Sep2507
  SwitchingQuylthulg: A pleasing rook sacrifice very similar to that seen in the Puzzle of the Day (Boleslavsky vs B Goldenov, 1947).
This one is much more complicated though as White's mates are not immediately obvious and Black threatens ...Qxd1 all the time. 24.Rxg7+! still leaves Black without defense, for example here click for larger view 25...Kf8 fails against 26.Nd7+! Nxd7 27.Bg7+ Ke7 28.Qg5+ f6 29.Bxf6+! Nxf6 30.Qg7# while 25...Ng4 26.Nxg4+ f6 is met by 27.Nh6+! Kxh6 28.Qh4+ Qh5 29.Qxf6+ Qg6 30.Qh4+ Qh5 31.Bg7+ Kxg7 32.Qxh5 and 25...Ng4 26.Nxg4+ Qxb2 by 27.Ne5+ Kf8 (27...Kf6 28.Qf4+ Kg7 29.Qg5+ Kf8 30.Nd7#) 28.Nd7+ Ke7 29.Qh4+ f6 30.Qxh7+ Kd8 31.Nxc5+ and mate next. If 27...Kxg7 then White finishes with 28.Ne5+ Kf8 29.Nd7+ Ke7 30.Qg5+ f6 31.Qxf6#. 

May0920   stacase: 24.Rxg7+ was the only move to consider 25 and 26 are automatics but then what? 27.Bg7+ another sacrifice? or 27.Ne3. Black's answer to Ne3 would have been 27...e5. and White's loss. As it turned out, all of Black's responses to the 2nd sacrificial offering don't work. 

May0920
  al wazir: I figured 24. Rxg7+ had to be the first move, but I didn't see the continuation. Not even close. 

May0920   Brenin: The lines of forces converging on g7, together with Black's threats on d1 and f2, make 24 Rxg7 the obvious move. After Black's only reasonable reply Kxg7 25 Qg3+ Ng4 26 Nxg4+ Kf8 (other Black responses are suicidal) White needs another focing move, and only 27 Bg7+ does the job. The role of the R on d1, closing the mating net, is critical here. 

May0920   Brenin: 26 ... f6 puts up more resistance, but then 27 Nh6+ wins, e.g. Kxh6 28 Qh4+ Qh5 29 Qxf6+ Qg6 30 Qh4+ Qh5 31 Bg7+ Kxg7 32 Qxh5, with Q+2P v R+B. 

May0920
  scormus: I was thrown off course in my efforts, by 25 .... Ng4. I'd been analyzing .... Kf8, after which W has a straightforward #ing sequence. .... Ng4 doesn't fair any better, but W's continuation is a bit less obvious, especially 27 Bg7+ 

May0920
  agb2002: Black threatens Qxd1+ and Bxf2+.
The first idea that comes to mind is 24.Rxg7+:
A) 24... Kxg7 25.Qg3+
A.1) 25... Kf8 26.Nd7+ Nxd7 (26... Ke7 27.Bxf6#) 27.Bg7+ Ke7 (27... Kg8 28.Bf6+ and mate in two) 28.Qg5+ f6 29.Bxf6+ Nxf6 (29... Kf7(8) 30.Qg7#) 30.Qg7#. A.2) 25... Kh6 26.Nxf7+ Kh5 27.Qg5#.
A.3) 25... Kh8 26.Nxf7#.
A.4) 25... Ng4 26.Nxg4+
A.4.a) 26... Qxb2 27.Ne5+
A.4.a.i) 27... Kf8 28.Nd7+ Ke7 29.Qg5+ f6 30.Qg7+ Kd8 31.Nxc5+ and mate next. A.4.a.ii) 27... Kf6 28.Qf4+ Kg7 (28... Ke7 29.Qxf7#) 29.Qg5+ Kf8 (29... Kh8 30.Nxf7#) 30.Nd7#. A.4.a.iii) 27... Kh6 28.Nxf7+ Kh5 29.Qg5#.
A.4.a.iv) 27... Kh8 28.Nxf7#.
A.4.b) 26... e5 27.Nxe5+ Kf8 (else as in A.4.a) 28.Nd7+ Ke7 29.Bf6+ Ke6 30.Qh3+ Qg4 31.Qxg4#. A.4.c) 26... f6 27.Nh6+ Kxh6 (27... Kf8 28.Qg8+ Ke7 29.Qf7#) 28.Qh4+ A.4.c.i) 28... Qh5 29.Qxf6+ Qg6 30.Qh4+ Qh5 31.Bg7+ Kxg7 (31... Kg6 32.Qf6#) 32.Qxh5 +  [Q+2P vs r+b]. A.4.c.ii) 28... Kg6 29.Qxf6+ Kh5 30.Qf7+
A.4.c.ii.1) 30... Kh6 31.Bc1+ wins decisive material. A.4.c.ii.2) 30... Kg5 31.Bc1+ Kh4 (31... Kg4 32.Qf4+ Kh5 33.Qg5#) 32.Qf6+ Kg4(h5) 33.Qg5#. A.4.c.ii.3) 30... Kg4 31.h3+ Kg5 (31... Kh4 32.Bf6#) 32.Bc1+ wins. A.4.c.ii.4) 30... Kh4 31.Bf6+ Kg4 32.Qg7+ Kf4(5) 33.Qg5+ Ke4 34.Qe5#. A.4.d) 26... Kf8 27.Bg7+
A.4.d.i) 27... Kxg7 28.Ne5+ as in A.4.a.
A.4.d.ii) 27... Kg8 28.Nf(h)6#.
A.4.d.iii) 27... Ke7 28.Qh4+ f6 29.Qxf6#.
A.4.e) 26... Kg6 27.Nh6+ Kxh6 (else mate in two) 28.Bc1+ wins decisive material. B) 24... Kh8 25.Nxf7+ Kxg7 26.Qxf6+ and mate next. C) 24... Kf8 25.Nd7+
C.1) 25... Nxd7 26.Rxf7+ Kxf7 (26... Kd8 27.Rd(f)xd7#) 27.Qg7#. C.2) 25... Kxg7 26.Qxf6+ Kg8 27.Qg7(h8)#.
C.3) 25... Ke7 26.Qxf6#. 

May0920   Brenin: <agb2002>: Wow! That's what I call detailed analysis. Thank you. 

May0920   Delboy: Very good for someone with a nonskyhigh rating to see all this in actual play. By the way, the win in the 27. ... Kxg7 28. Ne5+ Kf6 line is not 29. Nd7+ (which may even lose after 29. ... Kf5), but 29. Qf4+ 

May0920   Jambow: <I figured 24. Rxg7+ had to be the first move, but I didn't see the continuation. Not even close.> Bout sums it up for me too, although I seem to miss lateral slides in puzzles and games lately all too often. As soon as I saw 23. Qg3+ it was like yes but of course. Nice puzzle cg.com 

May0920   mel gibson: I saw the first move within 3 seconds
and I could tell that Black was in big trouble.
Stockfish 11 says mate in 21:
24. Rxg7+
(24. Rxg7+
(♖g3xg7+ ♔g8xg7 ♕c3g3+ ♘f6g4 ♘e5xg4+ f7f6 ♘g4h6+ ♔g7xh6 ♕g3f4+ ♔h6h5
♖d1d5+ e6e5 ♕f4f5+ ♔h5h6 ♕f5xf6+ ♔h6h5 ♕f6f7+ ♔h5g5 ♗b2c1+ ♗c5e3
♗c1xe3+ ♕e2xe3 ♕f7g7+ ♔g5f5 g2g4+ ♔f5e4 ♕g7xh7+ ♔e4f3 ♕h7h3+ ♔f3f4
f2xe3+ ♔f4g5 ♕h3h5+ ♔g5f6 ♕h5f5+ ♔f6g7 ♖d5d7+ ♖e8e7 ♖d7xe7+ ♔g7h6
♖e7h7+) +M21/73 235) 

May0920
  perfidious: The initial offer is obvious, due to Black's simple threat, but some of the play which follows is far from easy to foresee; in particular, the continuation 26....f6 appears to offer succour, but avails him naught. 

May0920   LewisKing: Still not seeing the win after 27. ... Kxg7 28. Ne5+ Kf6 29 Qf4+ Kg7 30. Qxf7+ Kh8 

May0920   Brenin: <LewisKing>: 30 Qg5+, instead of Qxf7+, and mate next move by Nxf7 or Nd7. 

May0920
  chrisowen: Cornered no? 

May0920
  Predrag3141: The puzzle about a week ago where it looked like White could proceed with the attack but he had to pause for Re3 made me think this could be the same. Pause for 27 Rf1, perhaps? But I started to see no such move would survive against 27 … Ne4. So, the sacrifice on g7 is a must. I was not worried about 27 Rg7+ Kg7 28 Qg3+ Ng4 29 Ng4+ because White regains most of his material with check, but actually he again loses against threats on f2 and d1 unless he can finish Black off. So I wrongly focused only on 28 … Kf8 and found 29 Nd7+. I guess I can take half credit. 

May0920   TheaN: A very long, but otherwise forcing Saturday that is pretty obvious for the get go. White <only> has Rxg7+ and the domination over the dark squares begins there. <24.Rxg7+> I don't think this deserves a ! as I don't see what White would do otherwise. The alternative for Qxf2+ and Qxd1+ is Rf1, but now Ne4! + suddenly does work as the knight's covering g3. Can Black decline this sac? Sometimes declines are tougher than accepts, but not here: to the corner is easy 24....Kh8? 25.Nxf7+! Kxg7 26.Qxf6+ Kf8 27.Nh6# but to the center doesn't work either. 24....Kf8? 25.Nd7+!, now that's a killer move: 25....Nxd7 (Ke7 26.Qxf6#; Kxg7 26.Qxf6+ Kg8 27.Qg7#) 26.Rxf7+! with 27.Qg7#. So <24....Kxg7 25.Qg3+> the same occurs: can Black move away? No. 25....Kh8?? 26.Nxf7#; 25....Kh6?? 26.Nxf7+ Kh5 27.Qg5#. 25....Kf8? is slightly more complicated but follows the same principle on the rook sac: 26.Nd7+! Nxd7 (Ke7 27.Bf6#, in this pattern realize the beautiful task of Nd7) 27.Bg7+, bishop first is crucial here as you need dark square fire power, Ke7 (Kg8 28.Bf6+ Qg4 29.Qxg4+ Kf8 30.Qg7#) 28.Qg5+ f6 (Nf6 29.Qxf6#) 29.Bxf6+ with 30.Qg7#. Lets not analyze 25....Qg4. If this is a way to prevent mate, it's going from death to torture. <25....Ng4 26.Nxg4+>. The obvious moves are 26....Qxb2 or 26....Kf8. I'll get back to those in tandem. How do the other king moves fair? 26....Kg8?? 27.Nh6+ Kf8 28.Qg8+ Ke7 29.Qxf7#; 26....Kg6 looks like suicide, though I have to be honest I didn't calculate it completely: 27.Nf6+ Kf5 (Kh6 28.Qh4+ Kg6 similar to main line) 28.Qh3+ and here my vision started to fail me a bit but of course 28....Kg6 (Kf4 29.g3+ Kf3 30.Qg2#) 29.Qxh7+ Kg5 30.h4+ (I had Bc1+!? now which wins too) Kf4 31.Bc1+ Ke5 (B/Qe3 32.Qe4#) 32.Nd7#. Pawn moves? White went from a discover into a discover threat with Ng4. 26....e5 does nothing after 27.Nxe5+; White went from e5 to g4 to e5, and e6 isn't a proper escape square. This mates exactly the same as after 25.Qg3+ K move, but without the Black knight to prolong mates. I'll get back to 26....f6! However, 26....Kf8 and 26....Qxb2 first. They're the obvious choices, as f6 looks incredibly eerie. I analyze them in tandem as the followup is incredibly similar. 26....Qxb2 27.Ne5+ is the easy followup and analysis will follow later. 26....Kf8 looks so much better though, no discover, threatening two pieces. <Now> comes the Saturday part. Bb2 in, Rd1 in, pressure on f2 and an exchange down. White's down and out?
click for larger view
27.Bg7+!!: in the Qxb2 variation Bb2 isn't even there. So we don't need it, and we lure the king back to the gfile. 27....Kxg7 (Kg8 28.Nf6#; Ke7 28.Qh4+ f6 29.Qxf6#) 28.Ne5+. And <now> we're on a similar path with 26....Kf8 and Qxb2, just with the queen on e2 instead of b2 which doesn't make a difference. 28....Kf6 (Kh8? 29.Nxf7#; Kh6? 29.Nxf7+ Kg5 30.Qg5#; Kf8? 29.Nd7+ Ke7 30.Qg5+ f6 31.Qxf6#; Qg4 29.Qxg4+) 29.Qf4+ Kg7 (Ke7 29.Qxf7#) 30.Qg5+ and the dominance over the dark squares comes to a conclusion; White built a staircase on g3f4g5, and finishes after 30....Kf8 31.Nd7# or 30....Kh8 31.Nxf7#. So, what about <26....f6!>? In fact, Black's <only> nonmated reply. Enabling f7, a light square, White will have to thread carefully to come out with a winning advantage. Bxf6+? for example, is wrong! <27.Nh6+!> the only move that keeps tabs on f7 so Black's forced to <27.....Kxh6 (Kh8 28.Nf7#; Kf8 28.Qg8+ Ke7 29.Qf7#) 28.Qh4+> threatening Qxf6+ to still win on the dark squares. <28....Qh5> if 28....Kg6 a long chase over the dark squares begins 29.Qxf6+ Kh5 30.Qf7+ Kh6 31.Bg7+ Kg5 32.Rd5+ exd5 33.Qf6+ Kh5 34.Qf5+ Kh4 35.g3#, however, after Qh5 White is <forced> to block off f7 forever and will only go up material with <29.Qxf6+ Qg6 30.Qh4+ Qh5 31.Bg7+> delivering the coup de grace <31....Kxg7 (Kg6? 32.Qf6#) 32.Qxh5 +> and when the huge dust cloud settles:
click for larger view
we are winning, but White has some work cut out for him still. Saturdays. Sigh. 

May0920   Brenin: It would be interesting to know how much of this Kanki saw when playing 23 Qc3, which lured the Black Q away from defence of the Kside to the attractivelooking e2, where threats such as 24 ... Qxd1, Bxf2+, Qxf2+ or Ne4 look unstoppable by White. Instead, 23 ... Qf5 would have maintained the pressure on f2, and defended the Kside, e.g. by rendering Qg5+ harmless in lines such as <agb2002>'s A.1. It would also have set a subtle trap: after the naturallooking 24 Ng4 Red8 25 NxN+ QxN, the threats on d1 and f2 mean that Black is winning, e.g. 26 Rxg7+ Kh8! 

May0920   Cheapo by the Dozen: What <al wazir> said. I didn't see how to unclutter the long diagonal and I didn't see the power of Nd7+ in various lines. 

May0920
  Check It Out: I did see 27.Bg7+ as a way to keep the attack going, but no way did I know if it worked. 
