patzer2: Here's my look at the game and the Sunday puzzle (24...?) with the chessgames.com Opening Explorer (OE) and Deep Fritz 14:
<1. e4 c5 2. Nf3 d6 3. d4 cxd4 4. Nxd4 Nf6 5. Nc3 g6 6. Be3 Bg7 7. f3 O-O 8. Qd2 Nc6 9. Bc4> This is the most popular move at this juncture in the Sicilian, Dragon, Yugoslav Attack (B78).
The second most popular move and the Fritz preference is 9. O-O-O as in Caruana vs Nakamura, 2015.
<9... Bd7 10. O-O-O Rc8 11. Bb3 Ne5 12. h4> This is the most frequently played move in the OE database. It was once the favorite of
Super GMs, but started to fall our of favor after
Anand vs Kasparov, 1995.
The second most popular move is 12. Kb1. It is more flexible and less committal than 12. h5, and is the Fritz first choice. Perhaps that's why 12. Kb1 is now the choice of Super GMs, as evidenced by
Anand vs Carlsen, 2008.
The fact White wins more with 12. Kb1 than with 12. h4 doesn't hurt its popularity. The chessgames.com OE database shows White won 52.1% of 259 games with 12. Kb1, as compared to 42.6% of 371 games won with 12. h4.
<12...h5 13. Bh6> Here 13. Bh6 is the second most popular move. Most often played is 13. Bg5 as in
Karjakin vs Carlsen, 2014.
The Fritz preference 13. Kb1, as in L Bruzon Batista vs V Cmilyte, 2012, is much less popular. This might be related to the fact Black has won more games (46.7% of 60 games in the OE) than White (33.3% of 60 games in the OE) after 13. Kb1.
<13... Nc4!> This strong move gives Black the best results of any try,
with Black winning 34.2% and losing 31.6% of 38 games in the OE after 13...Nc4.
In the not so distant past, the most frequently played move was 13... Bxh6 as in C Aravindh vs J Herman, 2011. However, 13...Bxh6 is now out of favor, and has since been replaced by 13...Nc4 as the popular choice.
<14. Bxc4 Rxc4 15. Bxg7 Kxg7 16. g4 hxg4 17. h5 Rh8 18. hxg6 fxg6 19. f4 e5 20. Rxh8?> This is the only game with this move in the OE. And rightly so, as 20. Rxh8? is a weak move which turns the game strongly in Black's favor.
Better is the Fritz choice 20. Nde2 = as in C de Holanda vs M Texeira, 1999.
<20... Qxh8 21. Nde2 Rc6 22. Qe3 b6 23. Nd5?!> This appears to
facilitate Black's plan to simplify to a won endgame with connected passed
Better perhaps is 23. fxe5 when White can maximize resistance and fight for a draw against Black's isolated (i.e. no longer connected) kingside pawns. After 23. fxe5, play might continue 23...dxe5 24. Kb1 Be6 25. Ng3 Qh6 26. Qd3 Qh2 27. a3 Bc4 28. Qe3 Qh6 29. Qe1 Qf4 30. b3 Be6 31. Rd3 Nh5 32. a4 Qf7 33. Nd5 Nf4 34. Rd2 Rc5 35. c4 Nh3 (-0.75 @ 21 depth).
<23... Qh3! 24. Ng3 Nxe4!!> This initiates the combination which solves today's Sunday puzzle.
<25. Nxe4 Qxe3+ 26. Nxe3 exf4> Now Black's plan of sacrificing a piece for two connected passed pawns becomes
<27. Nd5 Bf5 28. Rd4 f3 29. Ne3 Bxe4 30. Rxe4 g3 31. Rg4 g2 32. Nxg2>
If 32. Kd2, Fritz indicates Black wins after Rc4! 33. Rg3 Ra4 34. a3 Re4
35. Nxg2 fxg2 36. Kd3 Re5 37. Rxg2 g5 38. b3 Kg6 39. a4 Kf5 40. Rf2+ Ke6 41.
Rg2 Kf6 42. Kd4 Kf5 43. Kd3 g4 (-1.92 @ 24 depth).
<32... fxg2 33. Rxg2 d5 34. Kd2 Rf6 35. Rg3 Rf2+ 36. Kd3 Kf6 37. Rh3 g5 0-1> White resigns this correspondence game, realizing Black has a won ending.
However, at first glance the win isn't so clear. Fritz had to get up to 21 depth before it found a decisive follow-up.
Fritz assesses it as a win for Black after 38. Rh8 Ke5 39. Re8+ Kf4 (-2.22 @ 25 depth) when play might continue 40. Ra8 g4 41. Rxa7 Kf3 42. Rf7+ Kg2 43. Rb7 Rf6 44. Kd4 g3 45. Kxd5 Rg6 46. Rf7 Kh3 47. Rh7+ Kg4 48. Rh1 Rg5+ 49. Kd4 g2 50. Rg1 Kg3 51. c4 Kf2 52. Rxg2+ Rxg2 (-6.96 @ 25 depth).