Cheating in chess

[Li Wu]

The issue with statistical evidence in court is that it relies much more so on good faith arguments. The justice system is set up that the defence will argue everything in one direction, and by definition, use bad faith arguments (otherwise they aren't worth their salt as a defence lawyer).

E.g.- I say that I took a random sample and produced X, which has a probability of 0.001%. The defence can straight up lie and also say:
- I took a "random" sample and produced Y (e.g. in for chess cheating- he chose the worst moves possibles of his client).

Maghsoodloo's lawyers can even say his client hasn't memorised a single move of theory and is actually using pure calculation from move 1. Historically he has shown that with pure calculation alone he can produce up to 30 moves in a row matching top engines (e.g. some long 30 move opening theory). Does the prosecution then have to prove that the defendant does, in fact, memorise moves?

@Matthew I heard that in the stream as well about Leko. I think he was playing on chess24 which as far as I'm aware doesn't ban engine users?

[Tim Spanton]

Defence lawyers cannot "straight up lie"

[Li Wu]

Objectively all the engine matching tests show is that the choice of the human player was the same as that of the engine. in the absence of physical evidence or admissions it's an inference that the reason is that the player consulted the engine during the game, rather than it was equivalent levels of skill, coincidence or the player working with the engine before the game and remembering moves or patterns.

Objectively the only thing you can be sure of are things that you have observed personally. Everything else relies on trust upon witnesses, news articles, and expert witnesses interpreting evidence you don't understand.

-rather than it was equivalent levels of skill, coincidence
This is by definition what this sort of evidence is. It assigns a probability that such moves can be produced by any human/Magnus Carlsen/the player in question.

-or the player working with the engine before the game and remembering moves or patterns.
This is why especially for high level GMs, these sites don't ban lightly, and thoroughly review the data through human eyes.

[Li Wu]

Defence lawyers cannot "straight up lie"

Fair, I don't know much about law. Chalk this up as exaggeration of "bad faith argument".

[Matthew Turner]

@Matthew I heard that in the stream as well about Leko. I think he was playing on chess24 which as far as I'm aware doesn't ban engine users?

My impression was that Leko was talking about Chess.com, but I might be wrong. To be honest, I don't really know very much about Chess24, but from their FAQ section

"Can I use computer help (engine) while playing?

No! The use of computer engines or other tools while playing is strictly prohibited and violates our Terms and Conditions. The use of computer engine help while playing (cheating) will usually be handled by closing the relevant account."

I think it is true that for all the sites there is a vastly higher proportion of cheats at 1800 than there is at 2500

[JustinHorton]

Interesting conversation on Chess24 with Peter Leko talking about his battle to get his online rating up from 1500 and getting crushed by 1800's

This is the big news though

[Li Wu]

Yeah I think it's for the legends of chess series on c24. I don't remember Leko playing on chessom.

Also regarding the Mike Postle poker cheating scandal/court case, there was a graph produced (I can't verify his winrate displayed on this graph-disclaimer):

https://i.imgur.com/66i3Tii.jpg

Showing his winrate from TV recorded poker play. Potripper was a superuser scandal back in 2008 on Absolute Poker/Ultimate Bet- he was playing with FULL KNOWLEDGE of his opponent's holecards.

This is an example of what statistical evidence looks like. You can obviously continue to argue here that it could all be coincidence- well there is a probability we can assign to this coincidence and it's quite low!

[JustinHorton]

This is why especially for high level GMs, these sites don't ban lightly, and thoroughly review the data through human eyes.

I dunno how often it will be necessary to say this, but we don't actually know precisely what the sites do, and this is especially so when it's not high level GMs.

[Matthew Turner]

Yeah I think it's for the legends of chess series on c24. I don't remember Leko playing on chessom.

That doesn't really help because it was an anonymous account. As a titled account they would probably start you off with a much higher rating.

[Li Wu]

This is why especially for high level GMs, these sites don't ban lightly, and thoroughly review the data through human eyes.

I dunno how often it will be necessary to say this, but we don't actually know precisely what the sites do, and this is especially so when it's not high level GMs.

You're right, and we have established that we place different levels of trust on those sites.

But also conversely we know quite a bit of the care they take for high profile players. Anyone continuing to argue that such players being banned can be innocent, without personal connection to the player, probably will never change their mind on statistical evidence or maybe will never have trust for online sites to be competent in using data analysis to catch cheats.

[JustinHorton]

But also conversely we know quite a bit of the care they take for high profile players.

Do we? Where can we find this information? How much do we know about in any detail?

[Li Wu]

That doesn't really help because it was an anonymous account. As a titled account they would probably start you off with a much higher rating.
[/quote]

This makes sense. Fwiw:

https://www.chess.com/member/itshafu

This is Svidler's account that was used by Kasparov, and he indeed lost to some 1600-1800 players that are now banned.

[Roger de Coverly]

This is by definition what this sort of evidence is. It assigns a probability that such moves can be produced by any human/Magnus Carlsen/the player in question.

That's where the interpretations and assumptions come in. There's an underlying assertion that rating calculations are reliable down to the level of individual moves. A hypothetical player who could match GM choices 19 moves out of 20 but puts a queen or equivalent en prise on the 20th would have a modest rating as results would be poor.

i suppose there are two ways this could get to courts. One is where FIDE or a national federation attempted a publicised ban and this was hotly contested. The other is a dispute between players where accusations were made and the accused party took civil action for defamation etc.

[Li Wu]

But also conversely we know quite a bit of the care they take for high profile players.

Do we? Where can we find this information? How much do we know about in any detail?

Can't know for sure like everything else but all the top GM streamers say this, including Naroditsky's vod that Matthew linked. They could also be inferring this without knowing it as well.

But I think it makes sense, they think it makes sense. And if indeed someone innocent were to be banned, hopefully we will see some defining moment for online chess cheat-detection.

[JustinHorton]

They could also be inferring this

Well quite.