This is a subtle issue, but one that I think would make puzzles more fun & challenging if it were resolved well.
Here's a puzzle I think is deceptively easy: lichess.org/training/7xLWW -- the rating on the puzzle is 1166.
The first move, 1.Nf6+ is very easy to spot, but the correct follow-up to 1...Kf8 is not as easy to see. However, if you play Nf6+, the computer responds with 1...Qxf6, the only move that isn't a forced mate. It's much easier to grab the queen with 2.Qxf6 than it is to find the mate (at least, in my opinion).
I'm not sure if the puzzle always uses the engine's best continuation, but it seems like it does, and the result is often I'll identify what I think is the best move without being able to completely calculate the entire forced line, and the computer will immediately give up a bunch of material to avoid a forced mate that I can't see. I'm left a little bit dissatisfied because I don't get the opportunity to learn the different mating attacks because the computer assumes that I can find them.
A few ideas:
- (Easier) In puzzles with a forced mate, have the puzzle engine choose the continuation with the longest forced mate over continuations that give up large amounts of material.
- (Harder) Have the puzzle engine sample the best N continuations, and identify the continuation that is the hardest for players (and likely, most thematic for the puzzle) by seeing which one is hardest for players.
What I'm suggesting is a bit double-edged; I'm not nearly good enough to avoid being surprised by the puzzle engine even when I think I've calculated correctly, but I think it'd improve puzzle quality generally to make players to identify forced mates.
Here's a puzzle I think is deceptively easy: lichess.org/training/7xLWW -- the rating on the puzzle is 1166.
The first move, 1.Nf6+ is very easy to spot, but the correct follow-up to 1...Kf8 is not as easy to see. However, if you play Nf6+, the computer responds with 1...Qxf6, the only move that isn't a forced mate. It's much easier to grab the queen with 2.Qxf6 than it is to find the mate (at least, in my opinion).
I'm not sure if the puzzle always uses the engine's best continuation, but it seems like it does, and the result is often I'll identify what I think is the best move without being able to completely calculate the entire forced line, and the computer will immediately give up a bunch of material to avoid a forced mate that I can't see. I'm left a little bit dissatisfied because I don't get the opportunity to learn the different mating attacks because the computer assumes that I can find them.
A few ideas:
- (Easier) In puzzles with a forced mate, have the puzzle engine choose the continuation with the longest forced mate over continuations that give up large amounts of material.
- (Harder) Have the puzzle engine sample the best N continuations, and identify the continuation that is the hardest for players (and likely, most thematic for the puzzle) by seeing which one is hardest for players.
What I'm suggesting is a bit double-edged; I'm not nearly good enough to avoid being surprised by the puzzle engine even when I think I've calculated correctly, but I think it'd improve puzzle quality generally to make players to identify forced mates.