It occurred to me that the same code used to generate puzzles could *perhaps* be used to aid in cheat detection. As far as I can tell, puzzles come from positions in which one and only one move leads to the best evaluation given by Stockfish. This code could be used to find such positions (when they arise or after the fact), and look at the pattern of time usage specifically for that sequence, probably looking for a (clear) outlier.
I think this could allow one to detect "obvious", straightforward moves that are too slow, such as atypically slow recaptures, and move sequences that were initiated by the player (like traps and mating nets) but that are played as if each move were independent from the others (and perhaps other game situations I'm not thinking of).
Thoughts welcome, but please don't do like this other fine, fine user on my previous post who insinuated I wanted to aid in cheat detection so as to defeat the system myself.
I think this could allow one to detect "obvious", straightforward moves that are too slow, such as atypically slow recaptures, and move sequences that were initiated by the player (like traps and mating nets) but that are played as if each move were independent from the others (and perhaps other game situations I'm not thinking of).
Thoughts welcome, but please don't do like this other fine, fine user on my previous post who insinuated I wanted to aid in cheat detection so as to defeat the system myself.
