Puzzle broken

The final move is drawish

@ambrona I'll gladly take you up on that offer, from the position at the end of the puzzle. White has temporizing moves with his K. Black does not, because it is the sole defender of his c-pawn. The result is that it is very hard to ever make White move the bishop, and the only situation in which he must, he can return it there before Black can threaten the g3 pawn. The critical position is what I've described in my previous post. The only way Black progresses is if White blunders.

Feel free to send me a challenge from the position after 51 ...Be5. I'm confident that we will draw.

@phoenixshade
Ok! I will challenge you, unrated. We are allowed to use all the assistance that we want, including engines. Do you agree?

@ambrona

You might well be right, but it is all not easy. In your line:

57... Bf2
58. Bd6 Be3
59. Bf8 (it does matter!) Bd4
60. Be7 Be5
61. Bh4

and I fail to see how I am losing my pawn :-)

@ambrona & everyone

actually, 59. Bf8 there is 60 .... Bf4 !!

Stupid computer really isn't much help here, but this might just do the trick. If it does, and if that is the only way (if otherwise Bf8 and the idea of Bh4 saves from getting out-tempo-d) , the whole thing is actually beautifull and deep.

Thumbed down my own post because even though Black cannot temporize on b4-a3, he can do so on c1-a3. If White attempts to cut this off (for example with Be3), Black can threaten on the h2-b8 diagonal. It's complicated, but essentially White will get zugzwanged. This is a nice study, one of those where Black has to make double attacks first on SQUARES from whence he has double attacks on the pawns. (And it is about 3 levels deep in this regard.) A very educational position worthy of study, and if you ever find yourself the victim of mysterious bishop double-attacks in games, analyze them to see if similar ideas (double attacking squares on key diagonals) preceded them.

The engine will find it eventually but it seems content to repeat every possible position twice en route to the solution.

Thanks, @ambrona , for the demonstration of the correct winning method. The engine is no help at all.

I was happy to help.

The engine may not be useful to understand how to win, but it at least tells you that Black is completely winning.
Of course, there exist positions where the engine will give an incredibly wrong evaluation, but that should never happen in an open position l like the puzzle one.

Really interesting. Thank you ambrona and pheonixshade.
I am annoyed with myself now for talking myself out of the solution in my first post. Basically I solved it and then talked myself out of my own solution, because I hadn't written down clearly enough what I had done.
That is 2 puzzles now that are not broken.