Cooked puzzle

lichess.org/training/slVlB

The puzzle above has the cook 28...Qxe4 apart from the intended solution of 28...Bh2+ 29.Kxh2 Qxb5
I played 28...Qxe4 thinking that if 29.Qe8+ Kh7 30.Qf7+ Bg7 and it's winning because I still have the threat of Qxg2# which when prevented by Bf3 I would have ...Qb1+ (thus even 28...Qxe4 29.Bf3 Qb1+ looks good for Black; and the engine agrees as well). The alternate line would be 28...Qxe4 29.Bf3 Qb1+ 30.Kf2 Qc2 31.Bxb7 Qxd2+

I haven't checked the position until very high depths but I believe 28...Qxe4 is also winning for Black and is thus a cook, and the puzzle needs to be removed altogether. I have already thumbs-down it and I thought I should report it in the forum as well (speaking of which, it would be nice to have a mega-thread in the "Lichess Feedback" forum section where people can comment any cooked puzzles they encounter rather than having to create a new topic everytime)

Q X G 7

Qxe4 Bf3 is a good reply

@Anon581 said in #1:
> I haven't checked the position until very high depths but I believe 28...Qxe4 is also winning for Black and is thus a cook
Lichess puzzles want you to find the best solution.
What you provided is a winning solution, but not the best solution.

The solution wins a great deal of material whereas lines like 28...Qxe4 29.Bf3 Qb1+ 30.Kf2 Qc2 31.Bxb7 Qxd2+ only show a small advantage for Black like 0.9.
The problem is that, a few moves down the line, the position reduces down to an ending of pawns and opposite coloured bishops which is notorious for leading to a draw even with a pawn deficit. If a win could be achieved, it would be a long grind compared to the ease of exploiting a lost queen.

It took you so long to write this post, switching on the engine for one second would have showed immediately that only Bh2+ wins. I have no clue how you come to the conclusion that the engine likes Qxe4. It is like -9 compared to -0.8.

Bh2 wins the queen and is immediate resignation. Qxe4 may give a minuscule advantage.