lichess.org
Donate

Idea that brings us closer to measuring the difficulty of moves

EntschuldigungUwUedited

Can you elaborate on why using Maia is not a good idea? Maia can give a list of moves and the likelihood that Maia thinks they are played at a given level. The more likely an inaccurate move is to be played, the harder the position is. Yes, it is not perfect, but I think that is the best we've got.

For example, in the following position, it suggests 52% of the time an inaccurate move would be played (set to 2600)
r3k2r/1p2b1p1/p1nppn2/q3p3/4P3/2N1B2P/PPP2P2/R2QKBR1 w Qkq - 0 14

For comparison, in this position, Maia only suggests an inaccuracy rate of 10% (set to 2600)
r3kb1r/pp1n1p2/2p2pb1/3p3p/3P4/2N1P1P1/PP2NP1P/R3KB1R w KQkq - 0 12

So I believe one can conclude that finding an accurate move in the former position is more difficult than in the latter.

The fundamental idea is: what if we asked 1000 titled players to find a good move; how many would give an inaccurate move as an answer? I think this is a good way to quantify the difficulty of finding a move. Obviously, we can not ask 1000 titled players each time, but I think Maia Chess is the best alternative here when it comes to availability.

Can you elaborate on why using Maia is not a good idea? Maia can give a list of moves and the likelihood that Maia thinks they are played at a given level. The more likely an inaccurate move is to be played, the harder the position is. Yes, it is not perfect, but I think that is the best we've got. For example, in the following position, it suggests 52% of the time an inaccurate move would be played (set to 2600) r3k2r/1p2b1p1/p1nppn2/q3p3/4P3/2N1B2P/PPP2P2/R2QKBR1 w Qkq - 0 14 For comparison, in this position, Maia only suggests an inaccuracy rate of 10% (set to 2600) r3kb1r/pp1n1p2/2p2pb1/3p3p/3P4/2N1P1P1/PP2NP1P/R3KB1R w KQkq - 0 12 So I believe one can conclude that finding an accurate move in the former position is more difficult than in the latter. The fundamental idea is: what if we asked 1000 titled players to find a good move; how many would give an inaccurate move as an answer? I think this is a good way to quantify the difficulty of finding a move. Obviously, we can not ask 1000 titled players each time, but I think Maia Chess is the best alternative here when it comes to availability.