33 Bxe5 seems to fail as it allows black to finish the exchange with check.. 33 ... Bxe5+ 34 Rxe5 Qxe5+ 35 Qf4 Qxf4+ 36 Kxf4 Bxg6 which is winning for black

33 Rxe5 Bxe5 ( Qxf6 34 Qxd4 and black is losing)

Now white has two options f7 and Qxd4 (I think f7 is much more forcing, so I will ignore Qxd4). The point being that Bxe5+ loses a tempo for white as black plays Qxe5+ and the reverse is true for black as Bxf4+ Qxf4+, so the bishops are stuck as both kings are on the b8-h2 diagonal!

33 f7

But how does black defend this pawn onslaught?

33 ... Rg7/Rxg6 34 f8=Q+ mating
33 ... Rf8/Rh8 34 g7 with double the trouble
33 ... Rc8 34 Rxc8+ Qxc8 (Kxc8 35 f8=Q+ winning) 35 Bxe5+ Ka8 36 g7 winning

33 ... Rd8
34 Qxd4 Bxf4+ ( Rxd4 35 f8=Q+ mating )
35 Kxf4 Bd5 (Qd6+ is fatal, Qd5 36 Qxd5 Bxd5 37 Ke5 tranposes into the same line)

36 Qe5+ Qxe5
37 Kxe5

at which point black can resign,

37 ... Ba2/Bb3 38 Kf6 wins
37 ... Bxf7 38 gxf7 wins
37 ... Bf3 38 g7 wins

I guessed <33.Rxe5>, but I failed to see <35.Qxd4>! I didn't see the pin on the bishop until I played it out on the board.

Black has a threat on g4 which needs to be parried before white pushes the pawns forward.

The obvious move is therefore 33. Rxe5, removing the knight.

After this it is surely a question of pushing the pawns, starting with 34. f7.

I haven't got time to work through all the variations today, but this must be the right idea.

V Okhotnik vs V Berezhnoi, 1981 (33.?)

White to play and win.

Material: R+P for B+N. The Black Kb8 has 1 legal move, a8 on the back rank. The White Bf4 x-rays Kb8 through the Black Bd6 and Ne5, limiting their moves. The advanced White passers Pf6 and Pg6 have queening possibilities, so White can threaten on both sides of the board. The Black Be4 is loose; and Ne5 is attacked twice, burdening its defenders, Bd6 and Qe6. Black has a credible counter-attack, e.g., Rg8 x-rays Kg3 through Pg6 and Pg4, and he has the powerful threat 33…Qxg4+ against Kg3. The candidate therefore must have defensive qualities. Throughout the following, Black counter-attack against Kg3 is just short of credibility.

Candidates (33.): Rxe5

33.Rxe5 (threatening 34.Rxe6 or 34.Rxe4)

33…Bxe5 [else, down at least a piece with no real counter-attack]

34.f7 (threatening 35.fxg8=Q+ Qxg8 36.Bxe5+, up a R; or 35.f8=Q+)

(1) 34…Rc8

[Bxf4+ 35.Qxf4+ K-any 36.fxg8=Q+ Qxg8 37.Qxe4 wins Be4 and Rg8]

34.Bxe5+ Qxe5 [Ka8 35.Rxc8+ Qxc8 36.g7 ends the game]

35.Qf4 (threatening 36.Qxe5+, 36.Rxc8+, or 36.f8=Q)

35…Qxf4+ 36.Kxf4 (threatening 37.Rxc8+ or 37.f8=Q)

36…Rf8 [or R moves along 8-th rank]

37.g7 (threatening 38.g8=Q or 38.gxf8=Q+)

White gains a crushing material advantage after the coronation.

(2) 34…Rd8 [Rf8 or Rh8 35.g7, making the threat in 34. indefensible]

Candidates (35.): g7, Re1

[35.g7 Qxf7 36.Bxe5+ Ka8 37.Qf4 Qd5 is unclear]

35.<Re1> (threatening 36.Rxe4)

<[At 15 plies, Toga evaluates the game move 35.Qxd4 at +6.71 P; my move 35.Re1 at a comfortable +3.43 P. I was aware of 35.Qxd4 as a possible candidate, but I was nervous about my board vision handling so many complications across the whole board so many plies ahead.]>

(2.1) 35…Bc6 [Be4-other is worse] [again, Bxf4+ 36.Qxf4+]

36.Rxe5 (threatening 37.Re5-any+)

36…Qa2 [else, drop the Q, e.g., 36…Qb3+ 37.Re3+] 37.Re8+

37…Kc8 [else, drop Rd8] 38.Rxd8+ Kxd8 39.f8=Q+

With the initiative and an extra Q, White is just mopping up the mess.

(2.3) 35…Qxg6 [Qd5 36.g7 wins]

36.Bxe5+ Ka8 [or Kc8] 37.Qf4 (threatening 38.Rxe4 or 38.f8=Q)

The candidate 33.Rxe5 was not too hard to find, period, because the tactical assessment indicated it. The threat 33...Qxg4+ made defense of Pg4 paramount. Thus, because aggressive moves are always preferable to defense, capture of Ne5 became attractive, as a forcing move. Because of the Ks, the tactics were bound to center around who faltered on the b8-h2 diagonal first. The candidate 33.Bxe5 meekly surrenders the critical terrain. In any case, a rapid (almost unconscious) calculation shows it fails. The candidate 33.Rxe5 therefore was my first and only initial candidate.

Yesterday, I expressed the view that because White had a completely sound position once the game turned in his favor, kibitzers were overestimating the importance of exact calculation. Although reassuring that one could calculate the win of an extra R, it was unnecessary to confirm Re1-e4 as a winning candidate.

Today, with White on a razor edge, calculation to an assuredly winning quiescence is much more important. According to Toga, e.g., my variation [35.g7 Qxf7 36.Bxe5+ Ka8 37.Qf4 Qd5 is unclear] is in fact winning at about +2.0 P (37.Qxd4 is better at about +4.0 P], but the exposure of the White Kg3 requires extensive exact calculation to confirm the win. In my view, there is nothing wrong in using an inferior but simpler variation to demonstrate the soundness of a winning candidate.

In the game of chess, you only need to make one move at a time.

<15.Bxc4?!> what an awesome move!

There are certain dangers when one castles queenside with a missing c pawn. Among many other things it invites a knight, under the right circumstances, to check on d6 (or d3 if playing black) and then fork the rooks on f7 (or f2).

33.Rxe5 Bxe5 34.f7 Rf8 35.g7 Qxf7 36.gxf8 Qxf8 37.Bxe5+

This leaves white a rook up for a pawn, with initiative. let's see what really happened.

John, I just don't see how white saw through the whole thing. After 36...Qe7, I thought he had lost.

click for larger view

It took nerves of steel to then play (correctly) 37 Qxe4, (instead of the safer Qe5+) knowing 37...Qg5+ was coming and the potential rook loss.

But 37 Qe5+ Qxe5+ Kxe5, below, allows for the bishop to trade itself for the two passed pawns, making an even game.

click for larger view

"The whole combination" includes <35.Qxd4>. You can <start> the combination with 33.Rxe5, justifying its soundness with an inferior variation <without> calculating the consequences of 35.Qxd4. When Move 35 arrives, the justification of 35.Qxd4 is straightforward.

This reiterates my previous kibitz, of course. I will confess to grinding a personal axe here. I am often torn Mon-Wed by my desire <not> to calculate, and the knowledge that if I do not post the variations right to mate, a lot of kibitzers will regard my "solution" as incomplete. I regard it as a flaw in practical analysis to do more calculation than necessary to justify a move.

I try not to do more than I need to, to accomplish a goal.

Black's response (33 ... Bxe5) was the most obvious one and probably the best, but I remember considering 33 ... Qxf6 (sacrificing the bishop and putting white's rook strongly on e4 to capture the inevitable ... Bxf4). In retrospect, I'm wondering if ... Qxf6 would have been better anyway.

Vladimir Okhotnik - Viktor Berezhnoi,

Analysis by Rybka 3 1-cpu: ply 15 time 5 min

White to play (33.?) "Very Difficult"

Material is roughly equal, with White having a rook and pawn for bishop and knight, but White's trump card appears to be the two connected passed pawns on f6 and g6. However, Black is threatening Qxg4+, and a move like 33. f7? is answered by 33...Qxg4+ 34. Kf2 Rxg6! and White is busted as the f-pawn is not going anywhere.

So we must get rid of the knight, which is also attacking one of our key pawns on g6, but 33. Bxe5 is strongly met by 33...Bxe5+ 34. Kh3 Rxg6 (or Qxf6) and Black is clearly better with White's king under attack (and pawns being captured). Also, 33. Bxe5 Bxe5+ 34. Rxe5 Qxe5+ is clearly bad. White needs to retain the dark square bishop, at least for now. But there is another option:

33. Rxe5!

In this position, Black's knight was as strong (or stronger) than White's rook, and eliminating it not only removes the threat on g4, but it also strengthens the pawn tandem.

33...Bxe5 34. f7!

Now Black's dark square bishop no longer covers f8, so Black is in trouble.

34...Rd8

If 34...Rf8 (or Rh8) 35. g7 queens a pawn and 34...Bxf4+ 35. Qxf4+ is obviously bad. Also 34...Rc8 35. Rxc8+ Qxc8 36. Bxe5+ Ka8 37. g7 is crushing.

35. g7 Qxf7

Forced, to stop White from queening.

36. Bxe5+ Ka8 37. Qxd4!!

Double-attacking the rook and bishop, and showing the weakness of Black's back rank as 37...Rxd4?? loses quickly to 38. Rc8#. Black's next move is pretty forced.

37...Qf3+ 38. Kh4 (forced) and now:

A) 38...Rg8 39. Qc4! Qf2+ (or 39...Bd5 40. Qc8+ Rxc8 41. Rxc8#) 40. Kh5! and Black has no answer to the threats of Qxg8+ and Qc8+, i.e. 40...Rb8 41. g8=Q.

B) 38...Re8 39. Qd7 Qf2+ 40. Bg3 Qf6+ 41. Kh3 and Black can't meet the dual threats of Qxe8+ and Rc8+, i.e. 41...Qe6 42. Qxe6 Rxe6 43. Rc8# or 41...Bc6 42. Qxe8+! Bxe8 43. Rc8#, or 41...Rb8 42. g8=Q Rxg8 43. Rc8+.

C) 38...Qd3 39. Qxd3 Bxd3 40. Rd1! and White will win the bishop next move with 41. Rxd3, due to the threat of the g-pawn queening.

Time to see how close I came; I often miss the game continuation!

~~~~~~

<33. Rxe5!,Bxe5 34. f7!,Rd8 35. g7 Qxf7 36. Bxe5+ Ka8 37. Qxd4!!,Qf3+ 38. Kh4 (forced) and now:

A) 38...Rg8 39. Qc4! Qf2+ (or 39...Bd5 40. Qc8+ Rxc8 41. Rxc8#) 40. Kh5! and Black has no answer to the threats of Qxg8+ and Qc8+, i.e. 40...Rb8 41. g8=Q.

B) 38...Re8 39. Qd7 Qf2+ 40. Bg3 Qf6+ 41. Kh3 and Black can't meet the dual threats of Qxe8+ and Rc8+, i.e. 41...Qe6 42. Qxe6 Rxe6 43. Rc8# or 41...Bc6 42. Qxe8+! Bxe8 43. Rc8#, or 41...Rb8 42. g8=Q Rxg8 43. Rc8+.

C) 38...Qd3 39. Qxd3 Bxd3 40. Rd1! and White will win the bishop next move with 41. Rxd3, due to the threat of the g-pawn queening.>

I do have one small quibble, though. In line B, TheBish claims there is no way to meet the two threats of Qxe8 and Rc8 with mate. Black, however, does manage to meet both threats with 41...Qd8! If we look a bit further, though, it becomes clear that after the prosaic 42.Qxd8+!,Rxd8 43.Rf1 there is no answer to 44.Rf8, queening the pawn.

No time to go 'Silman' on this bad boy so I gave myself a minute. I had either the rook or the bishop taking the knight on e5 and then pushing the pawn to f7. Instinctively, I want to capture with the rook. Let's check.

---