Through trial and error,I finally saw g-6 which attacks the Queen, blocks the Rook from getting to f-6, gains a tempo allowing the White Queen to come crashing through to "crush da enemy"

As Dzechiel pointed out, I thought about playing 23.e4 so that 24.Qd2 (forking rook and knight) could be played but this runs into 23...Qc5 and the white queen is pinned.

23.Bh7 Qxd1 24.Rxd1 Rxh7 and white has lost a bishop and rook for queen--not a huge advantage but perhaps white can play 25.Qd2 Nc6 26.Qxd6+ Ke8 and 27.Qxf6. But I like <alshatranji>'s line better: 23...Qxg2+ 24.Qxg2 Rxg2+ 25.Kxg2 Rxh7--simple and probably best with approximate equality.

23.Bg6 is promising because it has a double-threat of Rxd5 and Qxf6+. At first glance it looks like 23...Qe6 stops both threats, but 24.Bf5 and the queen is trapped. So one try is 23...Qxd1 24.Qxf6+ Kd7 (24....Kd8 or 24...Kc8 25.Qxg8+ is an easy win) 25.Qxf7+ Kc6 26.Be4+ and 27.Rxd1.

However, none of this make much sense to me given the position. With the center essentially locked (at least until Black opens it), as well as four out of five of White's major pieces on the queenside and king side castling in White's future, I would have thought beginning a king side attack on White's already weak king side would have been more advantageous. Black places rook(s) on the g-file, add the bishop and h pawn and Black has a credible king side attack. The only real issue would have been to get the queen involved.

15. ... Nb4 followed by d5 would have been a better play aiming for a king side attack. Even the immediate 15. ... h5 followed by 16. ... h4 sacrificing the f6 pawn would have been more playable. In short, what was Shirov thinking with 15. .. Na5? ??? lol

After 16. 0-0 Ke7 17. Qe1 h5 18. b3 Qc5 19. Nd5+ Bxd5 20. exd5 Qxd5, White plays the obvious 21. Rd1. At this point, the warning lights should have been going off for Shirov. He plays 21. ... Rcg8 but in response to 22. Qf2 he plays 22. ... Rh6. Slightly better would have been 22. ... Qc5 or Qc6. They are both losing as well after White plays b4 and then Black's central pawns are mowed down. But, Black gets to keep his queen and White's task is anything but easy given that Black still has both rooks and his queen.

Looking at the puzzle position, I noted five things. One, Black's knight is just a bystander and moreover, it actually competes with Black's queen should it need to move. Two, Black's queen is resting unguarded on a white square subject to a disclosed attack by White's rook. Three, Black has virtually no control over the light squares. Four, Black's king and parallel rook are protecting the doubled f6 pawn which is being attacked by White's queen and rook. Five, Black is threatening mate via the rook on g8 and queen on d5 (therefore, if White is going to move his queen, he must check the Black king).

Given the above, a bishop move seemed obvious as the strongest candidate moves. Among them were, Bxa6, Bh7 and Bg6. Bxa6 nets a pawn and not much else. Bh7 forces an exchange to a two rook vs. knight and rook ending. Probably winning for White, but Black's central pawns coupled with White's weak king side make it anything but easy, much less conclusive.

Observations: My queen wants to take Pf6 in the worst way, but it's guarded by Rh6. My Pg2 is facing danger from the black queen and Rg8. I would love to move my bishop somewhere useful so that my discovered attack on the black Q would have some bite.

When all of this is taken into consideration, the g5 square starts flashing. By plunking my bishop there with 23.Bg5, I block both rooks from doing their thing, threaten Qxf6+, and force Black to move his queen.

But where? Not 23...Qe6 (to guard Pf6) since 24.Bf5! traps the queen. Other Q moves are simply lame because they allow 24.Qxf6+ with massive losses to follow.

<Maclouly> has pointed out that Black gets what appears to be a playable game with <xg2+! xg2 hxg6 xg6 xg6+> -- It's all rather forcing that I can see... and Black emerges down the exchange but up two pawns. By material count it's even... and with all those pawns on the board... it's way too early (for me at least) to resign.

Damn! you're right!

I nexted the Bg6 move (however I tought it was the best move)because of Qe5, and I didn't see this Bf5 infernal trap !!!

Today it the second move was the most difficult.

<23.?>

White: a2, b3, c2, e3, g2, h2, Bd3, Rd1, Rf1, Qf2, Kg1

Black: a6, b7, d6, e5, f7, f6, h5, Na5, Rg8, Rh6, Qd5, Ke7

Material: -+/

Candidates: <[23.Bg6]>

DISCOVERY. ON. QUEEN. It was the first alert that jumped out from the position. And to be honest, the solution is not very difficult afterwards. We want to use it, and want to take out the defender of f6. We can do that, in one blow.

-ML-
<23.Bg6> take note that 23.Bh7?! allows 23....Rxg2†! 24.Qxg2 Qxg2† 25.Kxg2 Rxh7 with some problems for White. Also winning, but this threat is way more severe. What can Black do here? Moving the Queen in order to safe f6 is deadly. Btw, I'll not consider any move that gives away the Queen for nothing. Ke6, due to Qxf6†, falls in that category, and any other move not depicted below.

/A\
<23....Qe6 24.Bf5 > whoops. Bye bye Queen. Critical move, though.

/B\
<23....Qc5> or Qc6 or Qb5, doesn't really matter; it might allow a check or two more so I think that in this category -Queen move without guarding f6-, Qc5 is best.

<24.Qxf6† Kd7 (Ke8 25.Qxf7† Kd8 26.Qxg8† Kd7 27.Qg7† or Kf8 25.Qxf7‡) 25.Qxf7† Kc6 26.Qxg8 > Black is lost: the King position is not really in Black's advantage, with a Rook down. Qxe3† with Kh1 and the pieces can go back in the box.

/C\
<23....Qxd1> original idea: getting Rook and Bishop for the Queen. White should not be greedy though.

<24.Qxf6† Kd7 (Kf8 25.Qxf7‡)> with Ke8, the same as in B happens forced, after which Rxd1 follows: 24....Ke8 25.Qxf7† Kd8 26.Qxg8† Kd7 27.Qg7† Kc6 28.Rxd1 .

<25.Qxf7† Kc6 (Kd8 26.Qxg8†) 26.Be4†!> a nice little intermediate move that does not allow Black to get Rook+Bishop, but only Rook for the Queen.

<26....d5 (Kc5 27.Rxd1 or Qd5 27.Qxd5† ) 27.Rxd1 > and is impossible due to 27....dxe5 leading to 28.Qxg8, so White keeps the Bishop. +/ with a deadly position wins easily.

I thought 23.Bg6, Qxd1 24. Rxd1?! (24.Qxf6+ wins easily), Rhxg6 25.Qd2,Rxg2+ 26.Qxg2,Rxg2+ 27.Kxg2 with a won endgame for white IMO which is why I didn't try and improve the line earlier on.

Also 23.Bg6,Qxg2+ 24.Qxg2?! transposes into the same endgame 24..Rhxg6 25.Rf2,Rxg2+ 26.Rxg2,Rxg2+ 27.Kxg2 but black can improve here and keep the other rook on putting up a much tougher fight. (e.g. 26..Rc8 or h8 maybe even g6?! draws after 27.Rxg6,fxg6 and no invasion squares...)

< Domdaniel: Major blind spot here. I saw no way through on the f-file, tried toying with fantasy lines involving a rook sac on d6, and reluctantly decided there was nothing better than the dubious 23.Bh7>

I initially thought 23.Bh7 as well, but 23..Qxg2 ruined the idea.

There are lots of half tries in this position though.

I also briefly considered 23.e4 planning on Qd2 attacking both Na5 and Rh6 but 23..Qc5! ruins that idea.

Lots of ideas to sift through in this, nice puzzle.