The Hardest Puzzle Ever Composed

to win white is supposed to promote pon to queen?

Solved!!!

https://www.youtube.com/watch?v=ncdsvKP0pPA  

I've seen this one before!?

is this up-side down or what?

and i bet komodo could do it.

I know this puzzle is famous and all, but I actually think its a dud. I know I know, I'm such an armchair GM, but hear my out. As soon as I mention the world "engine" you'll think "THIS PUZZLE IS TOO HARD FOR ENGINES THAT IS YOUR PROBLEM." But I'm not talking about the initial position. The mainline variation has black playing an inferior move, and that is what had me thinking.

As posted on the threat before, here is the mainline variation:

This is the mainline, but it is NOT forced. People conviently like to skip over this variation:

My challenge is for you to prove that this position is in fact a win for white. It seems like a draw, though I am aware that stockfish prunes heavily, so I am legitimately curious, not trying to change everything. I don't know how to set my stockfish so that it spends its time unpruning lines it initially skipped over, instead it is sitting here looking 50 or more moves deep which doesn't help since it already missed the winning variation if there is one.

Can you find a better continuation than stockfish? I'm curious if this win has been proven.

or also by this black gets win

You guys claiming that Black will win are wrong.  Have an engine play the game after the pawn promotes to a knight, you're missing the winning combination (which is a bishop move).

xman720: your postion after 17...Kxg4 is a tablebase win for white.

There is a recent article on this site by GM Gregory Serper with some of the "most difficult" problems, including the one under discussion:

https://www.chess.com/article/view/the-most-difficult-chess-puzzle

But please do not believe that anything you find on this site can be catalogued as the "the most difficult problem" or even "very difficult". There is a universe of beauty and difficulty you will only discover once you join a problemist society in your country. That's where the pro's are, not in the coffee pubs visited by Karpov, Kasparov and Anand.

You guys are talking about the HARDEST puzzle. Can I talk about 2nf hardest puzzle please? The problem is I am on my phone and can't post any diagram. After 2hrs I'll post one. It's IMPOSSIBLE to solve

Actually, 4...Kg4 is better than the routine 4...Nf7+?, which loses spectacularly. Despite this, even 4...Kg4 eventually loses. The puzzle leads to a forced win for White. I remember how amazed I was when I saw the solution for the first time. What a gem!

By chance, I know more about the history of this study than anybody else (outside the composer himself).

It was authored by dutch composer Gijs van Breukelen somewhere between 1973 and 1976. We were members of the same chess club and he showed it to me as a peer composer. I didn't realize at the time what a great piece of work had been crafted and that was very negligent of me.

We lost touch until about 10 years ago I received a message from him with a request to confirm that I had seen his composition in the seventies (someone else was claiming it). Unfortunately, I couldn't recall the precise position which is pretty critical in matters of authorship. Since that time some original memories re-emerged regarding the knight play and the Bishop ending up on a4 and now I am pretty certain that it is and always was Gijs's work.

So after all these years: Congratulations Gijs! It is a wonderful study!

I remember reading about the history of the puzzle on gameknot.com. Interestingly, some people incorrectly claim the puzzle was composed by Jim Plaskett.

Recent pic of Gijs van Breukelen:

http://www.ajedrezdeataque.com/03%20Finales/Listado/Van_Breukelen.htm

Gijs has a minute digital footprint which seems in line with his modest personality. After some searching, I managed to locate him in a small village between 2 of the largest rivers in the Netherlands. It's an area I occasionally pass by on my long distance bike rides and I will try and visit him the next time I'm there. Curious to learn about the current authorship status of "la danza del elefante" that made him famous not just in endgame circles but in the wider chess community as well. I do not know if Gijs still actively creates endgame studies but I did find results indicating that he still plays chess at club and tournament level.

I'm curious, why is it that engines cannot get this? I put Stockfish 7 to the challenge and at depth 43 it evaluated it at like -1.79 and advised d8=Q...? Even after NF6+ Kg7 Nh5+ Kg6 It still couldn't find Be2+... ! (once I showed it Be2 it immediately changed it's evaluation and found the rest)

Engines have trouble finding this puzzle because their default settings prune aggressively, especially stockfish.

You let your stockfish run to depth 43. This is great, but it's useless because stockfish already dismissed the winning move. Somehwere down the mainline of the puzzle solution it gave up to early and decided that the line was losing. Those 43 moves are assumming that the first move played is d8=Q.

The way to get engines to get this puzzle and other "anti-engine" puzzles is to reconfigure them (some engines do this easier than others) so that they prune less and are more "open-minded" to seemingly crazy moves. This will let them calculate down the correct lines and see the winning moves.