Chess Stack Exchange is a question and answer site for serious players and enthusiasts of chess. It only takes a minute to sign up.
Sign up to join this community
I’m a big fan of chess cheat detection apps, like Irwin that was created by lichess.org.
But I heard that there is a new bot called Leela that is teaching herself how to play chess, and to think in a human-like way. Is that true, and are there any ways to detect it?
Leela works by performing an exceptionally sophisticated positional evaluation at a relatively shallow search depth, whereas most chess engines work by performing a simple evaluation at as deep a search as possible. In theory this should produce a more positional style of play, and it does seem to be an effective strategy versus today's best conventional chess engines.
So in general, the moves selected by Leela will be different from those selected by Stockfish (Irwin's reference engine) in key positions. But the differences in move evaluation may be small from Stockfish's point of view; it's not that Stockfish thinks Leela's moves are bad per se, just that it doesn't reliably choose them for itself, because it doesn't see their unique advantages relative to a move that it has deeply researched tactically.
However, there are other characteristics of engine-users' play that differ from that of strong human players, and are observable and measured by Irwin.
Move timing statistics are an important one, and that probably won't change with Leela. Good human players recognise key moves, where many possibilities of nearly equal apparent strength exist, and plan ahead during them, taking more time. Then they play quickly when their opponent responds in the way they expected, or in easy positions where there is only one obvious good (or even legal) move. An engine user doesn't do this - for every move, they must enter the opponent's move into the engine, and wait a few seconds to see what to do.
Another characteristic which Irwin checks for is large variations in playing style, either between games or during a game. A player who plays some games extremely well with very few mistakes, but then plays other games (at the same time control) rather badly with several blunders, is probably a weak player who uses an engine for his good games. This is especially true if he seems to play his strong games against players with high ratings. Or a player might find himself losing a game, then get out the engine, in which case he will show blunders and a declining position, followed by extremely accurate play which salvages the game.
In short: yes, I believe Irwin can detect engine users, even if the engine in question happens to be Leela.
Unfortunately, if the cheater is smart enough, he's never going to be caught!
No matter how strong the cheat detector may be, if you only use the engine to help you out in one critical position, there is not enough data to get you caught.
Finally, the only thing that can be proven is similarity between your play and the engine's play, which does not constitute proof of cheating
