Puzzle with four equally good moves, but only one is "correct"

lichess.org/training/rjJjO

After 1... fxg5 2. b4 the only accepted solution is Kd5, even though h4, g4, and Ke6 win as well according to Stockfish at depth 40.

Ke6 isn't playable because you would walk into a check, if you push one of your queenside pawns, you will eventually need to come back if white plays c6. The puzzle can't handle this many variations and so it chooses the one that starts by eliminating all chances of screwing up.

@LyonFins said in #2:
> Ke6 isn't playable because you would walk into a check
Uh, how exactly?
> if you push one of your queenside pawns, you will eventually need to come back if white plays c6.
So what? 2... h5 3. c6 Kd6 for example still wins because White can't stop the pawns.
> The puzzle can't handle this many variations
Yes, that's why it shouldn't be a puzzle. The puzzle-finding algorithm does (very rarely) mess up and I'm pointing out one case where it did so it can be removed.
> and so it chooses the one that starts by eliminating all chances of screwing up.
To be fair, it is the most intuitive move, and I did play it after I saw it was winning, but a puzzle should only have one solution and not one intuitive one and a bunch of others. Also, there is no way an endgame this straightforward would be nearly 2300 rated if there weren't a significant number of people finding a different (but "wrong") winning move.

@noname1477 My Bad, ke6 is possible i was thinking of d6 but if your gonna play that move the logic is to move the king to stop the pawn later so why not just do that in one move. Otherwise your correct there are a million ways to win this.

Hello,

Thanks for the report, the puzzles will be deleted. In the meantime I've disabled it, meaning no other players should face it barring direct access via URL.