Annotated chess game

Yes, but don't tell me that this move was foreseen when playing Qxh7+  )

I was wondering the same thing! Astonishing recovery. But looking back from here, Black need not have opened up the position by 19...f5, but instead offered an exchange of Knights by 19...Ng6! When material ahead, it is often good policy to exchange pieces (not pawns so much if there aren't many of them). If White declined the offer, the knight would have been well placed to support the pawn-break ...e5, throwing White back onto the defensive.

And Black has seen that if the King strays onto the g-file, White unmasks a deadly battery: Nxe6ch.

In order to make the game more interesting, black could have played NxNf5, in sight of Rg5+ if black saves the queen.

I think White does well to keep it simple: 23.Nxf5 Qf6 24.Rg5ch Qxg5 25.Nxg5 exf5, and black retains a winning edge.

Although I quite like this recapture, Black might have been better advised to take with the Knight, forcing White to concern himself with the fork threat ...Nf3ch , and possibly the threat of ...Bxh2 as well. That extra moment of time this would have bought could have been used to bring the other knight out to f6 thence to e4. Even though the Black c-pawn would have been lost, Black would have been quite well placed, I think.

Here, black apparently loses the bishop after Rh1. If black goes Kg6, white has the nice e4!, threatening Bf4+, winning the Bh2. Black can't avoid his fate, but it interesting to try Nh4 (a move from check), Nf6 (it fails to exNf5, defending the Ne6...). Anyway: the apparently worse Kh6, tempts white to go Nf4+. This check won't give much to white, and I think that allows black to save the bishop. Kg2 is "threatened", so white has to go something like Kf2. Now Nf4+ wins the bishop, so black is not saved by Nf6 (but it is interesting to see throug it). It seems that Rh1 wins for white, if I am not omitting something.

The position is becoming quite strange - in an interesting way! Both sides might well find themselves with difficulties mobilizing their Q-side pieces.

This is starting to look very horrible for Black...

To summarize: 28.Rf2ch! Kh1! 29.Nf3... [A] 29... Nf1 (to stop 30.Rxh2#) 30.Kxf1 ... (renewing the Rxh2# threat) 30...Bd6 (it doesn't matter where the bishop goes, even g1, the result is exactly the same) 31.Rg2! ... (Broadening the attack. Now White threatens 32.Rg1#. The only defence Black has against it? 31...Bh2...); [B] 29...Bg1 30.Ba3 Bxf2ch 31.Kxf2ch Nf1 32.Rxf1#.

A shambolic position!

Totally surrealistic!   It was an exchange of courtesies among the players...

Instead of ...Nf1, 32...Nb6! was indicated, protecting the c-pawn, and freeing up Black's Q-side.

It is possiblle Black was hoping after 35.Rxd7 to weave some kind of mating net. This actually looks feasible: Position after 35.Rxd7 35...Bd3 36.Rg7 ... (threatening to unmask a battery against the Black King) 36...Bc2ch 37.Ke2 Ng3ch 38.Kd2 Nf1ch 39.Ke2 and a draw by repetition. So it seems that 34...Rc8 was a very clever move!

I must admit, I missed that drawing resource!