Cheating in chess

[John Upham]

I don't know who told you this, but they are probably just making it up. The typical software disregards any moves which have been played before (theory) and any forced moves (for instance after a blunder or in a position where you are just mating with a Queen + K vs K). What moves are left are "undecided" and those moves are matched to the first three choices of Stockfish, and compared against another program.

This was contained in an official reply from a well known UK chess organisation considering an appeal for a lichess 5 match ban.

I doubt if this organisation was "making it up".

I concede that I might have misunderstood the detail of the communication.

[Roger de Coverly]

OK, but what is meant by (theory) here

To do it properly, you would have an enormous collection of games, continually updated and use it to build a tree of positions that had been seen before. Even more rigorous would be to add in published analysis from sources like New in Chess Yearbooks and Informator even if it yet to be played in practice.

Whilst one might suspect that elite GMs do something like this in their pet variations, do we think chess servers can be bothered?

Even in 1969, theory could stretch a long way.

These two games for example
https://www.chessgames.com/perl/chessgame?gid=1158749
(Clarke - Gligorich Hastings)
https://www.chessgames.com/perl/chessgame?gid=1132367
(Stein - Reshevsky Interzonal play off)

[Adam Raoof]

The typical software disregards any moves which have been played before (theory)

OK, but what is meant by (theory) here

true, this should read (="theory") - all that has gone before in practice

Adam

[JustinHorton]

So theory means "has been played before"? If so I have some questions.

[Matthew Turner]

I have been made aware of (at least) some of the criteria used to "decide" if a game has been played with external assistance.

One of those which struck me was the opening move cut-off is considered to be move 8.

i.e. moves 1-8 are disregarded and moves 9 onwards are considered to be ripe for analysis.

Therefore, if one was to play a line that theory is know to say move 25 then if you replicate moves 9 -25 then 16 moves may match with the engine used by the jury.

So you could play some well-known line of the Marshall Attack, Sveshnikov or anything with long forcing lines of theory and fall foul of the jury.

John,
There is an initial screening which has to use some sort of 'arbitrary' cut-offs. A full Regan test excludes all moves which are theory, defined as anything played by 2300+ players.

[Roger de Coverly]

So theory means "has been played before"? If so I have some questions.

What chess servers say they do and what they actually do could be different matters. The chess.com game analysis which you quoted in your blog indicates that the public facing chess.com game review doesn't have much depth.

[Roger de Coverly]

A full Regan test excludes all moves which are theory, defined as anything played by 2300+ players.

That's not a particularly strong definition, particularly in obscure, old fashioned or rarely played variations.

[John Upham]

John,
There is an initial screening which has to use some sort of 'arbitrary' cut-offs. A full Regan test excludes all moves which are theory, defined as anything played by 2300+ players.

OK, so I if was to trot out lengthy lines of theory (which I am entitled to do) I could be become designated a cheater because my rating says I am not permitted to know such things.

As a former active correspondence player I can and would do so. Am I placing myself at risk by having such knowledge?

[NickFaulks]

I don't know who told you this, but they are probably just making it up.

Ken said this himself in one of his papers. I've read so many that I'm not going to hunt for it now.

[Matthew Turner]

John,
You would not fall foul of a full Regan test. You may fall foul of one of the platform's detection system; In all fairness I am not sure that there is any way the platforms could avoid such a possibility. However, I think this would be a rare occurrence and I that there is every chance that on explaining the situation (to the provider) the ban would be overturned.

[JustinHorton]

there is every chance that on explaining the situation (to the provider) the ban would be overturned.

What is your reason for saying so

I don't know who told you this, but they are probably just making it up.

Ken said this himself in one of his papers. I've read so many that I'm not going to hunt for it now.

Is it possible that the eight moves thing means "we never look at anything before move eight, but you can still be in theory after that"? (Or course, as ever, it would be better to know than speculate.)

[Roger de Coverly]

You may fall foul of one of the platform's detection system; In all fairness I am not sure that there is any way the platforms could avoid such a possibility.

They could try not applying a test based on the principle of not being good enough to make the same moves as an engine without consulting it during the game. That's in the early middle game where memory or prior engine analysis are potential factors.

The more suspicious games to my mind are where a player appears to be busted but finds a long series of only moves to keep the game going. But that could equally be skill, even where such skill has not been on display in previous games.

[Matthew Turner]

Justin,
I know there are certain number of bans overturned and this would appear to be an example where a player could relatively easily argue their case. A human would look at the appeal and I imagine the opening theory aspect would be easily verifiable, so I would have thought an appeal could well be successful.

[JustinHorton]

You don't think you're assuming too many things here?

[NickFaulks]

Is it possible that the eight moves thing means "we never look at anything before move eight, but you can still be in theory after that"?

I don't remember anything which gave me that impression.