That also looks like a close race after 50...a3 51. Kxb7 a2 52. Nxc5 a1=Q 53. Kxc7, but maybe Black can still win it.

Black two pawns are toast on the c file.

White has a potential pass pawn on d file, that is protected by the knight and the king. Which would be hard for a black Queen to take out.

I did keep looking for the win beyond what I posted above, and I'm saying I had trouble finding something convincing. It looks like Black would have to trade his queen for White's d-pawn, leaving two pawns versus knight, and White's pieces are often close enough to stop Black's pawns. Do you have any specific improvements?

We do both agree that 50...Rb1 deserves a "?" and Young Capablanca stole a 1/2 point in this game. I don't know what the time controls were for this game... Maybe Corzo was low on time?

I am not a Grandmaster so here are the games analyzed by engines. The first line shows the engines played the same moves on the 2nd line shows a new branch.

Games below was set to 20 mins for all starting from black to move at 50.....

< Hiarcs played the following. >

50... a3 51. ♔xb7 a2 52. ♔xc7 a1=♕ 53. ♘xc5 ♔g8 54. ♔c6

54... Qa8+ 55. Kb5 Qe8+ 56. Kb6 Qe2 57. d6 Qxc4 58. Nb7 Qc8 59. Nc5 Kf7 60. d7 Qd8+ 61. Kc6 Ke7 62. Kd5 g4 63. Ke4 Qa8+ 64. Kd4 Qa1+ 65. Kc4 Qa2+ 66. Kc3 Qxh2 67. d8=B+ Kxd8 68. Ne6+ Ke7 69. Nc5 Qc7 70. Kd4 Qxc5+ 71. Kd3 Kd6 72. Kd2 g3 73. Ke2 g2 74. Ke1 g1=Q+ 75. Kd2 Qcd4+ 76. Ke2 Qdd1#

< Fruit 2.1 played >

50... a3 51. ♔xb7 a2 52. ♔xc7 a1=♕ 53. ♘xc5 ♔g8 54. ♔c6

54...Qd4 55. Kb5 g4 56. Nd7 Qb2+ 57. Kc6 Qxh2 58. Nf6+ Kg7 59. Nxg4 Qf4 60. Nxh6 Qxc4+ 61. Kd6 Kxh6 62. Ke5 Kg6 63. d6 Qc5+ 64. Kf4 Qxd6+ 65. Ke4 Kf6 66. Ke3 Kf5 67. Kf2 Qd2+ 68. Kf3 Qe1 69. Kg2 Kg4 70. Kh2 Kf3 71. Kh3 Qh1# 0-1

< Crafty 19.18 played >

50...a3 51.♔xb7 a2 52.♔xc7 a1=♕ 53.♘xc5

I have been arguing that 50...Ra7 is an easier win than 50...a3. Obviously, computers don't need to consider the concept of "easier win" so they just play greedy chess. But here are some specific comments.

<Problem with 50....Ra7 is the following... 51 Nc1 a3 52. Na2... blocking the pass pawn. Black still can win thou.>

I didn't give the 51. Nc1 line because it is so easy for Black: 50...Ra7 51. Nc1 a3 52. Na2 g4 53. Kb8, after which there are many ways to win, such as 53...Ra4 54. Kxc7 Rxc4 55. d6 Rd4 54. d7 h5 etc. and Black's obviously many moves ahead in this race.

Moreover, after the engine's 50...a3, White also can play 51. Nc1 which just transposes to lines similar to above (one example of a Black win is 50...a3 51. Nc1 Rb1 52. Na2 g4 53. Kxc7 h5 54. d6 h4 55. d7 Rd1, etc.) so the Nc1 possibility is not really a distinctive feature for comparing the two lines.

So far, I've had time to work on the line given by <Hiracs> and <Fruit 2.1>: 50... a3 51. Kxb7 a2 52. Kxc7 a1=Q 53. Nxc5 Kg8

Here, I varied with <54. Nb7> instead of the passive 54. Kc6 given by the engines. The game continued 54...Qe5+ 55. d6 Qxh2 56. c5 Kf7 57. c6 g4 58. Kc8 g3 59. d7 g2 60. d8=Q g1=Q 61. Nd6+ Ke6 62. Qd7+ Kd5 63. Nc4+ Ke4 64. c7 Qhf2 65. Qe6+ Kd4 66. Kb8 Qb1+ 67. Qb6+ Qxb6 68. Nxb6, with a drawn position.

(For the record, gnuchess kept trying with 68...Qg3 69. Kb7 Qg7 70. Kb8 Qe5 71. Kb7 Qe4+ 72. Kb8 Qf4 73. Kb7 Qf3+ 74. Kb8 Qc6 75. c8=Q Qxb6+ 76. Ka8, and now the position is a Nalimov tablebase draw.)

http://www.vistula.risp.pl/historia...

So I searched a bit more for a translator and the moves that we were talking about is below. Taken from the page above.

http://www.poltran.com/pl.php4

< brown text is where the moves were suggested >

black text is where the actual moves played.

B) 9...0-0?! 10.c3! Black had the opportunity of securing a clear advantage : 9...Bxd2+ 10.Qxd2 (or 10.Kxd2 0-0 clearly in Black's favor) d6! 11.Nc4! 0-0 clear advantage Black ; 9...Qh4 also seems to lead to Black's clear advantage.

C) 12.f3! Clearly not 12.Nxe4 h5! and... Bxe, winning!

D)14.Ke2. Better was 14.Kd1! with idea Kc2 would have avoided the problem that will arise later in the game.

E)18...e5. Missing 18...Rxd5! 19.c4 Rh5! 20.Nh6! e5 21.Nf5 (only move) Bc8! 22.Nxe7! Bxh3 23.Nd5 Na6 Black's small advantage.

F) 31.Rxf3?! Capablanca misses 31.Kxf3! bxc5 32.Kg3 Rh5 (32...g5 33.Rf4! Rxf4 (33...Rh5 34.Rf7! Rh4 (34...Kg8 35.Rxc7 Rxc7 36.Nxc7 Kf7 37.e4 White clear advantage) 35.Rxc7! Rxc7 36.Nxc7 Bd7 37.Re1 White clear advantage) 34.exf4 Bd7! 35.fxg5 Bxb5 36.cxb5 Rg8 37.h4! hxg5 38.Re1 White advantage) Rh5 33.Rf7 (or 33.e4 White small advantage) Re8! 34.Re1 Ree5 45.e4 White small advantage.

G) 33.Kxf3?! Better was 33.Bxd6! cxd6 (only move) 34.Kxf3 Rf8+ (34...Rh3+ 35.Ke4 Re8+ 36.Kd4 Rexe3 (36...Rd8 37.Rh1=) 37.Nxd6 Rd3+ 38.Ke5 Rhe3+! 39.Kf4 Black small advantage) 35.Kg3 Re4 36.Re1 Rf6 Black small advantage.

H) 38...Ree8? 38...Ref7! 39.Re2 (39.Re3 Rf1! 40.h4 gxh4+ 41.Kxh4 Rh1+ 42.Kg3 (42.Rh3 Re1! 43.Kg3 Kg7 Black winning) Rg8+! 43.Kf2 Rh2+! 44.Kf3 Rh3+ 45.Kf2 Rxe3! 46.Kxe3 h5! Black winning) Rf3+ 40.Kg2 (40.Kg4? R8f4+ 41.Kh5 Kg7 and Rh4#) h5! 41.h4! gxh4! 42.Nxa5 (or 42.e5 h3+ 43.Kh2 Rf2+ 44.Rxf2 Rxf2+ 45.Kxh3 dxe5 46.Nxe5 Rxa2 Black winning) h3+ 43.Kh2 Rf2+! 44.Rxf2 Rxf2+ 45.Kxh3 Rxa2 46.Nc6 Rb2 Black winning

I) 40.Rxe5?. Capablanca had to play 40.Nxa5 Ra8! 41.Nb7! Rxa2 42.Nxc5! Rff2 43.h3 Black advantage.

J) 43.a4?! Three better alternatves which are better : 43.Nxc5 Rxa2 44.h4 Kg7 45.Kg4 gxh4 (or 45...Kg6 46.hxg5 hxg5 47.Na4 Black winning) 46.Kxh4 Black winning ; 43.h4 gxh4! 44.Kxh4 a4 45.bxa4 Re4+ 46.Kh5 Rxc4! 47.Kxh6 Rh4+! 48.Kg5 Rd4! 49.Kf6 c4! Black winning ; 43.h3 Rxa2 44.Nxc5 Kg7 Black winning.

K) 50...Rb1?? The position which attracted the most comments. Personnally and with Fritz 8 as a helper, it appears that there are 2 ways that would have easily won the game for Corzo : the one that is the most popular amongst the gentlemen who poster variations above 50...a3 51.Kxb7 a2 (only move) 52.Kxc7 (52.Nxc5 a1Q 53.Ne6(53.Kxc7 transposes) Qa2 54.Nxc7 Qxc4 winning) a1Q 53.Nxc5 Kg8! wins for Black ; another way, which hasn't been discussed in here, seems to be 50...Rb2! 51.Kxc7 (51.Nxb2 a3 52.Na4 a2 53.Kxc7 a1Q 54.Nxc5 which transposes) Rxh2 52.d6 Rd2 53.Nxc5 g4 winning.

L) 51...Rd1. Stronger was : 51...a3! 52.d6 a2 53.d7 a1Q 54.d8Q+ Kh7 55.Kc6! Rd1! and Corzo could still have played for the win, even if it is not that easily obtained. Here, would the white queen give check on the 7th rank, the black queen would interpose on g7.

" The first two games were won quickly by [Corzo], but something in the third, which was a draw, showed me that he had his weaknesses and gave me the necessary courage and confidence. From there on he did not win a game... The victory made me, morally at least, the champion of Cuba. I was then twelve years old. "

Corzo responds with 1...f5, the Dutch, again trying to take advantage of the 13 year-old's lack of book knowledge and experience. Capablanca opens with a Colle-Zukertort type system against Corzo's Queen's Indian type setup.

Capablanca gets into trouble in the opening by 8. Bxe4 [-0.50] when better was 8. O-O [0.14].

Capablanca then incurs a weak f3 pawn to obtain an open g-file by 12. f3 exf3 13. gxf3.

By 18. d5 Capablanca, in an inspired but unsound conception, sacrifices his d5 pawn with a view to also sacrificing his e-pawn, so that his Bd2 can bear down on Corzo's g7. But Corzo defends well and still hangs on to a slight advantage

By 28.Nb5 Capablanca posts his N on a strong post attacking the c7 base of Corzo's Q-side pawns, which will tie one of Corzo's Rs down defending c7. Capablanca's N can be only dislodged by Corzo's KB.

So in what is either a blunder (unlikely) or an intuitive positional sacrifice, Capablanca allows Corzo's B to pin one of Capablanca's Rs and win the exchange.

Then by 30. Bxc5 Capablanca exchanged Black's strong N leaving him with the R & N v 2 Rs endgame that he had presumably aimed for. But according to Houdini that position after 33...bxc5 was a win for Black [-1.91].

51. ... Rd1 Corzo belatedly realised that, although his a-pawn will queen one move before Capablanca's d-pawn, the latter will queen with check! (51... a3 52. d6 a2 53. d7 a1=Q 54. d8=Q+). Capablanca had played the endgame well, if not flawlessly, but Corzo's R, passed a-pawn and 2 to 1 K-side pawn majority, should have won against Capablanca's N, passed d5 pawn and c4 pawn, the latter blocked by Corzo's c5 pawn. But Corzo's weak endgame play saw him, instead of advancing his K-side majority, against which Capablanca would have no answer, waste two tempi advancing his passed a-pawn when Capablanca's N would stop it.

53. ... a2 After this it is a draw because now these moves by Black's a-pawn are wasted tempi because White's N can stop it queening. The winning moves for Corzo were 53... h5 or 53... g4

54. Nb3 g4 Too late Corzo discovers what would have been the winning method, advancing his K-side majority and queening his g-or h-pawn. Now Capablanca only just has the time to queen his d5 pawn, winning Corzo's R, then win Corzo's c5 pawn, and be about to queen his c-pawn, one move after Corzo would have queened his g-pawn, but with nothing decisive that Corzo could do.

Capablanca's in the comment above, from his "My Chess Career," shows that the 13 year-old was surprised that Corzo couldn't see the endgame win, when he clearly could. Hooper & Brandreth in "The Unknown Capablanca" commented: