Alekhine - Capablanca 1927 once more

[Jouni]

[D]8/5pk1/r5pp/P7/3R3P/6P1/5PK1/8 w - - 0 54

From Sune Larsson endgame test. "1927 Alekhine won the 34th game in the battle vs Capablanca to become World
Champion. He did it by playing 1.Ra4! in the above position. Black's rook has
to block the pawn and white has at his disposal an endless reserve of rook
moves along the a-file. Again, the future for black spells "Zugzwang". The
general rule is that a passed pawn should be supported by a rook from the rear.
That's why ideas of 1.Rd5? are wrong. This is a knowledge that have to be
implemented in many chessprograms."

Is this true? I noticed, that in Pavel Hase test Houdini and Rybka played Rd5 and won! So I played 6 top engines from this position after A Ra4 and B Rd5 (game in 20 minutes). In A white won 25/30 games and in B
24/30. Average length of wins were 28 moves in A and 26 in B. So I have feeling both moves are equally winning. BTW Komodo got best score in test even if it were only 1 cpu program..

Jouni

[Ajedrecista]

Hello Jouni:

[D]8/5pk1/r5pp/P7/3R3P/6P1/5PK1/8 w - - 0 54

From Sune Larsson endgame test. "1927 Alekhine won the 34th game in the battle vs Capablanca to become World
Champion. He did it by playing 1.Ra4! in the above position. Black's rook has
to block the pawn and white has at his disposal an endless reserve of rook
moves along the a-file. Again, the future for black spells "Zugzwang". The
general rule is that a passed pawn should be supported by a rook from the rear.
That's why ideas of 1.Rd5? are wrong. This is a knowledge that have to be
implemented in many chessprograms."

Is this true? I noticed, that in Pavel Hase test Houdini and Rybka played Rd5 and won! So I played 6 top engines from this position after A Ra4 and B Rd5 (game in 20 minutes). In A white won 25/30 games and in B
24/30. Average length of wins were 28 moves in A and 26 in B. So I have feeling both moves are equally winning. BTW Komodo got best score in test even if it were only 1 cpu program..

Jouni

I have done fast analysis with MultiPV = 2 with Houdini, Deep Rybka 4.1, Critter 1.2, Komodo 3 and StockFish 2.1.1 (all of them 32-bit). They agree in giving higher evaluations to Rd5 than to Ra4 by different margins. Analysis with 1024 MB of hash under Arena 2.0.1 in an Intel Pentium D930 (2999 MHz and 2 cores):

Code: Select all

FEN: 8/5pk1/r5pp/P7/3R3P/6P1/5PK1/8 w - - 0 1 

Stockfish-211-32-ja:
  1/1	00:00	         115	144	+1,53	Rd4a4
  1/1	00:00	         115	144	+1,77	Rd4d5
---------------------------------------------------------------------------
  2/2	00:00	         224	281	+1,29	Rd4a4 Kg7f6
  2/2	00:00	         224	281	+1,53	Rd4d5 Kg7f6
---------------------------------------------------------------------------
  3/3	00:00	         556	697	+1,45	Rd4a4 g6g5 h4xg5 h6xg5
  3/3	00:00	         556	697	+1,77	Rd4d5 h6h5 f2f3
---------------------------------------------------------------------------
  4/4	00:00	       1.121	1.378	+1,29	Rd4a4 Kg7f6 Kg2f3 Kf6e5 Ra4e4+ Ke5d5
  4/4	00:00	       1.121	1.406	+1,93	Rd4d5 h6h5 Rd5b5 Ra6c6
---------------------------------------------------------------------------
  5/6	00:00	       2.275	2.747	+1,29	Rd4a4 Kg7f6 Kg2f3 Kf6e5 Ra4e4+ Ke5d5
  5/6	00:00	       2.275	2.747	+1,81	Rd4d5 Ra6e6 g3g4 Kg7f6 Kg2f3
---------------------------------------------------------------------------
  6/7	00:00	       5.414	6.414	+1,25	Rd4a4 Kg7f6 Kg2f3 Kf6e5 Kf3e3 f7f5
  6/7	00:00	       5.414	6.414	+1,85	Rd4d5 Kg7f6 Kg2f3 Kf6e6 Kf3e4 f7f5+ Ke4d4
---------------------------------------------------------------------------
  7/10	00:00	       8.322	9.510	+1,41	Rd4a4 Kg7f6 Kg2f3 Kf6e5 Kf3e3 f7f5 f2f4+ Ke5d5 Ra4d4+ Kd5c5
  7/10	00:00	       8.322	9.510	+1,85	Rd4d5 Kg7f6 Kg2f3 Kf6e6 Kf3e4 Ra6a8 Rd5b5
---------------------------------------------------------------------------
  8/16	00:01	      43.659	44.323	+1,25	Rd4a4 Kg7f6 Kg2f3 Kf6e5 Kf3e3 Ke5d5 h4h5 Kd5c5 h5xg6 f7xg6 Ra4a1 Kc5b5
  8/16	00:01	      43.659	45.055	+1,97	Rd4d5 Kg7f6 Kg2f3 Kf6e6 Kf3e4 f7f6 Rd5b5 f6f5+ Ke4d4 Ra6d6+ Kd4e3 Rd6c6 Rb5b6
---------------------------------------------------------------------------
  9/16	00:01	      45.640	45.640	+1,25	Rd4a4 Kg7f6 Kg2f3 Kf6e5 Kf3e3 Ke5d5 h4h5 Kd5c5 h5xg6 f7xg6 Ra4a1 Kc5b5
  9/16	00:01	      45.640	45.640	+1,97	Rd4d5 Kg7f6 Kg2f3 Kf6e6 Kf3e4 f7f6 Rd5b5 f6f5+ Ke4d4 Ra6d6+ Kd4e3 Rd6c6 Rb5b6
---------------------------------------------------------------------------
 10/16	00:01	      53.367	52.526	+1,25	Rd4a4 Kg7f6 Kg2f3 Kf6e5 Kf3e3 Ke5d5 h4h5 Kd5c5 h5xg6 f7xg6 Ra4a1 Kc5b5
 10/16	00:01	      53.367	52.526	+1,97	Rd4d5 Kg7f6 Kg2f3 Kf6e6 Kf3e4 f7f6 Rd5b5 f6f5+ Ke4d4 Ra6d6+ Kd4e3 Rd6c6 Rb5b6
---------------------------------------------------------------------------
 11/19	00:01	     115.093	100.870	+1,25	Rd4a4 Kg7f6 Kg2f3 Kf6e5 g3g4 f7f5 g4xf5 Ke5xf5 Ra4a1 Kf5e5 Kf3e3 Ke5d5 Ra1d1+ Kd5c5
 11/19	00:01	     115.093	100.870	+1,97	Rd4d5 Kg7f6 Kg2f3 Kf6e6 Kf3e4 f7f6 Rd5b5 f6f5+ Ke4d4 Ra6d6+ Kd4e3 Rd6c6 Rb5b6
---------------------------------------------------------------------------
 12/21	00:01	     174.990	136.604	+1,21	Rd4a4 Kg7f6 Kg2f3 Kf6e5 g3g4 f7f5 g4xf5 Ke5xf5 Kf3e3 Kf5e5 Ke3d3 Ra6d6+ Kd3e2 Rd6a6 Ke2e3 Ke5d5 Ra4d4+ Kd5c5
 12/21	00:01	     174.990	136.604	+1,97	Rd4d5 Kg7f6 Kg2f3 Kf6e6 Kf3e4 f7f6 Rd5b5 Ke6d6 Ke4d4 Kd6c6 Rb5c5+ Kc6d7 g3g4 Ra6d6+ Kd4e4
---------------------------------------------------------------------------
 13/23	00:01	     249.848	173.746	+1,17	Rd4a4 Kg7f6 Kg2f3 Kf6e5 g3g4 f7f5 g4xf5 Ke5xf5 Kf3e3 Kf5e5 Ke3d3 Ke5d5 h4h5 g6g5 f2f3 Kd5c5 Ra4c4+ Kc5b5
 13/23	00:01	     249.848	175.701	+1,89	Rd4d5 Kg7f6 Kg2f3 Kf6e6 Kf3e4 Ra6c6 Rd5b5 Rc6c4+ Ke4d3 Rc4a4 h4h5 g6g5 f2f3 Ra4a2
---------------------------------------------------------------------------
 14/23	00:01	     358.056	220.342	+1,17	Rd4a4 Kg7f6 Kg2f3 Kf6e5 g3g4 f7f5 g4xf5 Ke5xf5 Kf3e3 Kf5e5 Ke3d3 Ke5d5 h4h5 g6g5 f2f3 Kd5c5 Ra4c4+ Kc5b5
 14/23	00:01	     358.056	220.342	+1,81	Rd4d5 Kg7f6 Kg2f3 Kf6e6 Kf3e4 Ra6c6 h4h5 f7f5+ Ke4d4 g6xh5 Rd5e5+ Ke6f6 Re5c5 Rc6a6 Rc5b5 Ra6d6+ Rb5d5 Rd6c6 f2f3
---------------------------------------------------------------------------
 15/30	00:02	   1.283.293	441.601	+1,13	Rd4a4 Kg7f6 Kg2f3 Kf6e5 Kf3g4 Ke5d5 h4h5 Kd5e5 h5xg6 f7xg6 Ra4a1 Ke5d5 Ra1d1+ Kd5c5 Rd1h1 h6h5+ Kg4g5 Kc5c4 Rh1d1 Kc4b5 f2f4
 15/30	00:02	   1.283.293	441.601	+1,93	Rd4d5 Kg7f6 Kg2f3 Kf6e6 Rd5b5 Ke6d7 Kf3e4 Kd7c6 Rb5d5 Ra6a8 Ke4d4 Kc6c7 Kd4c4 Kc7c6 h4h5 Kc6c7 h5xg6 f7xg6
---------------------------------------------------------------------------
 16/30	00:03	   1.455.864	456.670	+1,13	Rd4a4 Kg7f6 Kg2f3 Kf6e5 Kf3g4 Ke5d5 h4h5 Kd5e5 h5xg6 f7xg6 Ra4a1 Ke5d5 Ra1d1+ Kd5c5 Rd1h1 h6h5+ Kg4g5 Kc5c4 Rh1d1 Kc4b5 f2f4
 16/30	00:03	   1.455.864	456.670	+1,93	Rd4d5 Kg7f6 Kg2f3 Kf6e6 Rd5b5 Ke6d7 Kf3e4 Kd7c6 Rb5d5 Ra6a8 Ke4d4 Kc6c7 Kd4c4 Kc7c6 h4h5 Kc6c7 h5xg6 f7xg6
---------------------------------------------------------------------------
 17/30	00:03	   1.721.416	474.873	+1,13	Rd4a4 Kg7f6 Kg2f3 Kf6e5 Kf3g4 Ke5d5 h4h5 Kd5e5 h5xg6 f7xg6 Ra4a1 Ke5e4 Ra1e1+ Ke4d4 Re1d1+ Kd4e4 f2f3+ Ke4e3 Rd1e1+ Ke3d4 Re1h1 h6h5+ Kg4g5 Kd4e3 f3f4 Ke3f3
 17/30	00:03	   1.721.416	474.873	+1,93	Rd4d5 Kg7f6 Kg2f3 Kf6e6 Rd5b5 Ke6d7 Kf3e4 Kd7c6 Rb5d5 Ra6a8 Ke4d4 Kc6c7 Kd4c4 Kc7c6 h4h5 Kc6c7 h5xg6 f7xg6
---------------------------------------------------------------------------
 18/30	00:04	   2.031.409	490.560	+1,13	Rd4a4 Kg7f6 Kg2f3 Kf6e5 Kf3g4 Ke5d5 h4h5 Kd5e5 h5xg6 f7xg6 Ra4a1 Ke5e4 Ra1e1+ Ke4d4 Re1d1+ Kd4e4 f2f3+ Ke4e3 Rd1e1+ Ke3d4 Re1h1 h6h5+ Kg4g5 Kd4e3 f3f4 Ke3f3
 18/30	00:04	   2.031.409	490.560	+1,93	Rd4d5 Kg7f6 Kg2f3 Kf6e6 Rd5b5 Ke6d7 Kf3e4 Kd7c6 Rb5d5 Ra6a8 Ke4d4 Ra8e8 h4h5 Re8e2 Rd5c5+ Kc6b7 h5xg6 f7xg6 f2f4 Re2d2+ Kd4e4 Rd2a2 Ke4e5 h6h5 Rc5b5+ Kb7c7
---------------------------------------------------------------------------
 19/31	00:06	   3.589.214	575.655	+1,21	Rd4a4 Kg7f6 Kg2f3 Kf6e5 Kf3g4 f7f5+ Kg4f3 g6g5 h4xg5 h6xg5 Kf3e3 Ke5d5 Ke3d3 Kd5c5 Ra4a1 Ra6d6+ Kd3c3 Rd6a6 f2f3 Ra6a7 Kc3d3 Ra7a6 f3f4 Ra6d6+ Kd3c3 g5xf4 g3xf4
 19/31	00:06	   3.589.214	577.136	+1,93	Rd4d5 Kg7f6 Kg2f3 Kf6e6 Rd5b5 Ke6d7 Kf3e4 Kd7c6 Rb5d5 Ra6a8 Ke4d4 Ra8e8 h4h5 Re8e2 Rd5c5+ Kc6b7 h5xg6 f7xg6 f2f4 Re2d2+ Kd4e4 Rd2a2 Ke4e5 h6h5 Rc5b5+ Kb7c7
---------------------------------------------------------------------------
 20/36	00:10	   6.978.739	652.950	+1,17	Rd4a4 Kg7f6 Kg2f3 Kf6f5 g3g4+ Kf5e5 Kf3e3 Ke5d5 h4h5 g6xh5 g4xh5 Kd5c5 Ra4f4 Ra6xa5 Rf4xf7 Kc5c4 Rf7c7+ Kc4d5 Rc7g7 Kd5c4 Ke3e4 Ra5xh5 f2f4 Rh5h4 Rg7g6 Kc4c5
 20/36	00:10	   6.978.739	653.929	+1,93	Rd4d5 Kg7f6 Kg2f3 Kf6e6 Rd5b5 Ke6d7 Kf3e4 h6h5 Ke4d4 Kd7c7 Kd4c4 f7f6 f2f3 Ra6e6 Kc4d4 Re6a6 Rb5c5+ Kc7b7 g3g4 h5xg4 f3xg4 Ra6d6+ Rc5d5
---------------------------------------------------------------------------
 21/45	00:18	  13.055.220	703.899	+1,33	Rd4a4 Kg7f6 Kg2f3 Kf6f5 g3g4+ Kf5e5 Kf3e3 Ke5d5 h4h5 g6xh5 g4xh5 Kd5e5 f2f3 Ke5d5 Ke3d3 Kd5c5 Ra4f4 Ra6xa5 Rf4xf7 Kc5b4 Rf7b7+ Kb4c5 Rb7h7 Ra5a6 Kd3e4 Ra6e6+ Ke4f5 Re6d6 f3f4 Kc5d4 Rh7e7
 21/45	00:18	  13.055.220	703.899	+1,97	Rd4d5 Kg7f6 Kg2f3 Kf6e6 Kf3e4 Ra6c6 h4h5 Rc6c2 Rd5b5 g6xh5 f2f4 Rc2a2 Rb5e5+ Ke6d7 Re5xh5 Ra2a4+ Ke4e3 Ra4a2 Ke3d3 Ra2a4 Kd3e3
---------------------------------------------------------------------------
 22/45	00:23	  16.767.575	713.969	+1,33	Rd4a4 Kg7f6 Kg2f3 Kf6f5 g3g4+ Kf5e5 Kf3e3 Ke5d5 h4h5 g6xh5 g4xh5 Kd5e5 f2f3 Ke5d5 Ke3d3 Kd5c5 Ra4f4 Ra6xa5 Rf4xf7 Kc5b4 Rf7b7+ Kb4c5 Rb7h7 Ra5a6 Kd3e4 Ra6e6+ Ke4f5 Re6d6 f3f4 Kc5d4 Rh7e7
 22/45	00:23	  16.767.575	713.969	+2,06	Rd4d5 Kg7f6 Kg2f3 Kf6e6 Kf3e4 Ra6c6 h4h5 Rc6c2 Rd5b5 g6xh5 f2f4 Rc2a2 Rb5e5+ Ke6d7 Re5xh5 Ra2a4+ Ke4e3 Kd7c7 Ke3d3 Ra4a3+ Kd3d4 Ra3a4+ Kd4c3 Ra4a3+ Kc3c4 Ra3xg3 Rh5xh6 Rg3g4
---------------------------------------------------------------------------
 23/45	00:27	  19.620.910	717.978	+1,33	Rd4a4 Kg7f6 Kg2f3 Kf6f5 g3g4+ Kf5e5 Kf3e3 Ke5d5 h4h5 g6xh5 g4xh5 Kd5e5 f2f3 Ke5d5 Ke3d3 Kd5c5 Ra4f4 Ra6xa5 Rf4xf7 Kc5b4 Rf7b7+ Kb4c5 Rb7h7 Ra5a6 Kd3e4 Ra6e6+ Ke4f5 Re6d6 f3f4 Kc5d4 Rh7e7
 23/45	00:27	  19.620.910	717.978	+2,06	Rd4d5 Kg7f6 Kg2f3 Kf6e6 Kf3e4 Ra6c6 h4h5 Rc6c2 Rd5b5 g6xh5 f2f4 Rc2a2 Rb5e5+ Ke6d7 Re5xh5 Ra2a4+ Ke4e3 Ra4a3+ Ke3d4 Ra3xg3 Rh5xh6 Rg3f3 Rh6f6 Kd7e7 Rf6f5 Rf3a3 Rf5b5 Ra3a4+ Kd4e5
---------------------------------------------------------------------------
 24/45	01:10	  53.457.653	761.635	+1,49	Rd4a4 Kg7f6 Kg2f3 Kf6f5 g3g4+ Kf5e5 Kf3e3 Ke5d5 h4h5 g6xh5 g4xh5 Kd5e5 f2f3 Ke5d5 Ke3d3 Kd5c5 Ra4f4 Ra6xa5 Rf4xf7 Kc5b4 Rf7b7+ Kb4c5 Rb7h7 Ra5a6 Kd3e4 Ra6e6+ Ke4f5 Re6d6 f3f4 Kc5d4 Kf5g4 Kd4d5 Rh7g7 Kd5d4 f4f5 Kd4e5
 24/45	01:10	  53.457.653	761.635	+2,10	Rd4d5 Kg7f6 Kg2f3 Kf6e6 Kf3e4 Ra6c6 h4h5 Rc6c2 Rd5b5 g6xh5 f2f4 Rc2e2+ Ke4d4 Re2h2 Rb5b6+ Ke6f5 Rb6b7 Rh2a2 Rb7xf7+ Kf5g4 Rf7g7+ Kg4f5 Rg7a7 Ra2a4+ Kd4c5 Ra4a3 Ra7a6 Ra3xg3 Ra6xh6 Rg3c3+ Kc5d4
---------------------------------------------------------------------------
 25/48	01:58	  89.025.527	750.674	+1,49	Rd4a4 Kg7f6 Kg2f3 Kf6f5 g3g4+ Kf5e5 Kf3e3 Ke5d5 h4h5 g6xh5 g4xh5 Kd5e5 f2f3 Ke5d5 Ke3d3 Kd5c5 Ra4f4 Ra6xa5 Rf4xf7 Kc5b4 Rf7b7+ Kb4c5 Rb7h7 Ra5a6 Kd3e4 Ra6e6+ Ke4f5 Re6d6 f3f4 Kc5d4 Kf5g4 Kd4d5 Rh7g7 Kd5d4 f4f5 Kd4e5
 25/48	01:58	  89.025.527	750.674	+2,06	Rd4d5 h6h5 Kg2f3 Kg7f8 Kf3e4 Kf8e7 f2f3 f7f5+ Ke4d4 Ra6c6 Rd5c5 Rc6a6 Kd4c4 Ke7d6 Rc5d5+ Kd6e6 Rd5c5 Ke6d6
---------------------------------------------------------------------------
 26/51	04:33	 168.258.680	614.188	+1,49	Rd4a4 Kg7f6 Kg2f3 Kf6e5 Kf3e3 Ke5d5 f2f3 Kd5c5 h4h5 g6g5 Ke3e4 Ra6e6+ Ke4f5 Re6a6 f3f4 g5xf4 Kf5xf4 Ra6f6+ Kf4e3 Rf6a6 Ra4f4 f7f6 Ke3d3 Kc5b5 Rf4f5+
 26/51	04:33	 168.258.680	614.188	+2,14	Rd4d5 Kg7f6 Rd5b5 Kf6e7 h4h5 Ke7d6 Kg2f3 Kd6c6 Rb5e5 Kc6d6 Kf3e4 g6g5 Re5d5+ Kd6c7 Rd5c5+ Kc7d7 Rc5b5 Kd7e7 Rb5b7+ Ke7f8 Rb7b8+ Kf8e7 Rb8b5 Ra6a8 Ke4d4 Ra8a6 Kd4c4 Ke7d6 Rb5f5 Ra6a7 Kc4b5 Ra7b7+ Kb5a4 Rb7d7 a5a6 Kd6e6 Rf5b5 f7f6 Rb5b6+ Ke6e5 f2f4+

Regards from Spain.

Ajedrecista.

[Terry McCracken]

[D]8/5pk1/r5pp/P7/3R3P/6P1/5PK1/8 w - - 0 54

From Sune Larsson endgame test. "1927 Alekhine won the 34th game in the battle vs Capablanca to become World
Champion. He did it by playing 1.Ra4! in the above position. Black's rook has
to block the pawn and white has at his disposal an endless reserve of rook
moves along the a-file. Again, the future for black spells "Zugzwang". The
general rule is that a passed pawn should be supported by a rook from the rear.
That's why ideas of 1.Rd5? are wrong. This is a knowledge that have to be
implemented in many chessprograms."

Is this true? I noticed, that in Pavel Hase test Houdini and Rybka played Rd5 and won! So I played 6 top engines from this position after A Ra4 and B Rd5 (game in 20 minutes). In A white won 25/30 games and in B
24/30. Average length of wins were 28 moves in A and 26 in B. So I have feeling both moves are equally winning. BTW Komodo got best score in test even if it were only 1 cpu program..

Jouni

That's no way to test which is better. The solid way and likely the only way to win with near best play is Ra4, solidly protecting the "a" pawn. Rd5?! doesn't achieve this goal from a technical standpoint.

The engine that played Rd5 doesn't "understand" this.

[zullil]

likely the only way to win with near best play is Ra4

So your claim is that, with best play, the following position is a draw?

[D]8/5pk1/r5pp/P2R4/7P/6P1/5PK1/8 b - - 1 1

[Terry McCracken]

likely the only way to win with near best play is Ra4

So your claim is that, with best play, the following position is a draw?

[D]8/5pk1/r5pp/P2R4/7P/6P1/5PK1/8 b - - 1 1

It may not be but yes it could be.

Grandmaster analysis is needed to confirm this. I mean top human grandmaster not machine, but a computer could be used to assist of course.

The safe win is Ra4. Rd5 is dubious but may win. It will take longer I would think.

[Jouni]

One addition: after 1 hour and 29 Rybka_plies 2 best analysis

Analysis by Rybka 4.1:

1. +- (2.38): 1.Rd5 Rc6 2.Rb5 Rc3 3.a6 Ra3 4.Rb6 h5 5.Kf1 Ra2 6.Ke1 Kh6 7.f3 Ra3 8.Ke2 Kg7 9.Kf2 Kf8 10.g4 Ke8 11.Kg3 Ra4 12.Rc6 Kd7 13.Rf6[] Ke7 14.g5 Kf8 15.Rc6 Ke8 16.Rb6
2. +- (1.80): 1.Ra4 Kf6 2.Kf3 Ke5 3.Ke3 h5 4.f4+ Kd5 5.f5 Ke5 6.fxg6 fxg6 7.Kf3 Rf6+ 8.Kg2 Ra6 9.Kh3 Kd5 10.g4 hxg4+ 11.Kxg4 Ke6 12.Kg5 Kf7 13.Ra1 Kg7 14.Kf4 Kf6 15.Ra4 Ke7 16.Kg5

I remember, that +2.xx is 95% win for Rybka.

Jouni

[Terry McCracken]

One addition: after 1 hour and 29 Rybka_plies 2 best analysis

Analysis by Rybka 4.1:

1. +- (2.38): 1.Rd5 Rc6 2.Rb5 Rc3 3.a6 Ra3 4.Rb6 h5 5.Kf1 Ra2 6.Ke1 Kh6 7.f3 Ra3 8.Ke2 Kg7 9.Kf2 Kf8 10.g4 Ke8 11.Kg3 Ra4 12.Rc6 Kd7 13.Rf6[] Ke7 14.g5 Kf8 15.Rc6 Ke8 16.Rb6
2. +- (1.80): 1.Ra4 Kf6 2.Kf3 Ke5 3.Ke3 h5 4.f4+ Kd5 5.f5 Ke5 6.fxg6 fxg6 7.Kf3 Rf6+ 8.Kg2 Ra6 9.Kh3 Kd5 10.g4 hxg4+ 11.Kxg4 Ke6 12.Kg5 Kf7 13.Ra1 Kg7 14.Kf4 Kf6 15.Ra4 Ke7 16.Kg5

I remember, that +2.xx is 95% win for Rybka.

Jouni

That doesn't mean the computers have found the most economical win.

Give this position to any of the top GM's and the evaluation will be different.

Computers find wins in strange ways when they lack certain knowledge and it's beyond their horizons.

I think that is what is happening here.

[jdart]

Computers find wins in strange ways when they lack certain knowledge and it's beyond their horizons.

This is true. But I think it is also true that a large proportion of pre-computer endgame analysis, even from GMs, is faulty.

--Jon

[Houdini]

[D]8/5pk1/r5pp/P7/3R3P/6P1/5PK1/8 w - - 0 54

That's why ideas of 1.Rd5? are wrong. This is a knowledge that have to be
implemented in many chessprograms."

Houdini 2 has an evaluation term for this situation, giving a bonus for the white rook behind the passed pawn and a penalty for the black rook in front of the passed pawn.
But apparently this is not enough to compensate for the strong, centralized position of the rook at d5.

Robert

[Terry McCracken]

Computers find wins in strange ways when they lack certain knowledge and it's beyond their horizons.

This is true. But I think it is also true that a large proportion of pre-computer endgame analysis, even from GMs, is faulty.

--Jon

Not this time. This is technique.