The choice between 35.Bxf5 or 35.Nxf5 is easy, if you consider which piece needs activating. I did not think even once about the candidate 35.Nxf5.

[ply 15/59 time 03:06 value +2.97]

35.<Bf5> e5 36.Rh4 e4 37.Bxg6 Re7 38.Qh5 Nxd5 39.Nxd5 Qf3+ 40.Kf1 Qxh5 41.Nxe7+ Kf8 42.Bxh5 Kxe7 43.g6 Kf8 44.Rf4+ Kg8 45.Rf6 a3 46.Rxd6 a2 47.Rd1

The variation suggests that in the puzzle position, booting Rd4 is a defensive priority, to permit Qc3 to defend the dark squares on the K-side.

My candiate moves were 35. Be6, Nf5, Rh4 and Bf5.

Yup, you are right, 35. Nf5 is not creating strong enough threats if the sac is declined.

One example:
35. Nf5? a3 36. Nh6+ Kg7 37. Rf4 a2 38. Rf7+ Kh8 39. Bf5 Qe5 40. Bxg6. Looks promisng for white as long as Black don't play 40. Qxg5+

The choice of sacrificing B or N was also important in the last Sunday puzzle: Kotronias vs N Davies, 1994. The default favors keeping the N, the more effective short-range piece, but today had some special features to reinforce that choice. Today, there was a dichotomy of K-side attack, on light and dark squares. Only Pg6 defended the light squares, so its elimination was critical to permit the Qd1 to infiltrate. Only Bf5 can enforce the elimination; Nf5 cannot.

Thanks for the compliment.

I try to condense down my answers/posts due to my education (science; condense it down to a math. formula) and because English is not my mother tongue.

But it is great pleasure for me to read yours, dzechiel's and other's posts, explaining more the reason and background of the moves. So please keep your style.

Once again you are right! After reading your post I checked the line again and found a mistake in my analyse.

Here the corrected version: 35. Nf5? gxf5 36. Qh5 Qxd4 37. Qxe8. I thought this is , but after Kg7 38. Qxe7+ Kg6 39. Qxd6+ Kxg5 40. Qxb6 Qxd5+ 41. Kh2 a3

If the line Toga found is similar, -2ps is gracious.

I like the defence <32...gxh5 33.Rh4 Qg7 34.Rxh5 Nbd7>

click for larger view

I understood the logic of the text moves of the puzzle through 37 Qb1+ but thought that 37…Nd3 would stop the attack. In that continuation, after 38 Rxd3 Qe5, below, black superficially looks OK.

click for larger view

Here, for example, if 39 Qxb6 then 39…Qe4+ wins the rook.

Now it gets interesting. The next two moves are key. First is 39 Rd4+ Kg7 (forced), followed by 40 Re4!

click for larger view

Black’s queen has two “safe” squares, c3 or g5. If 40… Qc3, then 41 Re6! sets up a lengthy mating combination. If 40…Qxg5+ then 41 Rg4 pins and wins the queen.

After a few dead-ends (Be6+, Qg4), I keep coming across the same problem - that blasted g6 pawn. So let's blow it away with 35. Bf5. Now on 35. ... gf, we have Nxf5 followed by Qh5 and white's attack looks overwhelming.

Okay, let's see if black can decline the bishop with 35. ... Kg7 36. Bxg6 Kxg6. Now I had thought to play 37. Qg4, but of course the game continuation of Qb1+ is much stronger. Hope I would have spotted it OTB.

Looking at it now, I think 37. Qg4 is too slow - black defends with Rf8 and it is difficult for white to make progress.

Toga agrees. The critical defensive move here is 33...Qg7. Yep, if his Ns are off cavorting over on the Q-side, Black really has to make sure someone is at home with the kids.

Can someone put 35. Be6+ Nxe6 through their chess programs to see if there are winning lines for white after the aforementioned move sequence. I found dozens of them, but I think I might be a bit too optimistic, so I need an objective view. Thanks in advance

[ply 15/53 time 02:49 value (to White) -2.03]

35.<Be6> Nxe6 36.dxe6 Rc8 37.Ng4 Qc6+ 38.Kg3 Qc5 39.Nh6+ Kg7 40.Nf7 Qc3+ 41.Rd3 Qc1 42.Rxd6 Qxd1 43.Rxd1 a3 44.Nd8 Rc7 45.Ra1 Nc4 46.Kf4

White needs to make real progress into the K-side or the Black Q-side Ps will win. While 35.Be6+ advances the Pd5 toward Kg8, the move does not root out Pg5, the base of the Black K-side defenses.

<Hm....Should I give up chess altogether? Second puzzle in a row, and I missed it completely>

The obvious reply doesn't work, but it's not that easy to even see that:

35 ... gxf5
36 Qh5 Qxd4
37 Qxe8+ Kh7
38 g6+ Kh6
39 Nxf5+

click for larger view

There is actually a reasonable amount of annotation needed for some of those moves, more than one is critical for white to avoid a draw. So as far as I'm concerned it's not that easy to see that you can play Bf5 even if you only consider the obvious reply. Even back when I played competitively, I would not have accurately seen that over the board.

What is the correct reply for black? I had to resort to Rybka:

35 ... Nc4
36 Rxc4 Qg7
37 Be6+ Nxe6 (otherwise mate)
38 dxe6

35.Nf5 fishy threat of 36.Nh6+ -- but Black shrugs it off, and White's Q didn't get in. Nothing.

35.Bf5 and either I get my Qh5, or he declines Kg7 36.Bxg6 Kxg6 37.Qg4 or Rh4 (I missed Qb1+). White's g-pawn is clearly untouchable, and I felt that Q+R+N+P vs. essentially naked K would be enough.

Black resigned as he will be mated in two-whether he interposes the queen on white's 44{ Rf4 or not.

44 ♖f4+ ♔g6 45 ♕h6# or

<1. Nf3 d6 2. d4 Bg4>

This position, is recorded in the CG.com Opening Explorer as first being played at Master level in Tartakower vs Spielmann, 1938. With his draw in this game, Spielman demonstrates the "Knights before Bishops" convention in opening development doesn't always have to be a hard and fast rule.

<3. c4 Nd7>

Black is aiming for an unbalanced position, where he can put his opponent into unfamiliar territory and gain an advantage. The late English Gm Anthony Miles employed it about as often as any strong Master, starting in the late 1980s. The soundness of the move can be testified to by the fact that some of Miles strongest opponents successfully used it against him, as in Miles vs Anand, 1990. Of course Miles had some success with it himself, as in
I Sokolov vs Miles, 1989. Other Masters who employed this opening on more than one occasion include GM Adams, GM Benjamin, GM Gulko, GM Jansa, GM Murray and GM Speelman.

<4. g3>

This is OK. However, more often played is 4. Nc3 as in R Markus vs I Rogers, 2004 or 4. Qb3 as in Vaganian vs K Shanava, 2008.

<4...Bxf3 5. exf3 Ngf6>

This is not bad, but a good alternative is the flexible 5... g6 which won in Ftacnik vs Chandler, 2006.

<6. f4>

Don't know if it was a novelty, but this is the first game in the Opening Explorer with this move. It's purpose is to pressure the Black position by restraining ...e5.

<6...c6 7. Nc3 g6 8. Bg2 Bg7 9. d5 O-O 10. O-O Nb6 11. Qe2 Rc8 12. Rd1 Re8>

Perhaps Black can neutralize White's space advantage with 12... cxd5 13. cxd5 Nc4 14. b3 Nd7 15. bxc4 Bxc3 16. Rb1 Nc5 , when play might continue 17. f5 Qd7 18. fxg6 fxg6 19. Be3 b6 20. Bf3 Qc7 21. Bg4 Rce8 22. f4 Bg7 23. Bxc5 Qxc5+ 24. Kg2 Rf6 25. Bd7 Ref8 26. Be6+ Kh8 27. Re1 Rb8 28. Rb5 Qc7 29. Reb1 Rff8 30. a4 a6 31. R5b3 Qc5 32. h3 Rb7 33. Qd3 Qa5 34. f5 gxf5 35. Bxf5 h6 36. Be6 Rbb8 37. Ra3 Qc5 38. Ra2 Rbe8 39. Re2 Rf6 40. Bg4 Ref8 41. Rxe7 Rf2+ 42. Kh1 Qd4 43. Be2 Qxd3 44. Bxd3 Rd2 45. Bg6 Bd4 46. Re8 Kg7 47. Rxf8 Kxf8 48. Rf1+ Kg7 49. Be8 Re2 50. Rf7+ Kg8 51. Rf4 Be5 52. Rf3 Rc2 53. Rb3 Bd4 54. Bc6 Kf7 55. Rb1 Rxc4 56. Kg2 =.

<13. Be3 cxd5 14. cxd5 Nfd7>

Perhaps Black can improve his drawing chances with 14...Qd7 , when play might continue 15. h3 h6 16. a4 Qc7 17. a5 Nc4 18. Nb5 Qd8 19. Bd4 Nxa5 20. Nxa7 Nb3 21. Nxc8 Nxa1 22. Rxa1 Qxc8 23. Ra7 Nxd5 24. Bxg7 Kxg7 25. Bxd5 Qc5 26. Bxf7 Kxf7 27. Rxb7 Qd5 28. Rb5 Qe6 29. Qxe6+ Kxe6 30. Kf1 Kd7 31. Ke2 Kc6 32. Rb3 e6 33. Kd3 .

<15. Bd4 a6 16. h4 Bxd4 17. Rxd4 Nf6 18. Nd1 Rc7 19. Ne3 Qc8 20. Kh2 Rc1 21. Bh3 Qc5 22. Rxc1 Qxd4 23. Rd1 Qc5 24. Qf3 Qb5 25. b3 Qa5 26. g4 Qxa2 27. g5 Nfd7 28. h5!?>

White is totally committed to his patient King-side attack. If it fails, Black could easily turn the tables and win. Interestingly, Fritz 8 at about 15 ply shows Black to be holding a near decisive advantage.

<29...Nc5 30. Rd4 Qc3>

Black is probably correct to take action in the center to counter White's attack. However, a little pressure on the Queen-side with 30... a5 31. f5 a4 might give black enough to hold the position after 32. hxg6 hxg6 33. fxg6 fxg6 34. Rh4 Nxd5 35. Nxd5 Qxf3+ 36. Kxf3 a3 37. Bf1 Rf8+ 38. Ke3 a2 39. Bc4 e6 40. Ne7+ Kg7 41. Bxa2 Re8 42. Nxg6 Kxg6 43. f4 b5 44. Bb1+ Kg7 45. Rh7+ Kg8 46. Rc7 b4 47. Rc6 Rd8 48. Kd4 b3 49. Kc3 Kg7 50. Rc7+ Kh8 51. Rc6 Kg8 52. f5 exf5 53. Bxf5 Rf8 54. Bg6 Rf3+ 55. Kb2 Rf2+ 56. Kc3 Rg2 57. Rxd6 Rxg5 58. Rc6 =.

<1. Qd1 a5 32. f5!? (maybe ?!)>

White makes no secret of his aggressive intentions with this attacking move. However, the alternative 32. hxg6 hxg6 33. f5 Kg7 34. Rh4 Qe5 35. Qf3 might allow Black to hold on for at least a draw after 35...Rg8 36. fxg6 fxg6 37. Be6 Nxe6 38. dxe6 Qxe6 39. Qxb7 d5 40. Qc7 Qd6 41. Qc3+ e5 42. Ng4 Nd7 43. Qh3 Kf8 44. Nf6 Nxf6 45. gxf6 Qxf6 46. Qc8+ Kf7 47. Qd7+ Kf8 48. Qc8+ Kf7 49. Qd7+ Kf8 50. Qc8+ Kf7 . 51. Rh7+ Rg7 52. Rh3 e4 53. Rc3 Qg5+ 54. Kf1 Qh5 55. Rc7+ Kf6 56. Qf8+ Kg5 57. Rxg7 Qd1+ 58. Kg2 Qg4+ 59. Kh2 Qh5+ 60. Kg1 Qd1+ 61. Kg2 Qg4+ 62. Kh1 Qd1+ 63. Kg2 =.

Yet, given the strength of the move black missed on his next turn, White might have been wise to play 32. hxg6!? after all.

<32... a4?>

Black doesn't fully realize the strength of the White attack, and misses an opportunity to turn the tables with winning chances after

32... gxh5! 33. Rh4 a4 34. Qxh5 Qg7 to (Friz 8 rates it -1.94 @ 15 ply);

or 32... Rf8! 33. hxg6 fxg6 34.
Rh4 Qd3 35. Qxd3 Nxd3 to .

These two alternatives were recommended by <Random visitor> earlier, and I'm sure he's alalyzed them in depth on his deep Rybka program. So, I'm pretty confident either 32...gxh5! or 32...Rf8! would have turned the tables on White in this game.

<33. hxg6 hxg6 34. fxg6 fxg6 35.Bf5!!>

This combined clearance and demolition sham sacrifice, which appears to initiate a decisive King-side attack, is apparently the move Black overlooked earlier.

<35...Kg7 36. Bxg6! Kxg6>

No help for Black is 36... Qxd4 37. Qxd4+ Kxg6 38. Qh4 .

<37. Qb1+! Kg7 38. Nf5+! Kg8 39. Nh6+>

Better is the immediate 39. Nxe7+!, with a mating attack. However, White gets a second chance after...

<39...Kg7 40. Nf5+ Kg8 41. Nxe7+!>

Now White sees the mate-in-five.

<41...Kf8 42. Qh7 Rxe7 43. Qh8+ Kf7>

Black resigns in lieu of 44. Rf4+ Kg6 (44...Qf6 45. Qxf6#) 45. Qh6#.