Cheating in chess

[JustinHorton]

Yeah I've seen all that and the marking on the player's account online, I'm just wondering why it is not marked on the app!

[Matthew Turner]

Yeah I've seen all that and the marking on the player's account online, I'm just wondering why it is not marked on the app!

I think people have observed this before, but I've no idea why it happens. I doubt it has any special meaning.

[JustinHorton]

We've noticed before that the wording seems to be different, but not that the marking's been meaning. (I agree it probably means nothing of any significance.)

[Li Wu]

Thought about things a bit more. Think there is not much point in discussions here about methods of these sites. We don't know what they do, although some might say it's reasonable to assume a few things (both good and bad mind- clearly I'm in the "good" camp).

People are free to say things that might be wrong with anti-cheating methodologies, but have disagreeing views on whether concerns can be met. If I see something that I can rebut, I will, but now with the caveat that I don't know if the sites are doing it. Whatever grievances aired may not be a thing as you don't know what the sites are doing either. You have different experiences and expectations than me, but we aren't working with much here unfortunately!

You can also take the view that hidden processes shouldn't be accepted in general, which is fine, too. We can get philosophical about processes which are public but difficult to understand (what's the difference?). However, specifically statistical evidence is definitely a thing that's scientific, reproducible, testable, improvable etc but relies more on good faith actors. And it works for chess too (just we don't know if this is the case for the sites, for sure).

One thing we can agree on- it's a shame that information isn't always public, and chesscom/lichess don't have much in way of accountability for a lot of decisions.

[Chris Rice]

We know Lichess use this computer analysis:



Though you'll probably have to get someone else to explain how it works!

[Wadih Khoury]

We know Lichess use this computer analysis:



Though you'll probably have to get someone else to explain how it works!

My guess:

1. The Centipawn column is self explanatory, though I guess they filter out obvious moves and/or theory. Standard deviation is interesting to note, as you can see that red and amber dots relate to low centipawn or SD
2. Move time is the advantage online providers have versus OtB. You can clearly see that someone moving in 4-6 seconds each move (average of 5 with a standard deviation of 1) in a 5+0 gme is more likely to be a cheat.
3. I suspect blurs is how many times the lichess window was out offocus. The top cheaters were 90%+ of the time not looking at lichess. Either they are very good at looking at Netflix while playing, or they were likely to be looking at an engine.
4. the combination of all of these give a probability of cheat (red sad face = cheater, amber sad is likely cheater)

[MartinCarpenter]

[*] Move time is the advantage online providers have versus OtB. You can clearly see that someone moving in 4-6 seconds each move (average of 5 with a standard deviation of 1) in a 5+0 gme is more likely to be a cheat.

This is I do wonder about - aren't most cheats manually reading moves off engines? That will automatically introduce a non trivial degree of randomness in the move times, and it would be very easy to manually introduce a few pauses.

Obviously if you're somehow using an engine to actually play the moves then it would be remarkably easy to program it to produce moves with an irregular time pattern.

[Matt Bridgeman]

Unfortunately there are videos on YouTube, some quite recent, with titles like ‘how to cheat on Lichess.’ So without Zoom and screen sharing there’s not a lot of hope in long plays, unless I suppose you are just doing events in very small groups of local, trusted club players.

[Kevin Thurlow]

You can imagine how excited I was to get an email from chess.com telling me that in 2020 I played 145 games and won 73 of them, had a top rating of 1979, and my best result was beating a 2724. I had no idea how many games I had played, or how many I had won, or indeed what the peak rating was. The win against the 2724 did surprise me - surely I would have remembered that? I checked my database, I lost to the 2724, I checked chess.com's record, I lost to the 2724. I don't want to jump to conclusions here, but if they can't work out the difference between a win and a loss, I'm not entirely confident that they can tell if someone is cheating.

[Matthew Turner]

You can imagine how excited I was to get an email from chess.com telling me that in 2020 I played 145 games and won 73 of them, had a top rating of 1979, and my best result was beating a 2724. I had no idea how many games I had played, or how many I had won, or indeed what the peak rating was. The win against the 2724 did surprise me - surely I would have remembered that? I checked my database, I lost to the 2724, I checked chess.com's record, I lost to the 2724. I don't want to jump to conclusions here, but if they can't work out the difference between a win and a loss, I'm not entirely confident that they can tell if someone is cheating.

This is indeed strange given you can easily look it up on the Chess.com site and find your best win was against ViktorTheCoolBoy 2376

[Kevin Thurlow]

Excitement follows excitement - chess.com have updated their terms of service https://www.chess.com/legal/user-agreement

Apparently you cannot use your username for any commercial purpose, which is interesting, especially for those people who play under their real name, and now presumably cannot write a book under their own names?

Oh, and they say, "If you have any questions, please reach out to our Support Team." (and teach them English.)

[Matthew Turner]

I got the new terms of service too. I started reading through and it seems like absolute nonsense, it feels like a knee jerk reaction to threats of legal action.

[Kevin Thurlow]

https://www.bbc.co.uk/news/world-55937157

Unbelievably, it got worse... Looking at BBC's Quiz of the week, Question 4's answer to a question about what game has been given up by some "famous" "gamer",

"It was Warzone. Vikkstar - Vikram Singh Barn to his mates - said the game, played by more than 50 million people worldwide since its release last March, was "in the worst state it's ever been" because of cheats. Hackers gain an advantage in various ways, such as using "aimbots" that allow them to shoot competitors without directing weapons accurately and "wall hacks" to spot other players through opaque objects. Last April, the game's publisher Activision said Warzone had "zero tolerance for cheaters" and had issued more than 70,000 permanent bans for transgressions."

Activision seem to be tougher than chess federations...

[Tim Spanton]

Marshmallows are tougher than most chess federations

[John Upham]

How does the so-called cheating algorithm behave when the player selects stronger moves than the cheat detection software uses?

For example if the player used one of the LC0 engines that could potentially best the cheat detection engine would the centipawn loss be affected?

Possibly if one was to use a stronger engine then one would be less likely to be accused?

I'm probably spouting nonsense.