Black threatens 56... fxe3.

The pawn on a7 hinders 56.b6. This suggests 56.Bxa7:

A) 56... Rxa7 57.b6

A.1) 57... Ra4+ 58.Kc5 (58.Kb5 Kd6 59.b7 Kc7 60.Kxa4 f3 61.Kb5 f2 62.Ka6 f1=Q+ 63.Ka7 Qa1#) Ra5+ 59.Kb4

A.1.a) 59... Kd6 60.b7 Kc7 61.Kxa5 f3 62.Ka6 f2 (62... Kb8 63.Kb6 f2 64.c7#) 63.Ka7 f1=Q 64.b8=Q+ Kxc6 65.Qc8+, draw.

A.1.b) 59... f3 60.Kxa5 f2 61.b7 f1=Q 62.b8=Q+ and White seems to have considerable winning chances.

A.2) 57... Kd6 58.bxa7 f3 59.a8=Q f2 60.Qa1 + -.

A.3) 57... f3 58.b7 (58.bxa7 f2 59.a8=Q f1=Q =) f2 (58... Ra8 59.bxa8=Q f2 60.Qa1 + -) 59.b8=Q+ followed by 60.Qb1.

B) 56... f3 57.Bc5 Rh2 58.b6 Rc2+ 59.Kb5 Rxc5+ 60.Kxc5 f2 61.b7 f1=Q 62.b8=Q+ looks similar to A.1.b.

C) 56... Kd6 57.b6 (57.Bb8+ Ke6 58.Bxf4 Rh4 59.b6 Rxf4+ 60.Kb5 Kd6 - +) f3 58.Bg3 f2 59.Bxf2 Kd6 60.c7

C.1) 60... Kd7 61.Kb5 f5 (61... Kc8 62.Kc6 + -) 62.Ka6 Kc8 63.Bg3 + -.

1. (-0.82): 57...Ra4+ 58.Kc5 f3 59.b7 Ra5+ 60.Kb4 f2 61.b8Q+ Ke4 62.Kxa5 f1Q 63.Qb4+ Ke3 64.c7 Qf5+ 65.Ka6 Qc2 66.Qe7+ Kf3 67.Qxf6+ Kg3 68.Qg7+ Kf4 69.Qf8+

Rybka 4 x 64

depth: 18 : 5 min :
Black error
(=0.00):31...Ne4. Best, Bd6, -0.23.

depth: 19 : 5 min :
White blunder
(-9.12):32.Ka1. Best, Rf3, =0.00.

depth: 17 : 3 min :
Black blunder
(-3.51):32...Nd2. Best, Qa4+, -9.12.

depth: 19 : 5 min :
Black error
(-2.81):36...Bc5. Best, Bd6, - 3.51.

depth: 22 : 4 min :
Black error
(-2.31):38...Rh1. Best, hxg5, -2.97.

depth: 27 : 15 min :
White blunder
(-7.04):52.c5. Best, Bf4, -4.07.

depth: 26 : 4 min :
Black blunder
(-4.01):53...Ke5. Best, Rh7, -7.04.

depth: 25 : 3 min :
Black blunder
(-0.75):55...f4. Best, Rh4+, -3.80.

depth: 24 : 2 min :
Black blunder
(+5.46):57...f3, Best, Ra4+, -0.75.

But the deep penetration into the position shows exactly opposite- black has to win. Read <Phony Benoni> or <patzer2>. After long hesitations I prefer don't support both of them. My line exactly repeat what wrote <M.Hassan>. I want only give the diagram after 64...Kxc6 This looks like a draw.

click for larger view

Nalimov Endgame Tablebases
Black to move

click for larger view

No, I can't. But I know it's a draw with White to play.

You may have seen the "Nalimov" tablebase mentioned. This is a database of endgame positions with six pieces or less that a computer has solved. You can refer to it at: http://www.k4it.de/index.php?topic=....

It will list all the possible moves in a position and give you the outcome of each with perfect play. For example, in this position White draws with 57.Qa8+, 57.Qc8+, 57.Qe8+, 57.Qb7+, and 57.Qb6+. All other moves lose, including 57.Qb2 which holds out for 73 moves.

+ endgames are quite difficult to explain, at least for non-masters like me, because there are so few guideposts and general rules. For instance, "Two passed pawns on the sixth rank defeat a rook" is such a general rule. As we have seen today, the individual position must be examined to determine if it's an exception, but the guidepost makes finding a candidate move simpler.

Here's one line of the draw after:

<57.Qa8+ Kd6 58.Qb8+ Ke6 59.Qc8+ Ke5 60.Qc7+ Ke4 61.Qb7+ Ke3 62.Qe7+ Kd3 63.Qd7+ Ke2 64.Qe7+ Kd2 65.Qd6+ Ke1 66.Qe6+ Kd1 67.Qd6+ Kc1 68.Qc6+ Kb1 69.Qb6+ Ka1 70.Qa6+ Qxa6+ 71.Kxa6 Kb2 72.Kb5 Kc3 73.Kc5 Kd3 74.Kd5 Ke3 75.Ke6 1/2-1/2>

You can see what happens: White checks the Black king from afar along the sixth rank, keeping the pawn under attack. Black can't interpose his queen or move the king to the f-file without losing the pawn. He tries moving the king to a1, but then Qa6+ forces a queen trade in a position where White can grab the pawn before Black can protect or promote it.

That's just one way, of course. You can use Nalimov to study the ending more carefully, and maybe come up with your own guideposts.

Just for fun, here's the 73-move win.:

My two cents:

In simplifed form. See below if you want refutations of alternative moves for either side. 58...Ra5+ 59. Kb4 Kd6 60. b7 Kc7 61. Kxa5 f3 62. Ka6 f2 63. Ka7 f1=Q 64. b8=Q+ is a fairly easy draw. 58...Ra1! 59. b7 Rc1+! 60. Kb6 Rb1+ 61. Kc5 f3 62. c7 f2 63. c8=Q f1=Q 64. b8=Q+ Rxb8 65. Qxb8+ is the best line for both sides. It looks like a draw since white has checks to start and the pawn has not gotten far, and this result is verified by the tablebase. However, this is black's best shot, it could be tricky.

57... f3? 58. b7 is completely lost.
57...Ra6? loses to both 58. b7 Rxc6+ (Rb6 c7) 59. Kb5 or 58. c7 57... Ra4+! is the most logical move, when 58. Kc5 is forced (all three other K moves lose to Kd6) and now: 58... Ra5+ is the simplest draw, and gives black some winning chances. Forced is: 59. Kb4 (Kc4? Kd6 0-1) Kd6 60. b7! (Kxa5? Kxc6 Ka6 f3 b7 f2 b8=Q f1=Q+ and white loses his queen) Kc7 61. Kxa5 f3 62. Ka6 f2 63. Ka7 f1=Q 64. b8=Q+ Kxc6 and white can give a perpetual without great difficulty. 58... Ra1! is also a draw. (these are the only two moves) 59. b7 (c7? Rc1+ and Kd6) Rc1+! (Rb1? c7 is just losing) 60. Kb6 Rb1+ (Black's idea: his rook holds the bpawn white black threatens Kd6) 61. Kc5! (the only move: 61. Kc7? Kd5 stops the pawns - Kd7 f3, all others lose to Kd6) and here black can repeat with Rc1+, but he has f3! 62. c7 f2 63. b8=Q f1=Q 64. c8=Q+ Rxb8 with good winning chances. Here the perpetual is harder (it IS a draw, verified by the tablebase, but this seems to have more chances) since white's king gets in the way of his queen.

White can't stop to capture the : 58...f3 (58...Ra8 59. c7 Rc8 60. b7 Rxc7 61. b8=Q, pinning the , and now black's f- doesn't promote) 59. c7 f2 60. c8=Q f1=Q+ 61. Kxa4.

At this point black has a lot of ways to check, but I think that white can hide his on the eighth rank and queen a second before black does.

That inadvertant "n" at the beginning of the first section is what produced the bad FEN message, as it made nine spaces on the rank instead of eight. Take it out, and you get this:

click for larger view