Annotated chess game

Just possibly 21...Rxc1 might have been the better option, with the idea 22.Nxd8?? Rxe1ch. White would instead have played 22.Rxc1. Black could then exchange on e6, or else try something a bit more combative: 22...Qf6 23.Nxf8 Qxb2, and, however White plays it, Black will recover the minor piece half of the Exchange, and obtain a zillion passed pawns on the Q-side; just where the bishop pair is aiming. On reflection, White would probably be ill-advised to exchange queens in that line.

I see it with not just discover check, but a masked battery, as well.

Considering how open the board looks, the position is still complicated. The bishop being indirectly protected by the queen and the blockaded pawn, Black might have looked at 23...a4, here. White would have had but one sensible reply: 24.Qd5, whereat comes 24...Rb8 25.Bc6 Bxb2! - and look at those black clouds gathering on the Q-side! This really is a fine game!

A neat strike!

If 27...Rf6, instead of underpromoting, I'd answer with 28.e8=Q. So, why is underpromoting to a N better then a regular promotion.

I can imagine that permitting the exchange of queens would have seemed counter-intuitive, leaving Black with the only piece on the board. But it so happens that White could not have exchanged queens - not safely anyhow: 27...Qc3! 28.exf8=Qch Kxf8 29.Qe2 ... (Not 29.Qxc3?? bxc3 30.Be4 a4, and the bishop can not stop both of Black's running pawns) 29...b3, and Black's two flank passers are well worth the bishop.

Accurate! 28.exf8=Qch would also have been winning, but the text is much more clever.

I would've missed what you posted above, Charles. But, what white plays next is also how I would've responded, as I would think that there's a mate soon to follow after that. Now, I realize that it's not so soon because I was thinking in terms of mate in 2 or mate in 3, but it'll take longer then that.

Taking with the bishop might not look as good as this capture, but it does have its points, possibly. If the king now moves, the game seems to transpose into the Qxf7ch lines. If the BQ moves anywhere but c3 or c1, White picks up the K-side pawns. The check doesn't gain Black much, so what about 30.Qc3, then? It turns out that White can after all afford to exchange queens, here: 30.Qxc3 bxc3 31.Bb3 ... (stymies both a- and c-pawns) 31...d5 (attempted decoy) 32.Kf1. White's K and B will hold up the Black pawns, after which White's K-side majority will win: 32...Kd6 33.Ke2 ... (the king has to get in amongst it ASAP; the K-side pawn majority is in reserve; it can wait) 33...Kc5 (say, to try and drive the pawns home) 34.Kd3 d4 35.Bc2! Kd5 36.f4! h5 37.h3 ... Black is practically in a zugzwang. The White K-side comes into its own. 37...Kc5 38.g4 hxg4 39.hxg4 Kd5 40.g5! ... The exclam is that this seemingly anti-positional move actually places Black in a zugzwang: he's bound to lose something, and he can't come at those White pawns. 40... Kc5 41.f5! Kd6 42.fxg6 ... (42.f6 also wins). The Black king has to abandon his pawns, and still won't be able to reel in the White.