Cheat detection

@ST4RSCR34M said in #1:

is there any software online that I can use to confirm my suspicions of having played a cheat? something I can pass a pgn to and it could give percentage chance of cheating going on.

be glad else you might be it.

Hello. I just had an Psychedelic moment in the last game. I just couldn’t believe my eyes.
So I was playing game after game (unrated 3,0 blitz) lost 5 in a row, then won one against a 2000- rated player. Then in the next game I met a 2400 rated player. The game lasted around 38 moves. I was white and it turned into a sicillian opening, where I play in a bit unusual style I have practiced a lot during several years. And I won. I just could not believe it. I have never won over such a high rated player before. In excitement I analyzed it with the comp.analyzer right afterwards.

The shock : I had only 5 cp loss ( 0 inaccuracies, 0 wrong moves, 0 blunders)
The opponent had 25 cp loss (0 inaccuracies, 0 wrong moves, but 1 blunder)

So if he reported me it clearly seems like I cheated. But the TRUTH: I did not. I just can not believe how I made it against such an good opponent that played without errors until the endgame. Miracles happen. It was amazing. Have never ever played a better game in my entire life in blitz. 99 % accuracy. Opponent 90 %.

If I’m getting banned for cheating, I can understand it. But it will be a false ban in this case ( with my hand on the bible or whatever, I can say from my heart it was no cheat)

Here is the game:

Lichess won't just look at some ACPL / percentage numbers. These highly depend on the course of the game and everyone can score good numbers every once in a while. On the contrary, a long streak of such perfect games is highly suspicious.

For a single game, it has to be proven that the moves came from an engine and that there's no likelihood you found them by yourself. E.g. move times could be a clear giveaway.

Lichess bans on more factors than just ACPL.

Of course it can be expected that your opponent reports you for it, but you shouldn't worry.

Chessbase centipawn analyses in Chessbase 16

@Eyon-chess said in #32:

Hello. I just had an Psychedelic moment in the last game. I just couldn’t believe my eyes. [...]
The shock : I had only 5 cp loss ( 0 inaccuracies, 0 wrong moves, 0 blunders)

I honestly think, it is an illusiion, anyone (software or human) could detect a cheater reliably, from one game, only.
And people who claim, someone cheated, just because there was a game with 0 inaccuracies clearly have not looked at much data - otherwise they would know, that this can happen even for relatively weak players (from time to time!). Also, playing a weak opponent can make it much easier to play "perfectly", when comparing to engine analysis and applying some simple metric like centipawn-loss, accuracy or whatnot.

Beyond that, I do think, that online-sites can detect (regular!) cheaters fairly reliably (over some number of games). But also, I think, that there will always remain uncertainty and that sophisticated cheating is a problem and will continue to be one, in the future, even more so. Sure, anti-cheating - measures can compare moves to the strongest engines (i.e. stockfish and / or Lc0, etc.), BUT it is also not too hard, to take one of those (even open-source!) engines and tweak them in a way, to prune "non-human" - like moves (at the cost of some playing-strength, but still playing top-human-level) and focus their search-space on more "natural" (human-like) moves, only. I doubt, that any automatic anti-cheating measure in online-play would ever be able to "detect" that.

Yes, the effort of the cheater would be much higher, than for the regular Joe, just downloading stockfish and firing that up, as it is, and I don't think, that many people would do it. But also, I doubt, that it hasn't been done somewhere on our planet, already.

Disclaimer: My claims are based on some understanding of the powers of current chess-engines - as a hobby-developer of a (weak!) engine and I don't encourage doing what I've said, above. But also (if I was even a strong chess-player), I wouldn't trust any serious prize-money - online-tournament - results, because of the views, I outlined, above.

I just saw this article. The person who wrote it claims that he has been cheating and nobody detected him:

"Occasionally, in online games where I have been testing out particular openings, I have used a chess engine (a dedicated computer program) to look for the best moves to try to get an understanding of positions.

Strictly speaking, that is cheating and if the chess platforms on which I play found out, I would be banned. But they never have, because after the first 15 or so moves I abandon the engine and just play on my wits, usually making the litany of blunders for which my chess is noted. Any suspicions anti-cheating systems have about my perfect play up to move 15 are allayed by my decidedly imperfect play over the next 30 or 40. I am a cheat who has got away with it."
https://www.theguardian.com/commentisfree/2022/oct/08/chess-cheat-magnus-carlsen-hans-niemann

Apparently detecting cheating online is impossible unless the person is dumb and uses chess engine the entire game...

@gogoLSFC said in #36:

I just saw this article. The person who wrote it claims that he has been cheating and nobody detected him:
[...]
Apparently detecting cheating online is impossible unless the person is dumb and uses chess engine the entire game...

Well, this seems fairly obvious, no?

I mean, if someone is "only" going to use an engine for a couple of moves or so and it's always the same first 15 or so - sure, that should be detectable, if people / software are looking for those patterns. But I don't think it takes much of an imaginative mind to come up with some scenario, where people "just" query an engine a few times, per game, irregularly, and get away with it - especially if they then end up loosing due to blundering away for the remainder of the game (like that person claims, they did).

I mean honestly: Based on what criterium should anyone detect such subtle "help" from an engine - unless you believe in esoteric software-power. xD

Personally, I honestly don't care, though - I'm not playing for money, am sticking to my true (low) rating-bracket, mostly, and I know, that anyone winning against me, with the help of some engine is really just hurting themselves, more than me. And I believe I'd be having the same mindset, even if I was a stronger player and playing at a much higher rating-bracket.

I'm not saying, that it doesn't matter - I just think, that it would be a more-or-less hopeless undertaking, trying to detect "smart" cheaters (who don't just always play stockfish-#1-moves). I think it is more useful, raising awareness for the possibility of cheating in online-chess, at any time, propagate fair-play and focus on making it impossible in an otb-setting, where results truely matter (i.e. with prize-money, etc.) and where detection should be much easier (possible!).

Apparently detecting cheating online is impossible unless the person is dumb and uses chess engine the entire game...

No. I mean, who even guarantees that the story is true ...

@Cedur216 said in #38:

No. I mean, who even guarantees that the story is true ...

I totally agree - who knows, whether or not it's just some poser.
...But that being said, I think just ignoring what he wrote would be missing the point, of this tread: I do think, in today's chess-online-world, it is entirely possible and not even unlikely that this happens countless times, every day. And personally, I don't see anyone preventing it (at least the "somewhat sophisticated" kind of cheating), in general - as opposed to otb, as long as we don't (yet) consider neural implants - but even that will be a concern, at some time, I'm sure, in the future and I'd be surprised, if I wouldn't see news of this, during my lifetime, even.

@Eyon-chess said in #32:

The shock : I had only 5 cp loss ( 0 inaccuracies, 0 wrong moves, 0 blunders)
The opponent had 25 cp loss (0 inaccuracies, 0 wrong moves, but 1 blunder)

It was a very simple game. Pieces were exchanged until your opponent's blunder going into a losing King and pawn ending.