Chess Player Strip Searched

[MartinCarpenter]

It isn't difficult to think up a seemingly workable code, the tricky thing is going to be doing it reliably enough. You're looking at 50/60 moves per game, and often you'll need to do it at real speed. Lots of possibility for amusing mix ups

[Jonathan Bryant]

It isn't difficult to think up a seemingly workable code, the tricky thing is going to be doing it reliably enough. You're looking at 50/60 moves per game, and often you'll need to do it at real speed. Lots of possibility for amusing mix ups

Transmission errors are always possible of course - see the original von Neumann story, for example. All the more reason to keep the code as simple as possible.

Another lesson from my magic days: the more complex the method the higher the probability of failure sooner or later.


You probably wouldn't be able to use the suggested method in a blitz finish, no. But I'm sure you could make a similar objection to all the other established methods of cheating too.

[Chris Rice]

I mentioned a while ago about a proposal for engine-matching move cheating allegations and that the process should look something like this:

1. Develop a system to identify when possible cheating may be occurring using matching moves.
2. Using whatever criteria has been decided on one would then proceed to identify players who are falling foul of this criteria.
3. Investigate further using whatever other evidence is available and if the matter can't be dismissed then a committee should be convened to interview/test player concerned.
4. Player gets interviewed/tested and gets a chance to address concerns before committee. Probably a good idea if these concerns were in writing. Here is the player's chance to say that he/she is the illegitimate spawn of Data from Star Trek which is why he/she/it can play like a computer or whatever argument they can come up with.
5. Committee comes to a decision.
6. Appeals Committee

As a follow up I contacted Professor Regan directly and he seems to agree with the structure above, here is his reply:

"Dear Chris,
Thanks for the input! Indeed, this body of opinion is represented on the committee—that’s why I flowed the post from talking about the committee to the box with “is it a proof?” Practically we are starting with due-process, and I have drafted an outline of issues and procedural checkpoints—roughly like yours. As for what to do during a tournament, I am at this very moment (up late) giving the other 9 a hands-on “scholium”, very hands-on… Are you an ECF official? I’d be happy to confer with one…

—Ken R."

I think there is a window of opportunity for an ECF official to get involved with the new FIDE Anti-Cheating Committee if anyone is willing?

[Christopher Kreuzer]

I think there is a window of opportunity for an ECF official to get involved with the new FIDE Anti-Cheating Committee if anyone is willing?

The best input to such a committee from national bodies would be those with: (i) experience in organising tournaments; (ii) those who are arbiters at tournaments; (iii) and those who would have to deal with such allegations at national body level (appeals, bans, etc).

If you look at past allegations, you can see that national bodies have varied somewhat in their approach, with some common approaches as well. It would be good if the lessons from previous incidents can be incorporated into the processes being devised by this committee.

[Roger de Coverly]

1. Develop a system to identify when possible cheating may be occurring using matching moves.

Be cautious with this. A number of players would find this objectionable. So you play a good game beating a higher rated player, particularly one that you had prepared some analysis for and find yourself accused of cheating. I would have thought suspicious behaviour should be an equal starting point.

[Chris Rice]

1. Develop a system to identify when possible cheating may be occurring using matching moves.

Be cautious with this. A number of players would find this objectionable. So you play a good game beating a higher rated player, particularly one that you had prepared some analysis for and find yourself accused of cheating. I would have thought suspicious behaviour should be an equal starting point.

It's not about being cautious it's about having robust procedures.

Certainly any allegations would have to involve more than just a one off game, I agree that cannot be correct, you would need to show a consistent pattern and then prove its as a result of cheating. As we've discussed heavily on this thread for engine matching moves this is not going to be an easy or a trivial matter (and that's if it can be achieved at all) so proper procedures for investigating such claims need to be drafted. Recipients of allegations should be able to see the evidence against them and be able defend themselves with due process. That's all I'm suggesting, because in the absence of proper procedures we could end up with witch hunts.

[Roger de Coverly]

Certainly any allegations would have to involve more than just a one off game, I agree that cannot be correct, you would need to show a consistent pattern and then prove its as a result of cheating.

What would prompt an investigation though? Would every game be scrutinised, or only those from tournaments where a player scored above the prediction of his rating? So would arbiters be simultaneously reporting norms and referring players to the Commission on anti-chess Activities?

[Chris Rice]

Certainly any allegations would have to involve more than just a one off game, I agree that cannot be correct, you would need to show a consistent pattern and then prove its as a result of cheating.

What would prompt an investigation though? Would every game be scrutinised, or only those from tournaments where a player scored above the prediction of his rating? So would arbiters be simultaneously reporting norms and referring players to the Commission on anti-chess Activities?

All great questions and its the criteria that would prompt an investigation that need to be decided on by the FIDE Anti-Cheating Committee. They are clearly not at that stage yet and the current status seems to be that they are working toward drafting concrete proposals at the FIDE General Assembly in Tallinn, Estonia, in the first week of October.
My guess is that the reason they brought Prof Regan in was due to his ability with statistical modelling. So he will look to establish normal distribution curves on tournaments and players etc and then investigate deviations from the curves. They would need to establish how much of a deviation from the norm prompts a suspicion.

[David Guthrie]

Surely, any half decent cheater is going to work out a way to throw in a few "human" moves often enough to beat any published criteria for checking games for move matching against computer engines.

Its a pity that mobile phone jammers are illegal in so many countries. I'm sure I've seen/heard it suggested before, but what's wrong with the low tech solution of requiring both players to place their turned-off phones in a bag/container by the side of the board before play starts where everyone can see them (and then prohbiting possession of a phone (even if turned off) during play)?Obviously this wouldn't stop determined cheaters but would take away the opportunity for cheating for more opportunistic / less determined ones.

[Roger de Coverly]

My guess is that the reason they brought Prof Regan in was due to his ability with statistical modelling. So he will look to establish normal distribution curves on tournaments and players etc and then investigate deviations from the curves. They would need to establish how much of a deviation from the norm prompts a suspicion.

There's an assumption underlying this which in my view has a fundamental flaw. It might follow, but doesn't have to follow, that matching the moves suggested by an engine is indicative of cheating. Curiously you have to assume that engines aren't capable of perfect play to give it weight. To the extent that the play of top humans and top engines can converge, then the fact that both select the same move is a question of skill rather than collusion. I was reminded, on a thread about exchange sacrifices, that as engines get more sophisticated, then the engine matching factor of Petrosian gets higher.

[Roger de Coverly]

. I'm sure I've seen/heard it suggested before, but what's wrong with the low tech solution of requiring both players to place their turned-off phones in a bag/container by the side of the board before play starts where everyone can see them (and then prohbiting possession of a phone (even if turned off) during play)?

There isn't really, the practical problem would be of theft risk if the player were to leave the board, or for that matter to be distracted during the game. Making it clearer that a phone when switched off at the start of the game remains switched off until the end of the game would help as well.

[PeterFarr]

The starting point couldn't just be the tournament winner out-performing their rating - that would apply to most tournaments presumably. But probably you could look at performances exceeding a rating by a given threshold.

Actually you should also look at massive under-performance as well, in the search for fixed norm tournaments, as was mentioned earlier in the thread. In fact I believe any tournament where norms are achieved should be subject to more rigorous checking, as its obvious there is a great deal of cheating going on of this type (as also noted in the separate thread on this). As Roger said earlier, this is not computer-related, although computer analysis should help in the investigations.

In the end, FIDE need to look at a range of sources; their own database of tournaments, players, arbiters and games is a huge resource to analyse, looking for unusual patterns, not just within games but in results. But I suspect the biggest initial source will still come from more direct sources - suspicious opponents etc.

Particularly given that this type of analysis is rather new, Roger does have a point that it's dangerous just to rely on statistical evidence as "proof"; it's circumstantial evidence that needs corroboration. But this isn't a fundamental flaw, it's just something of which to be mindful. if nothing else, statistical approaches should help narrow down the focus of where to investigate for suspicious activity, which is potentially a big efficiency gain - you are not then looking for a needle in a haystack.

David's point about masking cheating by throwing in human moves is a good one, but then that means the cheater also limits his gains; to some extent it depends how self-disciplined they are.

I have a feeling that cheaters actually get steadily less cautious the more that they succeed, and it's this that can lead to their downfall. I imagine that the scam idea of running totally fictitious tournaments actually started with a few games in a real tournament being bought and sold to achieve the correct norm results, leading on to more systematic rigging, and finally to the realization that you could do the whole thing from the comfort of your own front room. But then it got so obvious that it was almost impossible to miss, even for FIDE.

[MartinCarpenter]

For checking games for suspicion, its easy. You do it for every single game that you can as a matter of routine. Pretty easy to do I think.

Not only is that the only fair way to do things you get a massive deterence effect too, and hopefully avoid the sorts of semi lynch mobs that we currently seem to be getting at times.

[Jonathan Bryant]

...Roger does have a point that it's dangerous just to rely on statistical evidence as "proof" ....

Well Roger's point was more that it's not right to draw conclusions from a single game. Take a large enough sample and it's a totally different story.

Robust procedures, as Chris put it.

I have a feeling that cheaters actually get steadily less cautious the more that they succeed, and it's this that can lead to their downfall.

You may well be on to something here. I'm quite sure that it would be possible to cheat in club games and never be caught if you consulted an engine on your phone say, once a game.

As a thought experiment I'm rather curious as to how much of a rating improvement this would be bring although obviously I'd never actually do it. Partly because I just don't want to and partly because I'm pretty sure that after a few games once wouldn't be enough and I'd do it twice and then it would be three times and then it would be every move.


Then again, it might be true that only the cheaters that have been caught follow this pattern. In theory, at least, there could be hundreds out there quietly following the 'restrained behaviour' pattern and we'd never know.

[Jonathan Bryant]

Oops. Double post. Delete delete.

Was it Freakonomics that had chapters on using statistical analysis to catch cheating sumo wrestlers and school teachers?

If you can use the method for that I don't see why you can't find a reliable way of doing it in chess.