Homemade studies [1]: Constrictor

This is the first study that I composed, and it had a larger impact than I expected. It has been discussed intensively in some chess forums and in the near future it will be part of a chess studies book... I'll attach references later. Enjoy! [Hint] Computers may (badly) misevaluate this position long time into the solution and I enjoy mocking the chess engines.

White to move and win.

Well, it looks like White's main obstacle is that he is about to lose a piece, upsetting his material advantage. If the d3-Knight is defended, Qg7+ picks up the Rook. I initially thought 1.Nf4 may be a good try, as 1...Qg7+ 2.Rg6 hxg6 3.Ne6+ would pick up the Queen, but then 1...Qd2+ complicates matters again.

However, this does lead us to some interesting lines. 1.Nf4 Qd2+ 2.Kg3 Qxc2 gives us a tempo, as White has no checks. In fact, it may even be possible to try 3.Kg4, allowing the King to participate in a possible attack. This could get complicated, and I don't have time right now to look at it in depth.

Indeed, white loses a piece in maximum 2 moves, but has the possibility to choose which one to give up... Except for the solution, White can barely hold for a draw, albait Black's counterplay is complex and by no means obvious. By comparison, the solution is strikingly elegant.

[Hint] It may not help much at this stage, but it may be intriguing to know that IF Black's pawn on a6 were back on a7, THEN White would not be able to win with best defence.

Wow, this is hard.

And welcome to the group, Kornrade!

I think this lies outside my meager solving ability.

You may have simply approached it the wrong way .

[Hint] There are 3 concepts that I squeezed inside this puzzle, all very unconfortable to chess engines, but quite natural to humans: [I use small white font in case someone wants to try it without this hint]

>>>

* fortress: apparently, in this puzzle, the strongest chess engines don't have so much of a problem with the fortress itself, but with the fortress as part of a greater plan

* long term strategy: make a simple plan and stick to it, even if it takes 20-30 moves or more to accomplish it. Here, the plan has 2 or 3 simple steps, easy to understand for humans.

* zugzwang: this is the most abstract and confusing aspect for computers. Of course, the great number of variations helps with the silicon confusion, but not so much for humans if they stick to the plan.

SPOILER here is the story of the puzzle, and some links starting from there also contain the full solution.

<<<

Of course, you can use computers to help you with the analysis. They won't spoil the fun

I cant get it.

I gave up and looked at the solution. I haven't looked at all of the lines yet, but the idea is brilliant! I would not have found that on my own.

Here are some of the variations, with explanations for better understanding.

Overwhelmingly hard until you know what the concept is. Then it becomes just very hard.

Beautiful one. The idea behind it is fantastic.

Wow. That is beautiful. Really one of the nicest puzzles I've ever seen.