1

Sometimes I feel that a human, particularly a human of about my elo rating, would not ever play so.

I never used an engine in my life for cheating, but I think it is not so hard.

I think, most of the cheaters does not automatize "well". Thus, he has no way to step with the engine on the spot, instead if he switches to it, he needs to duplicate the state. This gets some time - maybe a minute - for him. After that, he starts to play astonishingly well.

If I experience this pattern, I report it, but there is a bitter feeling. Obviously reporting someone beause I lost is no good. I think I should do this only if I am sure, but I can never be sure.

There are rumors that chess.com has some very sophisticated technology to detect engine cheaters, but I doubt that they could really do it. I believe that they do their maximum (like utilizing some AI to measure the "human-ness" of the steps of the players), but I think anything what they could do, can be easily gamed (like using the engine only for some, critical steps of the party).

Do maybe some other patterns exist, which could raise the doubt?

6
  • (I am about 1100 on the chess.com, playing nearly exclusively 10 min games)
    – Gray Sheep
    Nov 18 '20 at 0:42
  • 1
    Losing as a 1100 player doesn't mean your opponent used a chess engine. Lot's of people can win against 1100 without an engine.
    – SmallChess
    Nov 18 '20 at 1:01
  • @SmallChess Yes. About 50% of the people I am playing with. And - maybe in 10% of them - I have some doubt.
    – Gray Sheep
    Nov 18 '20 at 1:05
  • 5
    Don't worry if your opponents were using engine or not. Improve your own games. Remove your blunders. As a 1100 level, I'm sure I can convincingly win without any engine help. So do many other players. Worry about it when you gain like 1000 rating points.
    – SmallChess
    Nov 18 '20 at 1:08
  • @SmallChess chess.com user stat shows that about half of their users are below my rating.
    – Gray Sheep
    Nov 18 '20 at 1:11
4

There are a range of factors sites use;

Such as play style, move depth , skill level vs move skill level. Engines are emotionless and are willing to just commit.

Things such as ;

  • Move depth not comparable to skill level
  • Move Time being consistent eg 5 seconds per move (hard to demonstrate her)
  • Consistent string of flawless games

Move depth not comparable to skill level

I have the following of a game played recently in an club vs club tournament on Lichess. This juniors game was detected as having cheating in it. Whilst it was being played people were suspicious by the two particular moves highlight below.

This highlighted moves are something you would expect to see of players rated 2000+ and isn't a natural move to find especially for juniors. The human brain sees the rook move and discards it almost immediately as a Rook trading for a Bishop. I have reviewed the games since and found that the move list was 1 for 1 the entire time for black against StockFish 10

[White "White "]
[Black "The Suspect"]
[Event "Hyper Aggressive Sacrifices (Blitz 3|2)"]
[FEN "2krr3/ppp2ppp/2n3b1/4n3/2P3P1/P3PN1P/1P1NB3/2KR1R2 b - - 0 18"]
[StartFlipped "0"]

1... Rd3! 2. Bxd3 Nxd3+ 3. Kc2 Nf2+ 4. Kc1 Nxd1 5. Kxd1

Another Example from the same game;

[White "White"]
[Black "The Suspect"]
[Result "0-1"]
[FEN "2kr4/ppp2ppp/6b1/8/2P1P1P1/P3rB1P/1PKN1n2/3R4 w - - 0 24"]

1. Rf1 Rxf3! 2. Nxf3 Bxe4+ 3. Kc3 Rd3+ 3. Kb4 Rxf3 3. Re1 Nd3+ 0-1

Not the answer you're looking for? Browse other questions tagged or ask your own question.