Was today's puzzle easy? I think it was easier than most Sundays, but certainly not worth the "got it immediately" comments that we usually get when the obvious first move wins.

Material: B for N. The White Kg2, Qg4, and Ng6 are on the g-file with the Black Qg7 and Kg8, guaranteeing some tactical fireworks. Such conjunctions favor the player with a piece that can discover an attack, namely, the White Ng6. The Black Kg8 is defended only by Qg7, Bf8, and Bh7, and the other pieces cannot assist Kg8 immediately, so a sacrificial attack can yield a local superiority on the K-side for White. The White Rh1 can capture Bh7, to permit "reloading" by the otherwise inactive Ra1, and possibly to displace Qg7 in some variations, to permit Ng6 to discover not just an attack, but a check on the g-file. The White Be3 can aid the attack, so the only inactive White piece is Bc2.

Candidates (33.): Rxh7

My 35.Rh8+ nets a draw by repetition.

Toga sticks with the game continuation up to:

33.Rxh7 Qxh7 34.Rh1 Qg7 35.Bh6 Qxh6 36.Ne7+ Kf7 37.Rxh6 Bxh6

38.Nf5 Bf8 39.Qh5+ Kg8 40.Qg6+ Kh8 41.Qxf6+ Kg8

It then deviates with

[plies 15/51 time 02:42 value +1.42]

<42.f4> Rb7 43.fxe5 dxe5 44.Qxe5 Raa7 45.d6 Rd7 46.Qf6 Rab7

47.b4 Nc4 48.Bb3 Nc6 49.Qe6+ Kh7 50.Bxc4 bxc4 51.Qd5 Rb6

52.e5 Bh6

When fed the game continuation, however,

<42.Qg6+> Kh8 43.Qe8

Toga likes it better than the "best play" at 15 plies given above:

[plies 15/46 time 00:11 value +3.36]

43...Nb7 44.Qf7 Nd7 45.Nh4 Bg7 46.Ng6+ Kh7 47.Qf5 Nf6

48.Nxe5+ Kg8 49.Qe6+ Kh7 50.Nf7 Nc5 51.e5+ Nfe4 52.Ng5+ Kh8

53.Qh3+ Kg8 54.Qh7+ Kf8 55.Bxe4 dxe5 56.Qf5+ Kg8 57.Qf7+ Kh8

However I am currently investigating better moves then 28 Nxg7 and here the computer is able to suggest some. Just increase the pressure with Kh2, Qh5, and doubling rooks on the g-file. Very prosaic but nevertheless much more powerful then the knight sacrifice. With 28 Kh2 Nf7 29 Qh5 Bf8 30 Rg1 Rb7 31 Rg3 Qxc3 32 Rag1 g5 I have just reached an evaluation of 5.87/20. If I would just know how to harvest that now... probably Bxg5.

click for larger view

Wasn't there really nothing better then 31 ... Qxc3? This move looks utterly wrong to me.

Hi, <Gregor Samsa Mendel>. Thanks for making the comment. I infer that generally, a N is preferable to a B when local superiority is required, e.g., for today's style of K-side attack. Presumably, the local superiority reflects the N's ability to control squares of both colors.

By the way, I have a different line here which I would like you to check out> 33. B x N. Then, if ... Q x N; Q x Q, B x Q later bringing both rooks on the king side ; if Black tries the weaker.. R x B; there follows R x B, Q x R; R - KR1, Q - KN2, R-KR8+ with the same Q +B duo

or

33. B-KR6 (Bh6), Q x N; Q x Q, B x Q; B x B, K x B, R x B once again with te ideabringing the rooks over to give mate

I have frequently been guilty of ISPUB.

According to Toga II 1.3.1 (with the full variation, which a human can improve near the end),

[plies 15/52+ time 01:11 value +5.05]

35.Kxh1 Ra7 36.Ne7+ Kf7 37.Nf5 Rab7 38.Qh5+ Kg8 39.Qg6+ Kh8

40.Qxf6+ Kg8 41.Kg2 b4 42.Qg6+ Kh8 43.Bg5 Nd7 44.Bxd8 bxc3

45.Qh5+ Kg8 46.Bg5 Rb6 47.Qg6+ Kh8

The game is a computer-killer (see posts below). If the computer is to be believed, there is no quick kill, but the variation starting from 34...Qxh1+ is like the game, except White retains Be3 (which is not going to make Black's life easier).

I found three aspects of this difficult Sunday puzzle solution particularly interesting:

1. The combination is apparently set up by the previously played demolition of pawn structure sacrifice 28. Nxg7!!. Indeed, the initial play after 28. Nxg7!!, until we reach the puzzle solution position with 33. Rxh7!!, appears to involve all forced best moves.

2. White's two best replies to 33. Rxh7! transpose, since either 33. ...Qxh7 34. Rh1 Qg7 or 33..Kxh7 34. Rh1+ Kg8 both reach the same position and represent the only reasonable lines of play (other replies lose quickly).

3. Black appears to be holding the balance with strong best responses for the first nine moves of this combination -- that is until White plays the stunning surprise move 43. Qe8!! . Indeed, as <johnlspouge> noted <this position is a computer killer> as some computer analysis (including my Fritz 8 @ 15 ply or so) indicates the position to be level even after 42...Kh8. However after the strong 43. Qe8!! is played, Black's opponent and the computers (after the move is forced on them for evaluation) quickly realize the strength of White's decisive threats.

P.S.: Just for fun, here's a few possibilities after the stunning 43. Qe8!!:

If 43... Nb7, then White wins with 44. Qf7! when play might continue 44...Re8 45. Qxb7 Rab8 46. Qf7 Rbd8 47. Nh4 Bg7 48. Bd1 Nd7 49. Nf5 Bf6 50. Bg4 a5 51. Qh5+ Kg8 52. Nh6+ Kh8 53. Be6 Re7 54. Nf7+ Kg8 55. Nxd8+ Kg7 56. Bxd7 Rxd7 57. Ne6+ Kg8 58. Qe8+ Kh7 59. Qxd7+ .

If 43...Kg8, as in the game continuation, then White scores the full point with 44. Ne7+ Kg7 45. Qg6+ Kh8 46. Qg8#.