@Cedur216 said in #45:
> First mistake is already that while chesscom's methods are unknown, so someone could claim they have no programs
Their specific algorithms and methods aren't public knowledge, but they have had them vetted by statisticians and demonstrated them to titled players - see
www.chess.com/article/view/online-chess-cheatingChess.com CEO Erik Allebest has spoken about it in a number of interviews this year (often in the context of a wider interview)
@Cedur216 said in #45:
>here it is well known to everyone that we have Irwin and Kaladin.
I'm not as familiar with the lichess cheat detection methods nor what statements they've made about it, so that's new to me and pretty interesting: so both are AI models looking for (a) play that is characteristic of chess engines, and (b) play that is uncharacteristic for the player. There are other things you can look for as well - such as whether the active window is losing focus and getting it back and the timing of moves - that could be perhaps be a part of one or either of those models. Has cheat detection on lichess always used AI? It seems to be the sort of thing that has become more powerful more recently.