> I would not even need to be given moves, just the answer on which move was way better
I feel like people keep looking past this line, and suggesting that the device transmit things like "caution" or whatever. In another thread, someone suggested it could warn you if the most tempting line was a bad one.
These all rely on the idea that Stockfish can think like a human and say "Carlsen will be thinking of X, I'll warn him that's a bad idea."
I've never seen any indication that Stockfish can know what a human is thinking, novice or grandmaster.
Instead Carlsen suggested something else: he asks about two moves, and Stockfish tells him which is rated higher.
That's totally doable by stockfish, but means all these suggestions require an input, such as a butt-squeeze detector.
Stockfish rates moves by chasing them down a game tree and evaluating positions. I can imagine several heuristics for a "non-obvious move here" buzzer based on that architecture:
- Hidden gem: The best move has similar value to others if exploring a few moves, but becomes clearly better at depth [n].
- Trap: There is a move that has similar value to others on shallow exploration, but offers opponent a checkmate at depth [n]
- Material exchange: The best move involves a major material loss in the first few moves.
I would love to see these tested in practice, but it seems very plausible to me that some combination of them could make a very useful one-bit (erm, one-butt?) signal.
> I've never seen any indication that Stockfish can know what a human is thinking, novice or grandmaster.
But it's incredibly easy to train ML models to predict what a human thinks is a good move. Then you just send a signal when "most probable move" is different than "chess engine strongest move".
I have worked on this problem. I would not say incredibly easy. Maia chess (linked in other reply) is pretty good but only has an accuracy of around 55% iirc, and that’s with enough data to train on. Grandmaster level games are rarer and have less data.
Most popular chess tournaments have a live audience. So that would be another vector.
Any delay would have to be significant like 30min to avoid players just waiting for the broadcast to match the board state during a critical move where they wanted to cheat.
And then if you see the players exit the playing hall while the board is 30min behind, do you just fast forward broadcast the remaining moves? That kind of takes the drama out of the most interesting part of the game. Or do you just keep the feed on delay and ask the players not to indicate if they won/lost yet? I’m sure their body language would be obvious.
For any tournament where it matters, they're being paid as a form of entertainer/athlete. It's not ideal, but it's also not unreasonable to put restrictions like that on them. It's already reasonably common to have things like interviews afterwards.
I feel like people keep looking past this line, and suggesting that the device transmit things like "caution" or whatever. In another thread, someone suggested it could warn you if the most tempting line was a bad one.
These all rely on the idea that Stockfish can think like a human and say "Carlsen will be thinking of X, I'll warn him that's a bad idea."
I've never seen any indication that Stockfish can know what a human is thinking, novice or grandmaster.
Instead Carlsen suggested something else: he asks about two moves, and Stockfish tells him which is rated higher.
That's totally doable by stockfish, but means all these suggestions require an input, such as a butt-squeeze detector.