3- There is a very efficient method to catch cheaters, and I believe it's almost infalible.

---

Now, let me explain those 3 ideas:

First, 2: Why statistical analysis is not a good way to catch cheaters?

Simply because, a cheater doesn't need to cheat at every move. He could cheat only every 2 or 3 moves. And then statistics would say nothing about it, since it's perfectly normal for a non-cheating player to have 50% engine matching.

Even if a player cheated at every move, he could not always pick the 1st choice of the engine. He could sometimes pick the first, sometimes the 5th best move, etc. in such a way that would be impossible to catch by statistics.

3- Is there a good method to catch cheaters? YES and it's a very guaranteed method.

I think that every player above such a level, for example, above 2500 should be required to do a TEST OF CHESS STRENGHT. This test could last a few hours, or a few days, in which the player's strenght would be tested, with full surveilance, in such a way that it would be impossible to cheat.

The result would tell the player's true strenght, and if it was much lower than the tournaments, he should not be allowed to play in tournaments, and should receive from that moment on, special investigation.

For this reason, as long as a player's strenght can be precisely TESTED, chess is NOT threatened by cheating. But we need that those measures become a rule, to identify the true, non-cheating chess strenght of players and remove from them the possibility of cheating.

Why not just play games in a Faraday Cage?

I do not think such testing gives anything what a regular elo rating lacks. Ratings reflect player's chess strength as well.

I think if a player is cheating in tournament games he would do that as well while solving the test. Therefore, we face the same problem that currently affects tournaments: making sure the individual does solve the test himself without any illegal assistance

But such a test would have 100% guarantee that it would be impossible to cheat.

For example, it could give random positions and ask what would the player play, instead of starting the game from the beginning.

Also, in such a test there could be no transmission of the game to anyone external who could have an engine.

The player would be scanned for any devices in his body, also.

Of course it makes sense.

trolls will troll, but if you want to really demonstrate this test could not be effective, you need serious arguments and follow logic. This test could have 100% guarantee that the player would have to play without the possibility of cheating. I think every player above 2500 should be required to do it from time to time, because as technology improves, the ways of cheating also do.

What kind of serious chess player wouldn't agree with a method that would catch cheaters, and give the guarantee and safety that the tournaments are clean and honest?

dude i dont think you can ever say 100% gurranteed, regardless of topic because you can never prove this to be the case.

i think precautions that are to be taken in near future are going to be sufficient for the time being, no electronic devices in playing area that are capable of acting as an engine or transmitting data(of course exceptions for doctors pagers and such). So you dont even have to be caught using one just that you have one on your person.  The irish chess union is taking on this stance i believe after a couple of incidents.

The issue of the tests you suggest (in low levels especially) is that it discourages rapid improvement de to accusations of cheating that could result from honest fair work.

the main soloution in my opinion is the education of new players because one of the most distressing things for me at tournaments is seeing certain high level schools encourage this culture of cheating to children in a win at all costs attitude. which when caught leads to a child leaving the sport and being branded as something he/she was taught to be.

cheating sadly probably wont ever be eradicated from the sport (or any sport) but if respect can be encouraged between players it can hopefully be reduced.

ejSalsac, I wouldn't wish any kind of discouragement for rapid progress of talented players, and I don't see how could a test do that. Also I suggested the test for GMs only, SPECIALLY because that's where the cheaters want to get. There is no use for cheating if it doesn't get the cheater to the GM level where the money and fame are.

well lots of low level tournaments offer cash prizes on a regular basis, and wouldnt be a bit late to start the test at the GM level opposed to catching it earlier on before they get the chance to get there damaging higher prestiged tournaments? 

and when someone exceeds the tests rating of their ability, accusations of cheating would ensue.

"and when someone exceeds the tests rating of their ability, accusations of cheating would ensue."

Of course not, the player would have the opportunity - and the duty - to make a new test at least each 100 points of improvement (or once a year), why not? If it's for the security of honest chess, I think every serious player would be interested and happy to do it.

You can't rely on one test to judge if someone is cheating. They could have a really good/bad day which would skew the results. Also, who pays for this long test? In order to make it useful you would have to take several precautions to make sure that they didn't cheat during the test which would require time and money. This doesn't sound practical at all and I doubt that many would be willing to take such a test if it lasts for hours/days as you said it might. It would be a colossal waste of people's time.

The most recent scandal involved an FM. Not over 2500. I would say most cheaters aren't over 2500 actually. It's easier to stay under the radar if you have bigger mistakes to capitalize on anyway, and weaker players will give you those bigger mistakes.

I'm a class-B player. If you let me use an engine's move for 5 moves a game, I'll take first prize at every tournament I enter, in my section, and no one will be the wiser, because 90+% of my moves will be -my- moves. It still will get me probably $10,000 over 4 tournaments. That's the kind of cheater I want to find a method of stopping.

"In order to make it useful you would have to take several precautions to make sure that they didn't cheat during the test which would require time and money. "

Would you rather let 1000 times more money to go to the hands of cheaters? There should be no economy, specially for a relatively insignificant spending, to secure chess against cheaters.

"They could have a really good/bad day which would skew the results."

If THAT is the reason, then the chance could be given for each player to chose the day or the test could be divided between many days.

"If it will be so easy to prevent people from cheating during this test, then why not just stop them cheating during the games?"

But that is clearly not happening as we have been seeing in the news.

Neither of these solutions convince me. I can never tell which day I will play best in the future or if I will play particularly badly on another day. Choosing the day of the test would therfore not eradicate fluke results (although it would be slightly more convenient for the person being tested). Spreading the test out over several days might help reduce fluke results but would waste even more time. The person would have to travel to and from the destination on several days which would be a large burden on their schedule. This would be especially bad if they didn't live near a place where they could be tested and would be travelling long  distances. It might not even solve the problem of fluke results since the person might be going through a particularly tough period which affects their chess ability.

The cost isn't worth the gain. Besides, what if they didn't cheat on the test but cheated during the tornuments? I mean, you have to stop the cheating wholesale, not just find it.

If you want to catch a cheater, you have to think like one.

See my post "Thinking Like a Cheater".

http://www.chess.com/forum/view/tournaments/thinking-like-a-cheater

The cost might be discourage some honest players from chess tournaments regardless of if the cost is insignificant compared to the amount made by cheaters. How much would this test cost (and please elaborate on how this test could be run to be sufficiently foolproof and not too expensive)? I don't know how much the average cheater makes so I don't know what figure you have in mind.

Now that is the real threat not foolish cheaters who use the engine for all their moves and get caught with statistical methods.

Even if the test shows that a player's level is say 2100 how can you prove he used the engine for just 3-5 critical moves? We need a method to determine the level of a single move - we would be able to say that a player constantly made 3-5 moves in each game which significantly surpass their natural strength - not an impossible task but extremely hard.

I think the effort should be involved in making the use of cheating technologies impossible/infeasible

"what if they didn't cheat on the test but cheated during the tornuments?"

Then the difference of performance would be evident, which is the whole point of the idea of a test.

For example, let's suppose Ivanov made such a test, probably the result would be a rating of about 1900-2000, which is undeniably inconsistent with his results in the tournaments where he cheated.

I wonder if anyone realises the futility of trying to stop people from cheating; it's the oldest trick any animal learns to do.  I dare say we are programmed to do it.  We are also programmed to hate anyone else that does.

This would work in case of Ivanov because his cheating is not intelligent: he uses the engine to make 90% of moves in his games which makes statistics tests trigger a warning even without any tests - he achieves better correlation with engine lines than any chess player in the history.

However I doubt it could really help in cases mentioned by Lucidish_Lux when the cheater plays in tournaments of his nominal strength and only uses the engine to find 3-5 critical moves. The performance would not be alarming because the opponents are of similar strength.

Ivanov is not smart as a cheater. How comes he can bet strong GMs and then lose against a 1500?

In case that Lucidish describes the tests could help after cheating in several tournaments if the cheater achieved higher rating but still solved the tests at his previous level.

If the cheater were smart enough, however he would just play a weaker tournament or two to have the rating go down. There are many "sound" ways to do that: not-so-obvious blunders (can happen to anyone) or making positions highly complicated and losing on time (this is not a rare case in complicated positions, is it?)

Rating manipulations are a separate form of cheating that I sometimes face. There are strong players who deliberately let their ratings drastically go down and then they play in B or C class tournaments and easily get 1st prizes because their real strength is much higher than any participant's. What is yet more disgusting, they are even able to earn money for decreasing their ratings - by offering to lose against their opponents in order to help them make higher rating! Some of the immoral opponents willingly pay them for that because they are focused on getting their rating higher by any means.

That is just to show how the system of tests might be compromised by smart bad guys whose only aim is to make more money on chess.

While I strongly agree with the idea that we should prevent cheating in chess, I have doubts if the proposed method is going to be sufficient - unless we find methods to rate single moves and be able to show all moves which were played with the help of an engine.