Top engines without tablebases don't play as good as top GMs

[Don]

I checked this position in youtube.com and realized that even a master like myself or used to be 19 years ago rated 2200, and since then have read several books on endings, plus I took a quick look at this position and the explanation and decided to play versus Komodo at 3 minutes per move using my AMD Athlon (tm) Duo Core Processor 4850e 4.50 GHz

http://www.youtube.com/watch?v=5SgNPwe_7B0

This is the initial position:
[D]2b5/2k3p1/p1p2pBp/P2p4/1P1P4/5PP1/7P/4K3 w - - 0 1"

Jorge,

This position is an outright win for white. I suspected that as soon as I saw the position but wasn't sure, now I am.

Black has no intrusion squares and white does. White can attack the 'a' pawn with his bishop and tie black down and that is probably one of the biggest factors. It's really pretty simple.

You say you are a master or used to be and I mean no disrespect but I cannot believe that you feel this is a good example of human superiority over computer - the concept of seeing if you can beat a computer with a winning position. When I said that I believe computers are superior to humans in the endgame, I didn't mean they could even overcome dead lost endings.

I played this position with computer vs computer many different times on many different levels and white won every single time. I also played Houdini vs Komodo which BOTH colors and white won both games. Then I played Komodo as white vs Houdni giving Houdini a 10 to 1 time odds advantage and Komodo still won. So the superiority of white is so large it can withstand a several hundred ELO difference in strength. All programs at any reasonable level finds the same basic plan without any trouble:

1. Invade with white king - to attack the g pawn.
2. Trade off the king side pawns by pushing them forward.

This is supported by harassment of the a pawn which black can do nothing about.

There is another reason that computers play better than humans which is demonstrated by Houdini in this particular starting position. Basically Houdini doesn't have a clue that white has a big advantage here, even on a deep search from the starting position Houdini only see's a score of 32 centi-pawns. And yet it has no trouble seeing the same plan as Komodo and I suspect ANY other reasonable chess program would see.

So YES, computers do not always evaluate position as well as humans but only on occasion is this actually fatal. The best move usually comes out of the search and not knowing this is a "win" doesn't mean it's certain to blunder the position.

In this position Komodo knows right away that white has a big advantage. You thought it was a demonstration of why humans evaluate better but it turns out that Komodo's evaluation is better than yours! Of course I know that our comparison is against the top Grandmasters.

I don't think you would win a single game as black or even draw from this position against Houdini, Komodo, Stockfish or any other half way decent program.

[pichy]

I checked this position in youtube.com and realized that even a master like myself or used to be 19 years ago rated 2200, and since then have read several books on endings, plus I took a quick look at this position and the explanation and decided to play versus Komodo at 3 minutes per move using my AMD Athlon (tm) Duo Core Processor 4850e 4.50 GHz

http://www.youtube.com/watch?v=5SgNPwe_7B0

This is the initial position:
[D]2b5/2k3p1/p1p2pBp/P2p4/1P1P4/5PP1/7P/4K3 w - - 0 1"

Jorge,

This position is an outright win for white. I suspected that as soon as I saw the position but wasn't sure, now I am.

Black has no intrusion squares and white does. White can attack the 'a' pawn with his bishop and tie black down and that is probably one of the biggest factors. It's really pretty simple.

You say you are a master or used to be and I mean no disrespect but I cannot believe that you feel this is a good example of human superiority over computer - the concept of seeing if you can beat a computer with a winning position. When I said that I believe computers are superior to humans in the endgame, I didn't mean they could even overcome dead lost endings.

I played this position with computer vs computer many different times on many different levels and white won every single time. I also played Houdini vs Komodo which BOTH colors and white won both games. Then I played Komodo as white vs Houdni giving Houdini a 10 to 1 time odds advantage and Komodo still won. So the superiority of white is so large it can withstand a several hundred ELO difference in strength. All programs at any reasonable level finds the same basic plan without any trouble:

1. Invade with white king - to attack the g pawn.
2. Trade off the king side pawns by pushing them forward.

This is supported by harassment of the a pawn which black can do nothing about.

There is another reason that computers play better than humans which is demonstrated by Houdini in this particular starting position. Basically Houdini doesn't have a clue that white has a big advantage here, even on a deep search from the starting position Houdini only see's a score of 32 centi-pawns. And yet it has no trouble seeing the same plan as Komodo and I suspect ANY other reasonable chess program would see.

So YES, computers do not always evaluate position as well as humans but only on occasion is this actually fatal. The best move usually comes out of the search and not knowing this is a "win" doesn't mean it's certain to blunder the position.

In this position Komodo knows right away that white has a big advantage. You thought it was a demonstration of why humans evaluate better but it turns out that Komodo's evaluation is better than yours! Of course I know that our comparison is against the top Grandmasters.

I don't think you would win a single game as black or even draw from this position against Houdini, Komodo, Stockfish or any other half way decent program.

Yes, this is a better position for white since the formation of the black pawns are on white squares on the Queen side it makes it much harder for black to defend both sides a better position for black but still with a small advantage for white would hac been this
one, and a top GM either Aonian or Kramnik can defend well both sides against any of the top 3 programs:
[D]2bk4/p1p3p1/3p1pBp/P7/1P1P4/5PP1/7P/4K3 w - - 2 1

[Don]

Yes, this is a better position for white since the formation of the black pawns are on white squares on the Queen side it makes it much harder for black to defend both sides a better position for black but still with a small advantage for white would hac been this
one, and a top GM either Aonian or Kramnik can defend well both sides against any of the top 3 programs:
[D]2bk4/p1p3p1/3p1pBp/P7/1P1P4/5PP1/7P/4K3 w - - 2 1

I have an update. On a deeper level believe it or not Houdini was not able to win this ending in one of the games - essentially it just refused to invade with it's king. For whatever reason Houdini is stuck on an evaluation of "32" which might be a draw score perhaps and cannot make progress. I don't know why but it's probably related to overly aggressive null move pruning or some search heuristic or possibly zugzwang because it reports several ply deeper but is not able to see anything.

But now withstanding the Houdini glitch I think I disagree with you in general about this. I don't believe any top player could avoid losing against Komodo from this position if you give Komodo white - I think this is a clear win for white.

I'm still curious about the Houdini glitch, there is still a very remote possibility that Houdini see's some kind of defense for black if he invades, so I'm going to run this one more time - giving Houdini black and a much longer time control than previously and I will look to see if there is some idea I missed.

[Don]

Yes, this is a better position for white since the formation of the black pawns are on white squares on the Queen side it makes it much harder for black to defend both sides a better position for black but still with a small advantage for white would hac been this
one, and a top GM either Aonian or Kramnik can defend well both sides against any of the top 3 programs:
[D]2bk4/p1p3p1/3p1pBp/P7/1P1P4/5PP1/7P/4K3 w - - 2 1

I have an update. On a deeper level believe it or not Houdini was not able to win this ending in one of the games - essentially it just refused to invade with it's king. For whatever reason Houdini is stuck on an evaluation of "32" which might be a draw score perhaps and cannot make progress. I don't know why but it's probably related to overly aggressive null move pruning or some search heuristic or possibly zugzwang because it reports several ply deeper but is not able to see anything.

But now withstanding the Houdini glitch I think I disagree with you in general about this. I don't believe any top player could avoid losing against Komodo from this position if you give Komodo white - I think this is a clear win for white.

I'm still curious about the Houdini glitch, there is still a very remote possibility that Houdini see's some kind of defense for black if he invades, so I'm going to run this one more time - giving Houdini black and a much longer time control than previously and I will look to see if there is some idea I missed.

I played this out once more letting Houdini defend at a very high level, taking basically all day to complete the game. Komodo played at a time control several times faster and still won using the same simple plan, so I am convinced that this is simply a win for white and there is no hidden defense. At least as far as my ability takes me (with help from Komodo and Houdini.)

Don

[Don]

Yes, this is a better position for white since the formation of the black pawns are on white squares on the Queen side it makes it much harder for black to defend both sides a better position for black but still with a small advantage for white would hac been this
one, and a top GM either Aonian or Kramnik can defend well both sides against any of the top 3 programs:
[D]2bk4/p1p3p1/3p1pBp/P7/1P1P4/5PP1/7P/4K3 w - - 2 1

I have an update. On a deeper level believe it or not Houdini was not able to win this ending in one of the games - essentially it just refused to invade with it's king. For whatever reason Houdini is stuck on an evaluation of "32" which might be a draw score perhaps and cannot make progress. I don't know why but it's probably related to overly aggressive null move pruning or some search heuristic or possibly zugzwang because it reports several ply deeper but is not able to see anything.

But now withstanding the Houdini glitch I think I disagree with you in general about this. I don't believe any top player could avoid losing against Komodo from this position if you give Komodo white - I think this is a clear win for white.

I'm still curious about the Houdini glitch, there is still a very remote possibility that Houdini see's some kind of defense for black if he invades, so I'm going to run this one more time - giving Houdini black and a much longer time control than previously and I will look to see if there is some idea I missed.

I played this out once more letting Houdini defend at a very high level, taking basically all day to complete the game. Komodo played at a time control several times faster and still won using the same simple plan, so I am convinced that this is simply a win for white and there is no hidden defense. At least as far as my ability takes me (with help from Komodo and Houdini.)

Don

If anyone is interested, here is the game played using xboard with time handicap of 4 to 1 in favor of Houdini. In reality the difference is not that much because I'm using 32 bit Houdini and 64 bit Komodo.


[Event "Computer Chess Game"]
[Site "greencheeks"]
[Date "2012.03.26"]
[Round "-"]
[White "Komodo"]
[Black "Houdini 1.5"]
[Result "1-0"]
[TimeControl "7200+60"]
[FEN "2b5/2k3p1/p1p2pBp/P2p4/1P1P4/5PP1/7P/4K3 w - - 0 1"]
[SetUp "1"]

{--------------
. . b . . . . .
. . k . . . p .
p . p . . p B p
P . . p . . . .
. P . P . . . .
. . . . . P P .
. . . . . . . P
. . . . K . . .
white to play
--------------}
1. Kf2 {+1.28/26} Kd6 {-0.50/33 20:00} 2. g4 {+1.28/24 4} Be6
{-0.49/32 2:28} 3. Kg3 {+1.28/25 4} Bc8 {-0.50/31 2:24} 4. Kh3
{+1.24/27 33} Ke7 {-0.50/32 2:28} 5. Kh4 {+1.22/25 4} Kd6 {-0.50/31 16:06}
6. Kh5 {+1.19/29 1:00} Ke7 {-0.97/30 13:35} 7. Bb1 {+1.17/29 1:04} Ke8
{-1.08/27 2:11} 8. Kg6 {+1.18/27 26} Kf8 {-2.64/27 11:18} 9. Bd3
{+1.14/24 5} Kg8 {-4.07/25 9:29} 10. Bf1 {+1.11/28 59} Bb7 {-2.86/24 2:11}
11. f4 {+1.38/27 12} Bc8 {-1.25/27 1:50} 12. g5 {+2.12/27 48} hxg5
{-3.33/24 9:24} 13. fxg5 {+2.21/27 19} fxg5 {-3.42/24 2:45} 14. Kxg5
{+2.21/27 16} Kf7 {-3.60/24 3:30} 15. Bd3 {+2.21/28 21} Ke8 {-3.48/24 1:33}
16. Kg6 {+2.67/28 1:07} Kf8 {-3.46/26 2:09} 17. Be2 {+2.62/29 59} Kg8
{-3.46/26 1:52} 18. h3 {+2.76/28 33} Kf8 {-3.46/26 1:37} 19. Bd3
{+2.92/28 44} Kg8 {-3.46/24 6:14} 20. Bf1 {+2.95/25 13} Kf8 {-3.91/24 3:35}
21. Be2 {+3.02/24 6} Bxh3 {-3.53/23 1:44} 22. Bxa6 {+3.03/24 7} Ke7
{-4.11/26 3:30} 23. Kxg7 {+3.40/28 47} Bg4 {-4.13/26 2:10} 24. Bd3
{+3.56/25 7} Kd7 {-5.34/27 4:54} 25. Kf6 {+4.28/29 54} Kc7 {-5.48/26 1:26}
26. a6 {+5.75/28 56} Kb8 {-7.72/27 4:52} 27. Ke7 {+6.22/29 45} Kc7
{-5.71/24 0.1} 28. a7 {+6.69/26 15} Kb7 {-7.64/23 44} 29. Kd6 {+7.42/28 49}
Bf3 {-9.88/25 1:24} 30. Bf5 {+7.74/30 55} Be4 {-10.27/26 1:34} 31. Be6
{+7.68/25 1:25} Bd3 {-10.25/27 1:41} 32. a8=Q+ {+10.21/27 44} Kxa8
{-11.30/26 0.1} 33. Kxc6 {+11.55/23 8} Kb8 {-10.45/28 1:06} 34. b5
{+10.21/24 6} Be2 {-14.48/28 3:45} 35. Bxd5 {+317.97/25 1:14} Bg4
{-10.62/26 0.1} 36. Bg8 {+317.97/21 1:35} Kc8 {-99.08/24 2:52} 37. d5
{+99.35/23 3:49} Bd7+ {-99.60/28 1:23} 38. Kc5 {+99.69/22 2:20} Kd8
{-99.62/28 2:25} 39. Be6 {+99.73/20 50} Bxb5 {-99.76/30 1:11} 40. Kxb5
{+99.73/18 15} Kc7 {-99.78/32 26} 41. Kc5 {+99.79/21 30} Kd8
{-99.80/30 0.1} 42. d6 {+99.79/19 53} Ke8 {-99.82/1 0.1} 43. Kb6
{+99.83/19 37} Kd8 {-99.84/26 0.1} 44. Kb7 {+99.85/19 2:54} Ke8
{-99.76/1 0.1} 45. Kc7 {+99.87/18 2:34} Kf8 {-99.88/1 0.1} 46. d7
{+99.89/17 33} Ke7 {-99.90/20 0.1} 47. d8=Q+ {+99.91/17 1:02} Kxe6
{-99.92/1 0.1} 48. Qg5 {+99.93/17 29} Kf7 {-99.94/1 0.1} 49. Kd6
{+99.95/17 45} Kf8 {-99.96/14 0.1} 50. Ke6 {+99.97/16 34} Ke8
{-99.98/1 0.1} 51. Qe7# {+99.99/17 1:09}
{Xboard adjudication: Checkmate} 1-0

[lech]

Lesson 2:
It is possible to get this possition on a very high ply. It needs 20 additional depths to verify that it is a wrong choise.

[d] 8/8/4k3/8/3PK3/7p/7P/8 w - - 0 1

I added 3 plies.
Would you like to test your liked engine: in how many depths (or how much time) is able to find the best move 1...Re6 with draw (0,00) pawn endgame?

[d]8/5k2/7r/8/3P4/4R2p/4K2P/8 b - - 0 1

[Don]

Lesson 2:
It is possible to get this possition on a very high ply. It needs 20 additional depths to verify that it is a wrong choise.

[d] 8/8/4k3/8/3PK3/7p/7P/8 w - - 0 1

I added 3 plies.
Would you like to test your liked engine: in how many depths (or how much time) is able to find the best move 1...Re6 with draw (0,00) pawn endgame?

[d]8/5k2/7r/8/3P4/4R2p/4K2P/8 b - - 0 1

I'm sure you can find positions that computers do not handle well. What is the point of all of this? I do not dispute that. I've been doing this for several decades now and I know more about the weaknesses of programs than you do. You are not imparting any special insight here.

Anyway, here is the analysis by our current development version of Komodo, it takes longer than before as expected.


30 -0.10 39.1M 0:28.96 Re6 Kf3 Rxe3+ Kxe3 Ke6 Ke4 Kd6 d5 Kd7 Ke5 Ke7 d6+ Kd8 Kd4 Kc8 Ke4 Kd7 Kd5 Kd8 Kc6 Kc8 Kb6 Kd7 Kc5 Kd8 Kb4 Kd7 Kc3 Kd8 Kd3 Kd7 Ke4 Kxd6 Kd4 Ke6 Ke4 Kd6
30 -0.10 37.1M 0:27.07 Re6 Kf3 Rxe3+ Kxe3 Ke6 Ke4 Kd6 d5 Kd7 Ke5 Ke7 d6+ Kd8 Kd4 Kc8 Ke4 Kd7 Kd5 Kd8 Kc6 Kc8 Kb6 Kd7 Kc5 Kd8 Kb4 Kd7 Kc3 Kd8 Kd3 Kd7 Ke4 Kxd6 Kd4 Ke6 Ke4 Kd6
29 -0.10 27.1M 0:21.07 Re6 Kf3 Rxe3+ Kxe3 Ke6 Ke4 Kd6 d5 Kd7 Ke5 Ke7 d6+ Kd8 Kd4 Kc8 Ke4 Kd7 Kd5 Kd8 Kc6 Kc8 Kb5 Kd7 Kc5 Kd8 Kc6
29 -0.10 26.2M 0:20.30 Re6 Kf3 Rxe3+ Kxe3 Ke6 Ke4 Kd6 d5 Kd7 Ke5 Ke7 d6+ Kd8 Kd4 Kc8 Ke4 Kd7 Kd5 Kd8 Kc6 Kc8 Kb5 Kd7 Kc5 Kd8 Kc6
28 -0.10 19.8M 0:16.44 Re6 Kf3 Rxe3+ Kxe3 Ke6 Ke4 Kd6 d5 Kd7 Ke5 Ke7 d6+ Kd8 Kd5 Kd7 Kc5 Kd8 Kd4 Kd7 Kd5 Kd8 Ke5 Kd7 Kd5
28 -0.10 18.5M 0:15.19 Re6 Kf3 Rxe3+ Kxe3 Ke6 Ke4 Kd6 d5 Kd7 Ke5 Ke7 d6+ Kd8 Kd5 Kd7 Kc5 Kd8 Kd4 Kd7 Kd5 Kd8 Ke5 Kd7 Kd5
27 -0.10 15.5M 0:13.34 Re6 Kf3 Rxe3+ Kxe3 Ke6 Ke4 Kd6 d5 Kd7 Ke5 Ke7 d6+ Kd8 Kd5 Kd7 Kc5 Kd8 Kd4 Kd7 Kd5 Kd8 Kd4
27 -0.10 15.1M 0:13.02 Re6 Kf3 Rxe3+ Kxe3 Ke6 Ke4 Kd6 d5 Kd7 Ke5 Ke7 d6+ Kd8 Kd5 Kd7 Kc5 Kd8 Kd4 Kd7 Kd5 Kd8 Kd4
26 -0.10 13.1M 0:11.86 Re6 Kf3 Rxe3+ Kxe3 Ke6 Ke4 Kd6 d5 Kd7 Ke5 Ke7 d6+ Kd8 Kd5 Kd7 Kc5 Kd8 Kd4 Kc8 Ke4 Kd8 Ke5 Kd7 Kd5 Kd8 Ke5
26 -0.10 13.0M 0:11.73 Re6 Kf3 Rxe3+ Kxe3 Ke6 Ke4 Kd6 d5 Kd7 Ke5 Ke7 d6+ Kd8 Kd5 Kd7 Kc5 Kd8 Kd4 Kc8 Ke4 Kd8 Ke5 Kd7 Kd5 Kd8 Ke5
25 -0.10 12.0M 0:11.11 Re6 Kf3 Rxe3+ Kxe3 Ke6 Ke4 Kd6 d5 Kd7 Ke5 Ke7 d6+ Kd8 Kd5 Kd7 Kc5 Kd8 Kd4 Kd7 Kd5 Kd8 Kc5 Kd7 Kd5
25 -0.10 11.8M 0:10.99 Re6 Kf3 Rxe3+ Kxe3 Ke6 Ke4 Kd6 d5 Kd7 Ke5 Ke7 d6+ Kd8 Kd5 Kd7 Kc5 Kd8 Kd4 Kd7 Kd5 Kd8 Kc5 Kd7 Kd5
24 -0.10 11.3M 0:10.67 Re6 Kf3 Rxe3+ Kxe3 Ke6 Ke4 Kd6 d5 Kd7 Ke5 Ke7 d6+ Kd8 Kd5 Kd7 Kc5 Kd8 Kd4 Ke8 Ke4 Kd8 Ke5 Kd7 Kd5 Kd8 Ke5
24 -0.10 11.1M 0:10.46 Re6 Kf3 Rxe3+ Kxe3 Ke6 Ke4 Kd6 d5 Kd7 Ke5 Ke7 d6+ Kd8 Kd5 Kd7 Kc5 Kd8 Kd4 Ke8 Ke4 Kd8 Ke5 Kd7 Kd5 Kd8 Ke5
23 -0.10 10.8M 0:10.27 Re6 Kf3 Rxe3+ Kxe3 Ke6 Ke4 Kd6 d5 Kd7 Ke5 Ke7 d6+ Kd8 Kd5 Kd7 Kc5 Kd8 Kd4 Kd7 Kd5 Kd8 Kd4
23 -0.10 10.7M 0:10.20 Re6 Kf3 Rxe3+ Kxe3 Ke6 Ke4 Kd6 d5 Kd7 Ke5 Ke7 d6+ Kd8 Kd5 Kd7 Kc5 Kd8 Kd4 Kd7 Kd5 Kd8 Kd4
22 -0.10 10.5M 0:10.04 Re6 Kf3 Rxe3+ Kxe3 Ke6 Ke4 Kd6 d5 Kd7 Ke5 Ke7 d6+ Kd8 Kd5 Kd7 Kc5 Kd8 Kd4 Kd7 Kd5 Kd8 Kd4
22 -0.10 10.4M 0:09.99 Re6 Kf3 Rxe3+ Kxe3 Ke6 Ke4 Kd6 d5 Kd7 Ke5 Ke7 d6+ Kd8 Kd5 Kd7 Kc5 Kd8 Kd4 Kd7 Kd5 Kd8 Kd4
22 -0.54 10.1M 0:09.80 Re6
22 -0.69 9.7M 0:09.40 Kf6 Kf3 Rh4 Re4 Rh8 Re5 Rh4 Ke3 Rh7 Ke4 Rh4+ Kd3 Rh7 Ra5 Rh8 Ke4 Rh4+ Ke3 Rh8 Kd3 Rb8 Ra2 Kf5 Rf2+ Ke6 Ke4
21 -0.70 2.6M 0:02.44 Kf6 Kd2 Rh8 Kd3 Kf7 Kc4 Rc8+ Kb5 Rd8 Rf3+ Ke6 Kc4 Rc8+ Kd3 Rh8 Ke4 Rh4+ Ke3 Rh7 Rg3 Rh4 Rg7 Kf6 Rc7 Rh8 Ke4 Re8+ Kf4 Re2 Rc6+ Ke7
21 -0.70 2.3M 0:02.15 Kf6 Kd2 Rh8 Kd3 Kf7 Kc4 Rc8+ Kb5 Rd8 Rf3+ Ke6 Kc4 Rc8+ Kd3 Rh8 Ke4 Rh4+ Ke3 Rh7 Rg3 Rh4 Rg7 Kf6 Rc7 Rh8 Ke4 Re8+ Kf4 Re2 Rc6+ Ke7
20 -0.70 1.8M 0:01.73 Kf6 Kd2 Rh8 Kd3 Kf7 Kc4 Rc8+ Kb5 Rd8 Rf3+ Ke6 Kc4 Rc8+ Kd3 Rh8 Ke4 Rh4+ Ke3 Rh7 Rg3 Kf5 Kd3 Ke6 Ke4 Rh4+ Ke3 Rh8 Rg7 Rb8
20 -0.70 1.7M 0:01.62 Kf6 Kd2 Rh8 Kd3 Kf7 Kc4 Rc8+ Kb5 Rd8 Rf3+ Ke6 Kc4 Rc8+ Kd3 Rh8 Ke4 Rh4+ Ke3 Rh7 Rg3 Kf5 Kd3 Ke6 Ke4 Rh4+ Ke3 Rh8 Rg7 Rb8
19 -0.70 1.3M 0:01.25 Kf6 Kd2 Rh8 Kd3 Kf7 Kc4 Rc8+ Kd5 Rd8+ Kc5 Rc8+ Kb5 Rd8 Rf3+ Ke6 Kc4 Rc8+ Kd3 Rh8 Ke4 Rh4+ Ke3 Rh8 Rf2 Re8 Kd3 Rh8
19 -0.70 1.2M 0:01.20 Kf6 Kd2 Rh8 Kd3 Kf7 Kc4 Rc8+ Kd5 Rd8+ Kc5 Rc8+ Kb5 Rd8 Rf3+ Ke6 Kc4 Rc8+ Kd3 Rh8 Ke4 Rh4+ Ke3 Rh8 Rf2 Re8 Kd3 Rh8
18 -0.70 1.1M 0:01.05 Kf6 Kd2 Rh8 Kd3 Kf7 Kc4 Rc8+ Kd5 Rd8+ Kc5 Rc8+ Kb5 Rd8 Rf3+ Ke6 Kc4 Rc8+ Kd3 Rh8 Ke4 Rh4+ Ke3 Rh5 Rf4 Rb5 Ke4 Rb2 d5+ Ke7
18 -0.70 1.1M 0:01.02 Kf6 Kd2 Rh8 Kd3 Kf7 Kc4 Rc8+ Kd5 Rd8+ Kc5 Rc8+ Kb5 Rd8 Rf3+ Ke6 Kc4 Rc8+ Kd3 Rh8 Ke4 Rh4+ Ke3 Rh5 Rf4 Rb5 Ke4 Rb2 d5+ Ke7
18 -0.86 803927 0:00.77 Rh4 Kd3
17 -0.70 555798 0:00.54 Rh4 Kd3 Rh8 Re5 Ra8 Re2 Kf6 Rf2+ Ke6 Ke4 Re8 Rf5 Kd6+ Re5 Rh8 Rd5+ Ke6 Rb5 Rh4+ Kd3 Rh7 Rb8 Ra7 Ke4
17 -0.70 533594 0:00.52 Rh4 Kd3 Rh8 Re5 Ra8 Re2 Kf6 Rf2+ Ke6 Ke4 Re8 Rf5 Kd6+ Re5 Rh8 Rd5+ Ke6 Rb5 Rh4+ Kd3 Rh7 Rb8 Ra7 Ke4
17 -0.72 247469 0:00.24 Kf6 Kd2 Rh8 Kd3 Kf7 Kc4 Rc8+ Kd5 Rd8+ Kc5 Rc8+ Kb5 Rd8 Rf3+ Ke6 Kc4 Rc8+ Kd3 Rh8 Ke3 Kd6 Rf6+ Kd5 Rf5+ Kd6 Ke4
16 -0.70 139417 0:00.13 Kf6 Kd2 Rh8 Kd3 Kf7 d5 Rh4 Kc3 Ra4 Re2 Kf6 d6 Ra6 Rd2 Ra8 Kb4 Ke6 d7 Rd8
16 -0.70 122906 0:00.12 Kf6 Kd2 Rh8 Kd3 Kf7 d5 Rh4 Kc3 Ra4 Re2 Kf6 d6 Ra6 Rd2 Ra8 Kb4 Ke6 d7 Rd8
15 -0.72 85112 0:00.08 Kf6 Kd2 Rh8 Kd3 Kf7 d5 Rh4 Kc3 Ra4 Re2 Kf6 d6 Ra6 Rd2 Ra8 Kb4 Ke6
15 -0.72 74406 0:00.07 Kf6 Kd2 Rh8 Kd3 Kf7 d5 Rh4 Kc3 Ra4 Re2 Kf6 d6 Ra6 Rd2 Ra8 Kb4 Ke6
14 -0.70 45815 0:00.04 Kf6 Kd2 Rh8 Kd3 Kf5 d5 Ra8 Kd4 Ra4+ Kc5 Ra5+ Kc6 Ra6+ Kc7 Ra7+ Kd6 Ra6+ Kd7 Ra7+ Ke8 Ra8+ Kf7 Ra7+ Re7 Ra2
14 -0.70 39843 0:00.04 Kf6 Kd2 Rh8 Kd3 Kf5 d5 Ra8 Kd4 Ra4+ Kc5 Ra5+ Kc6 Ra6+ Kc7 Ra7+ Kd6 Ra6+ Kd7 Ra7+ Ke8 Ra8+ Kf7 Ra7+ Re7 Ra2
13 -0.70 24914 0:00.02 Kf6 Kd2 Kf5 Kd3 Rh8 d5 Ra8 Kd4 Ra4+ Kc5 Ra5+ Kc6 Ra6+ Kc7 Ra7+ Kb6 Ra2 Rxh3 Ke5 Rd3
13 -0.70 21676 0:00.02 Kf6 Kd2 Kf5 Kd3 Rh8 d5 Ra8 Kd4 Ra4+ Kc5 Ra5+ Kc6 Ra6+ Kc7 Ra7+ Kb6 Ra2 Rxh3 Ke5 Rd3
12 -0.70 16664 0:00.02 Kf6 Kd2 Kf5 Kd3 Rh8 d5 Ra8 Kd4 Ra4+ Kc5 Ra5+ Kc6 Ra6+ Kc7 Ra7+ Kb6 Ra2 Rxh3
12 -0.70 14198 0:00.02 Kf6 Kd2 Kf5 Kd3 Rh8 d5 Ra8 Kd4 Ra4+ Kc5 Ra5+ Kc6 Ra6+ Kc7 Ra7+ Kb6 Ra2 Rxh3
11 -0.70 10791 0:00.01 Kf6 Kd2 Kf5 Kd3 Rh8 d5 Ra8 Kd4 Ra4+ Kc5 Ra5+ Kc6 Ra6+ Kc7 Ra7+ Kb6 Ra2
11 -0.70 9999 0:00.01 Kf6 Kd2 Kf5 Kd3 Rh8 d5 Ra8 Kd4 Ra4+ Kc5 Ra5+ Kc6 Ra6+ Kc7 Ra7+ Kb6 Ra2
10 -0.70 6439 0:00.01 Kf6 Kd2 Kf5 Kd3 Rh8 d5 Ra8 Kd4 Ra4+ Kc5 Ra5+ Kc6 Ra6+ Kc7 Ra7+ Kb6
10 -0.70 5318 0:00.01 Kf6 Kd2 Kf5 Kd3 Rh8 d5 Ra8 Kd4 Ra4+ Kc5 Ra5+ Kc6 Ra6+ Kc7 Ra7+ Kb6
9 -0.70 3302 0:00.00 Kf6 Kd2 Kf5 Kd3 Rh8 Rf3+ Ke6 Ke4 Rh4+ Ke3 Rh7
9 -0.70 2806 0:00.00 Kf6 Kd2 Kf5 Kd3 Rh8 Rf3+ Ke6 Ke4 Rh4+ Ke3 Rh7
8 -0.72 1884 0:00.00 Kf6 Kd2 Kf5 Kd3 Rh8 Rf3+ Ke6 Ke4 Rh5
8 -0.72 1646 0:00.00 Kf6 Kd2 Kf5 Kd3 Rh8 Rf3+ Ke6 Ke4 Rh5
7 -0.72 1270 0:00.00 Kf6 Kd2 Kf5 Kd3 Rh8 Rf3+ Ke6 Ke4 Rh5
7 -0.72 1136 0:00.00 Kf6 Kd2 Kf5 Kd3 Rh8 Rf3+ Ke6 Ke4 Rh5
6 -0.70 826 0:00.00 Kf6 Kd2 Kf5 Kd3 Rh8 Rf3+ Ke6
6 -0.70 729 0:00.00 Kf6 Kd2 Kf5 Kd3 Rh8 Rf3+ Ke6
5 -0.70 523 0:00.00 Kf6 Kd2 Kf5 Kd3 Rh8 Rf3+ Ke6
5 -0.70 460 0:00.00 Kf6 Kd2 Kf5 Kd3 Rh8 Rf3+ Ke6
4 -0.75 323 0:00.00 Kf6 Kd2 Kf5 Kd3
3 -0.75 238 0:00.00 Kf6 Kd2 Kf5 Kd3
2 -0.75 106 0:00.00 Kf6 Kd2 Kf5 Kd3
1 -0.82 33 0:00.00 Kf6 Rf3+

[Don]

I just realized my output here from xboard is invalid. I had two instances running and I was experimenting a bit with different moves so the hash table had been pre-loaded. I pasted in the wrong copy.

I'm running it again and it doesn't want to force the trade on depth 31.

However, I'm not sure what this proves except that you can find position computers play much better and some humans play much better.

Lesson 2:
It is possible to get this possition on a very high ply. It needs 20 additional depths to verify that it is a wrong choise.

[d] 8/8/4k3/8/3PK3/7p/7P/8 w - - 0 1

I added 3 plies.
Would you like to test your liked engine: in how many depths (or how much time) is able to find the best move 1...Re6 with draw (0,00) pawn endgame?

[d]8/5k2/7r/8/3P4/4R2p/4K2P/8 b - - 0 1

I'm sure you can find positions that computers do not handle well. What is the point of all of this? I do not dispute that. I've been doing this for several decades now and I know more about the weaknesses of programs than you do. You are not imparting any special insight here.

Anyway, here is the analysis by our current development version of Komodo, it takes longer than before as expected.


30 -0.10 39.1M 0:28.96 Re6 Kf3 Rxe3+ Kxe3 Ke6 Ke4 Kd6 d5 Kd7 Ke5 Ke7 d6+ Kd8 Kd4 Kc8 Ke4 Kd7 Kd5 Kd8 Kc6 Kc8 Kb6 Kd7 Kc5 Kd8 Kb4 Kd7 Kc3 Kd8 Kd3 Kd7 Ke4 Kxd6 Kd4 Ke6 Ke4 Kd6
30 -0.10 37.1M 0:27.07 Re6 Kf3 Rxe3+ Kxe3 Ke6 Ke4 Kd6 d5 Kd7 Ke5 Ke7 d6+ Kd8 Kd4 Kc8 Ke4 Kd7 Kd5 Kd8 Kc6 Kc8 Kb6 Kd7 Kc5 Kd8 Kb4 Kd7 Kc3 Kd8 Kd3 Kd7 Ke4 Kxd6 Kd4 Ke6 Ke4 Kd6
29 -0.10 27.1M 0:21.07 Re6 Kf3 Rxe3+ Kxe3 Ke6 Ke4 Kd6 d5 Kd7 Ke5 Ke7 d6+ Kd8 Kd4 Kc8 Ke4 Kd7 Kd5 Kd8 Kc6 Kc8 Kb5 Kd7 Kc5 Kd8 Kc6
29 -0.10 26.2M 0:20.30 Re6 Kf3 Rxe3+ Kxe3 Ke6 Ke4 Kd6 d5 Kd7 Ke5 Ke7 d6+ Kd8 Kd4 Kc8 Ke4 Kd7 Kd5 Kd8 Kc6 Kc8 Kb5 Kd7 Kc5 Kd8 Kc6
28 -0.10 19.8M 0:16.44 Re6 Kf3 Rxe3+ Kxe3 Ke6 Ke4 Kd6 d5 Kd7 Ke5 Ke7 d6+ Kd8 Kd5 Kd7 Kc5 Kd8 Kd4 Kd7 Kd5 Kd8 Ke5 Kd7 Kd5
28 -0.10 18.5M 0:15.19 Re6 Kf3 Rxe3+ Kxe3 Ke6 Ke4 Kd6 d5 Kd7 Ke5 Ke7 d6+ Kd8 Kd5 Kd7 Kc5 Kd8 Kd4 Kd7 Kd5 Kd8 Ke5 Kd7 Kd5
27 -0.10 15.5M 0:13.34 Re6 Kf3 Rxe3+ Kxe3 Ke6 Ke4 Kd6 d5 Kd7 Ke5 Ke7 d6+ Kd8 Kd5 Kd7 Kc5 Kd8 Kd4 Kd7 Kd5 Kd8 Kd4
27 -0.10 15.1M 0:13.02 Re6 Kf3 Rxe3+ Kxe3 Ke6 Ke4 Kd6 d5 Kd7 Ke5 Ke7 d6+ Kd8 Kd5 Kd7 Kc5 Kd8 Kd4 Kd7 Kd5 Kd8 Kd4
26 -0.10 13.1M 0:11.86 Re6 Kf3 Rxe3+ Kxe3 Ke6 Ke4 Kd6 d5 Kd7 Ke5 Ke7 d6+ Kd8 Kd5 Kd7 Kc5 Kd8 Kd4 Kc8 Ke4 Kd8 Ke5 Kd7 Kd5 Kd8 Ke5
26 -0.10 13.0M 0:11.73 Re6 Kf3 Rxe3+ Kxe3 Ke6 Ke4 Kd6 d5 Kd7 Ke5 Ke7 d6+ Kd8 Kd5 Kd7 Kc5 Kd8 Kd4 Kc8 Ke4 Kd8 Ke5 Kd7 Kd5 Kd8 Ke5
25 -0.10 12.0M 0:11.11 Re6 Kf3 Rxe3+ Kxe3 Ke6 Ke4 Kd6 d5 Kd7 Ke5 Ke7 d6+ Kd8 Kd5 Kd7 Kc5 Kd8 Kd4 Kd7 Kd5 Kd8 Kc5 Kd7 Kd5
25 -0.10 11.8M 0:10.99 Re6 Kf3 Rxe3+ Kxe3 Ke6 Ke4 Kd6 d5 Kd7 Ke5 Ke7 d6+ Kd8 Kd5 Kd7 Kc5 Kd8 Kd4 Kd7 Kd5 Kd8 Kc5 Kd7 Kd5
24 -0.10 11.3M 0:10.67 Re6 Kf3 Rxe3+ Kxe3 Ke6 Ke4 Kd6 d5 Kd7 Ke5 Ke7 d6+ Kd8 Kd5 Kd7 Kc5 Kd8 Kd4 Ke8 Ke4 Kd8 Ke5 Kd7 Kd5 Kd8 Ke5
24 -0.10 11.1M 0:10.46 Re6 Kf3 Rxe3+ Kxe3 Ke6 Ke4 Kd6 d5 Kd7 Ke5 Ke7 d6+ Kd8 Kd5 Kd7 Kc5 Kd8 Kd4 Ke8 Ke4 Kd8 Ke5 Kd7 Kd5 Kd8 Ke5
23 -0.10 10.8M 0:10.27 Re6 Kf3 Rxe3+ Kxe3 Ke6 Ke4 Kd6 d5 Kd7 Ke5 Ke7 d6+ Kd8 Kd5 Kd7 Kc5 Kd8 Kd4 Kd7 Kd5 Kd8 Kd4
23 -0.10 10.7M 0:10.20 Re6 Kf3 Rxe3+ Kxe3 Ke6 Ke4 Kd6 d5 Kd7 Ke5 Ke7 d6+ Kd8 Kd5 Kd7 Kc5 Kd8 Kd4 Kd7 Kd5 Kd8 Kd4
22 -0.10 10.5M 0:10.04 Re6 Kf3 Rxe3+ Kxe3 Ke6 Ke4 Kd6 d5 Kd7 Ke5 Ke7 d6+ Kd8 Kd5 Kd7 Kc5 Kd8 Kd4 Kd7 Kd5 Kd8 Kd4
22 -0.10 10.4M 0:09.99 Re6 Kf3 Rxe3+ Kxe3 Ke6 Ke4 Kd6 d5 Kd7 Ke5 Ke7 d6+ Kd8 Kd5 Kd7 Kc5 Kd8 Kd4 Kd7 Kd5 Kd8 Kd4
22 -0.54 10.1M 0:09.80 Re6
22 -0.69 9.7M 0:09.40 Kf6 Kf3 Rh4 Re4 Rh8 Re5 Rh4 Ke3 Rh7 Ke4 Rh4+ Kd3 Rh7 Ra5 Rh8 Ke4 Rh4+ Ke3 Rh8 Kd3 Rb8 Ra2 Kf5 Rf2+ Ke6 Ke4
21 -0.70 2.6M 0:02.44 Kf6 Kd2 Rh8 Kd3 Kf7 Kc4 Rc8+ Kb5 Rd8 Rf3+ Ke6 Kc4 Rc8+ Kd3 Rh8 Ke4 Rh4+ Ke3 Rh7 Rg3 Rh4 Rg7 Kf6 Rc7 Rh8 Ke4 Re8+ Kf4 Re2 Rc6+ Ke7
21 -0.70 2.3M 0:02.15 Kf6 Kd2 Rh8 Kd3 Kf7 Kc4 Rc8+ Kb5 Rd8 Rf3+ Ke6 Kc4 Rc8+ Kd3 Rh8 Ke4 Rh4+ Ke3 Rh7 Rg3 Rh4 Rg7 Kf6 Rc7 Rh8 Ke4 Re8+ Kf4 Re2 Rc6+ Ke7
20 -0.70 1.8M 0:01.73 Kf6 Kd2 Rh8 Kd3 Kf7 Kc4 Rc8+ Kb5 Rd8 Rf3+ Ke6 Kc4 Rc8+ Kd3 Rh8 Ke4 Rh4+ Ke3 Rh7 Rg3 Kf5 Kd3 Ke6 Ke4 Rh4+ Ke3 Rh8 Rg7 Rb8
20 -0.70 1.7M 0:01.62 Kf6 Kd2 Rh8 Kd3 Kf7 Kc4 Rc8+ Kb5 Rd8 Rf3+ Ke6 Kc4 Rc8+ Kd3 Rh8 Ke4 Rh4+ Ke3 Rh7 Rg3 Kf5 Kd3 Ke6 Ke4 Rh4+ Ke3 Rh8 Rg7 Rb8
19 -0.70 1.3M 0:01.25 Kf6 Kd2 Rh8 Kd3 Kf7 Kc4 Rc8+ Kd5 Rd8+ Kc5 Rc8+ Kb5 Rd8 Rf3+ Ke6 Kc4 Rc8+ Kd3 Rh8 Ke4 Rh4+ Ke3 Rh8 Rf2 Re8 Kd3 Rh8
19 -0.70 1.2M 0:01.20 Kf6 Kd2 Rh8 Kd3 Kf7 Kc4 Rc8+ Kd5 Rd8+ Kc5 Rc8+ Kb5 Rd8 Rf3+ Ke6 Kc4 Rc8+ Kd3 Rh8 Ke4 Rh4+ Ke3 Rh8 Rf2 Re8 Kd3 Rh8
18 -0.70 1.1M 0:01.05 Kf6 Kd2 Rh8 Kd3 Kf7 Kc4 Rc8+ Kd5 Rd8+ Kc5 Rc8+ Kb5 Rd8 Rf3+ Ke6 Kc4 Rc8+ Kd3 Rh8 Ke4 Rh4+ Ke3 Rh5 Rf4 Rb5 Ke4 Rb2 d5+ Ke7
18 -0.70 1.1M 0:01.02 Kf6 Kd2 Rh8 Kd3 Kf7 Kc4 Rc8+ Kd5 Rd8+ Kc5 Rc8+ Kb5 Rd8 Rf3+ Ke6 Kc4 Rc8+ Kd3 Rh8 Ke4 Rh4+ Ke3 Rh5 Rf4 Rb5 Ke4 Rb2 d5+ Ke7
18 -0.86 803927 0:00.77 Rh4 Kd3
17 -0.70 555798 0:00.54 Rh4 Kd3 Rh8 Re5 Ra8 Re2 Kf6 Rf2+ Ke6 Ke4 Re8 Rf5 Kd6+ Re5 Rh8 Rd5+ Ke6 Rb5 Rh4+ Kd3 Rh7 Rb8 Ra7 Ke4
17 -0.70 533594 0:00.52 Rh4 Kd3 Rh8 Re5 Ra8 Re2 Kf6 Rf2+ Ke6 Ke4 Re8 Rf5 Kd6+ Re5 Rh8 Rd5+ Ke6 Rb5 Rh4+ Kd3 Rh7 Rb8 Ra7 Ke4
17 -0.72 247469 0:00.24 Kf6 Kd2 Rh8 Kd3 Kf7 Kc4 Rc8+ Kd5 Rd8+ Kc5 Rc8+ Kb5 Rd8 Rf3+ Ke6 Kc4 Rc8+ Kd3 Rh8 Ke3 Kd6 Rf6+ Kd5 Rf5+ Kd6 Ke4
16 -0.70 139417 0:00.13 Kf6 Kd2 Rh8 Kd3 Kf7 d5 Rh4 Kc3 Ra4 Re2 Kf6 d6 Ra6 Rd2 Ra8 Kb4 Ke6 d7 Rd8
16 -0.70 122906 0:00.12 Kf6 Kd2 Rh8 Kd3 Kf7 d5 Rh4 Kc3 Ra4 Re2 Kf6 d6 Ra6 Rd2 Ra8 Kb4 Ke6 d7 Rd8
15 -0.72 85112 0:00.08 Kf6 Kd2 Rh8 Kd3 Kf7 d5 Rh4 Kc3 Ra4 Re2 Kf6 d6 Ra6 Rd2 Ra8 Kb4 Ke6
15 -0.72 74406 0:00.07 Kf6 Kd2 Rh8 Kd3 Kf7 d5 Rh4 Kc3 Ra4 Re2 Kf6 d6 Ra6 Rd2 Ra8 Kb4 Ke6
14 -0.70 45815 0:00.04 Kf6 Kd2 Rh8 Kd3 Kf5 d5 Ra8 Kd4 Ra4+ Kc5 Ra5+ Kc6 Ra6+ Kc7 Ra7+ Kd6 Ra6+ Kd7 Ra7+ Ke8 Ra8+ Kf7 Ra7+ Re7 Ra2
14 -0.70 39843 0:00.04 Kf6 Kd2 Rh8 Kd3 Kf5 d5 Ra8 Kd4 Ra4+ Kc5 Ra5+ Kc6 Ra6+ Kc7 Ra7+ Kd6 Ra6+ Kd7 Ra7+ Ke8 Ra8+ Kf7 Ra7+ Re7 Ra2
13 -0.70 24914 0:00.02 Kf6 Kd2 Kf5 Kd3 Rh8 d5 Ra8 Kd4 Ra4+ Kc5 Ra5+ Kc6 Ra6+ Kc7 Ra7+ Kb6 Ra2 Rxh3 Ke5 Rd3
13 -0.70 21676 0:00.02 Kf6 Kd2 Kf5 Kd3 Rh8 d5 Ra8 Kd4 Ra4+ Kc5 Ra5+ Kc6 Ra6+ Kc7 Ra7+ Kb6 Ra2 Rxh3 Ke5 Rd3
12 -0.70 16664 0:00.02 Kf6 Kd2 Kf5 Kd3 Rh8 d5 Ra8 Kd4 Ra4+ Kc5 Ra5+ Kc6 Ra6+ Kc7 Ra7+ Kb6 Ra2 Rxh3
12 -0.70 14198 0:00.02 Kf6 Kd2 Kf5 Kd3 Rh8 d5 Ra8 Kd4 Ra4+ Kc5 Ra5+ Kc6 Ra6+ Kc7 Ra7+ Kb6 Ra2 Rxh3
11 -0.70 10791 0:00.01 Kf6 Kd2 Kf5 Kd3 Rh8 d5 Ra8 Kd4 Ra4+ Kc5 Ra5+ Kc6 Ra6+ Kc7 Ra7+ Kb6 Ra2
11 -0.70 9999 0:00.01 Kf6 Kd2 Kf5 Kd3 Rh8 d5 Ra8 Kd4 Ra4+ Kc5 Ra5+ Kc6 Ra6+ Kc7 Ra7+ Kb6 Ra2
10 -0.70 6439 0:00.01 Kf6 Kd2 Kf5 Kd3 Rh8 d5 Ra8 Kd4 Ra4+ Kc5 Ra5+ Kc6 Ra6+ Kc7 Ra7+ Kb6
10 -0.70 5318 0:00.01 Kf6 Kd2 Kf5 Kd3 Rh8 d5 Ra8 Kd4 Ra4+ Kc5 Ra5+ Kc6 Ra6+ Kc7 Ra7+ Kb6
9 -0.70 3302 0:00.00 Kf6 Kd2 Kf5 Kd3 Rh8 Rf3+ Ke6 Ke4 Rh4+ Ke3 Rh7
9 -0.70 2806 0:00.00 Kf6 Kd2 Kf5 Kd3 Rh8 Rf3+ Ke6 Ke4 Rh4+ Ke3 Rh7
8 -0.72 1884 0:00.00 Kf6 Kd2 Kf5 Kd3 Rh8 Rf3+ Ke6 Ke4 Rh5
8 -0.72 1646 0:00.00 Kf6 Kd2 Kf5 Kd3 Rh8 Rf3+ Ke6 Ke4 Rh5
7 -0.72 1270 0:00.00 Kf6 Kd2 Kf5 Kd3 Rh8 Rf3+ Ke6 Ke4 Rh5
7 -0.72 1136 0:00.00 Kf6 Kd2 Kf5 Kd3 Rh8 Rf3+ Ke6 Ke4 Rh5
6 -0.70 826 0:00.00 Kf6 Kd2 Kf5 Kd3 Rh8 Rf3+ Ke6
6 -0.70 729 0:00.00 Kf6 Kd2 Kf5 Kd3 Rh8 Rf3+ Ke6
5 -0.70 523 0:00.00 Kf6 Kd2 Kf5 Kd3 Rh8 Rf3+ Ke6
5 -0.70 460 0:00.00 Kf6 Kd2 Kf5 Kd3 Rh8 Rf3+ Ke6
4 -0.75 323 0:00.00 Kf6 Kd2 Kf5 Kd3
3 -0.75 238 0:00.00 Kf6 Kd2 Kf5 Kd3
2 -0.75 106 0:00.00 Kf6 Kd2 Kf5 Kd3
1 -0.82 33 0:00.00 Kf6 Rf3+

[lech]

I am a programmer and chess composer and I can produce such items or a lot more uncomfortable.
This is not my goal, though,
I'm working on Sting SF 2.0, which solves many problems (not all).
It is good that people have the illusion that they can play better than good engines.

[pichy]

Yes, this is a better position for white since the formation of the black pawns are on white squares on the Queen side it makes it much harder for black to defend both sides a better position for black but still with a small advantage for white would hac been this
one, and a top GM either Aonian or Kramnik can defend well both sides against any of the top 3 programs:
[D]2bk4/p1p3p1/3p1pBp/P7/1P1P4/5PP1/7P/4K3 w - - 2 1

I have an update. On a deeper level believe it or not Houdini was not able to win this ending in one of the games - essentially it just refused to invade with it's king. For whatever reason Houdini is stuck on an evaluation of "32" which might be a draw score perhaps and cannot make progress. I don't know why but it's probably related to overly aggressive null move pruning or some search heuristic or possibly zugzwang because it reports several ply deeper but is not able to see anything.

But now withstanding the Houdini glitch I think I disagree with you in general about this. I don't believe any top player could avoid losing against Komodo from this position if you give Komodo white - I think this is a clear win for white.

I'm still curious about the Houdini glitch, there is still a very remote possibility that Houdini see's some kind of defense for black if he invades, so I'm going to run this one more time - giving Houdini black and a much longer time control than previously and I will look to see if there is some idea I missed.

I played this out once more letting Houdini defend at a very high level, taking basically all day to complete the game. Komodo played at a time control several times faster and still won using the same simple plan, so I am convinced that this is simply a win for white and there is no hidden defense. At least as far as my ability takes me (with help from Komodo and Houdini.)

Don

You are correct the position is a win for White since most of white pawns are further advanced than Black Pawns. I modified this position one last time, which should make the position dead draw, and depending on who moves first either White or Black should have a slightly advantage

PS: Now lets see who play the ending better Houdini or Komodo without tablebase?

[D]2bk4/8/p4pBp/Pp4p1/1PpP4/5PP1/7P/4K3 w - - 0 1

This is the analysis by Houdini 1.5a :

FEN: 2bk4/8/p4pBp/Pp4p1/1PpP4/5PP1/7P/4K3 w - - 0 1

Houdini_15_w32:
4/7 00:00 337 0 +0.02 Ke1d2 Kd8e7 Kd2c3 Ke7d6 Bg6e4
5/8 00:00 534 0 +0.02 Ke1d2 Kd8e7 Kd2c3 Ke7d6 Bg6e4
6/12 00:00 823 0 +0.06 Ke1d2 Kd8e7 Kd2c3 Ke7d6 Bg6e4 f6f5
7/12 00:00 1.448 0 -0.03 Ke1d2 Kd8e7 Kd2c3 Ke7d6 Bg6b1 f6f5 f3f4
8/12 00:00 2.088 0 0.00 Ke1d2 Kd8e7 Kd2c3 Ke7d6 g3g4 Bc8b7 Bg6e4 Bb7d5
9/15 00:00 4.710 314.000 -0.06 Ke1d2 Kd8e7 Kd2c3 Ke7d6 g3g4 Bc8b7 Bg6e4 Bb7d5 h2h3
10/15+ 00:00 7.105 61.000 -0.01 Ke1d2
10/15 00:00 8.338 69.000 0.00 Ke1d2 Kd8e7 Kd2c3 Ke7d6 g3g4 Bc8b7 Bg6e4 Bb7c6 h2h3 Bc6d5
11/17- 00:00 12.681 99.000 -0.05 Ke1d2 Kd8e7
11/24 00:00 27.699 199.000 -0.08 Ke1d2 Kd8e7 h2h4 g5xh4 g3xh4 Ke7d6 Kd2c3 f6f5 h4h5 Kd6d5 Bg6f7+ Bc8e6 Bf7xe6+ Kd5xe6
12/24+ 00:00 45.140 298.000 -0.03 Ke1d2
12/24 00:00 67.589 390.000 -0.04 Ke1d2 f6f5 Kd2c3 Kd8e7 h2h3 Ke7f6 Bg6h5 f5f4 g3xf4 Bc8xh3 f4xg5+ Kf6xg5 Bh5f7 Kg5f4 Bf7h5
13/24- 00:00 75.898 419.000 -0.08 Ke1d2 f6f5
13/24 00:00 117.051 576.000 -0.01 g3g4 Kd8c7 Ke1d2 Bc8d7 Bg6e4 Bd7c6 Kd2c3 Kc7d6 Kc3c2 Bc6d5 Kc2c3 Kd6c6 Kc3c2 Kc6d6 Kc2c3
14/24 00:00 143.282 654.000 0.00 g3g4 Kd8c7 Ke1d2 Bc8d7 Bg6e4 Bd7c6 Kd2c3 Kc7d6 Kc3c2 Bc6d5 Kc2c3 Bd5c6 Kc3c2
15/24 00:00 180.817 769.000 0.00 g3g4 Kd8c7 Ke1d2 Bc8d7 Bg6e4 Bd7c6 Kd2c3 Kc7d6 Kc3c2 Bc6d5 Kc2c3 Kd6c6 Kc3c2 Kc6d6 Kc2c3
16/25 00:00 245.433 933.000 0.00 g3g4 Kd8c7 Ke1d2 Bc8d7 Bg6e4 Bd7c6 Kd2c3 Kc7d6 Kc3c2 Bc6d5 Kc2c3 Kd6c6 Kc3c2 Kc6d6 Kc2c3
17/26 00:00 327.235 1.105.000 0.00 g3g4 Kd8c7 Ke1d2 Bc8d7 Bg6e4 Bd7c6 Kd2c3 Kc7d6 Kc3c2 Bc6d5 Kc2c3 Kd6c6 Kc3c2 Kc6d6 Kc2c3
18/30 00:00 471.444 1.335.000 0.00 g3g4 Kd8c7 Ke1d2 Bc8d7 Bg6e4 Bd7c6 Kd2c3 Kc7d6 Kc3c2 Bc6d5 Kc2c3 Kd6c6 Kc3c2 Kc6d6 Kc2c3
19/30 00:00 628.867 1.526.000 0.00 g3g4 Kd8c7 Ke1d2 Bc8d7 Bg6e4 Bd7c6 Kd2c3 Kc7d6 Kc3c2 Bc6d5 Kc2c3 Kd6c6 Kc3c2 Kc6d6 Kc2c3
20/30 00:00 948.034 1.819.000 0.00 g3g4 Kd8c7 Ke1d2 Bc8d7 Bg6e4 Bd7c6 Kd2c3 Kc7d6 Kc3c2 Bc6d5 Kc2c3 Kd6c6 Kc3c2 Kc6d6 Kc2c3
21/34 00:00 1.325.314 1.992.000 0.00 g3g4 Kd8c7 Ke1d2 Bc8d7 Bg6e4 Bd7c6 Kd2c3 Kc7d6 Kc3c2 Bc6d5 Kc2c3 Kd6c6 Kc3c2 Kc6d6 Kc2c3
22/35 00:00 1.894.664 2.182.000 0.00 g3g4 Kd8c7 Ke1d2 Bc8d7 Bg6e4 Bd7c6 Kd2c3 Kc7d6 Kc3c2 Bc6d5 Kc2c3 Kd6c6 Kc3c2 Kc6d6 Kc2c3
23/35 00:01 2.821.999 2.332.000 0.00 g3g4 Kd8c7 Ke1d2 Bc8d7 Bg6e4 Bd7c6 Kd2c3 Kc7d6 Kc3c2 Bc6d5 Kc2c3 Kd6c6 Kc3c2 Kc6d6 Kc2c3
24/38 00:01 4.178.661 2.476.000 0.00 g3g4 Kd8c7 Ke1d2 Bc8d7 Bg6e4 Bd7c6 Kd2c3 Kc7d6 Kc3c2 Bc6d5 Kc2c3 Kd6c6 Kc3c2 Kc6d6 Kc2c3
25/39 00:02 6.679.205 2.615.000 0.00 g3g4 Kd8c7 Ke1d2 Bc8d7 Bg6e4 Bd7c6 Kd2c3 Kc7d6 Kc3c2 Bc6d5 Kc2c3 Kd6c6 Kc3c2 Kc6d6 Kc2c3
26/40 00:04 11.722.390 2.740.000 0.00 g3g4 Kd8c7 Ke1d2 Bc8d7 Bg6e4 Bd7c6 Kd2c3 Kc7d6 Kc3c2 Bc6d5 Kc2c3 Kd6c6 Kc3c2 Kc6d6 Kc2c3
27/40 00:06 16.511.379 2.778.000 0.00 g3g4 Kd8c7 Ke1d2 Bc8d7 Bg6e4 Bd7c6 Kd2c3 Kc7d6 Kc3c2 Bc6d5 Kc2c3 Kd6c6 Kc3c2 Kc6d6 Kc2c3
28/49 00:08 23.646.088 2.801.000 0.00 g3g4 Kd8c7 Ke1d2 Bc8d7 Bg6e4 Bd7c6 Kd2c3 Kc7d6 Kc3c2 Bc6d5 Kc2c3 Kd6c6 Kc3c2 Kc6d6 Kc2c3
29/49 00:12 34.106.617 2.815.000 0.00 g3g4 Kd8c7 Ke1d2 Bc8d7 Bg6e4 Bd7c6 Kd2c3 Kc7d6 Kc3c2 Bc6d5 Kc2c3 Kd6c6 Kc3c2 Kc6d6 Kc2c3
30/49 00:19 54.627.522 2.806.000 0.00 g3g4 Kd8c7 Ke1d2 Bc8d7 Bg6e4 Bd7c6 Kd2c3 Kc7d6 Kc3c2 Bc6d5 Kc2c3 Kd6c6 Kc3c2 Kc6d6 Kc2c3
31/50 00:30 87.195.676 2.837.000 0.00 g3g4 Kd8c7 Ke1d2 Bc8d7 Bg6e4 Bd7c6 Kd2c3 Kc7d6 Kc3c2 Bc6d5 Kc2c3 Kd6c6 Kc3c2 Kc6d6 Kc2c3
32/50 00:48 135.795.208 2.791.000 0.00 g3g4 Kd8c7 Ke1d2 Bc8d7 Bg6e4 Bd7c6 Kd2c3 Kc7d6 Kc3c2 Bc6d5 Kc2c3 Kd6c6 Kc3c2 Kc6d6 Kc2c3
33/53 01:58 324.693.472 2.745.000 0.00 g3g4 Kd8c7 Ke1d2 Bc8d7 Bg6e4 Bd7c6 Kd2c3 Kc7d6 Kc3c2 Bc6d5 Kc2c3 Kd6c6 Kc3c2 Kc6d6 Kc2c3
34/54 02:34 434.209.905 2.803.000 0.00 g3g4 Kd8c7 Ke1d2 Bc8d7 Bg6e4 Bd7c6 Kd2c3 Kc7d6 Kc3c2 Bc6d5 Kc2c3 Kd6c6 Kc3c2 Kc6d6 Kc2c3
35/60 04:54 832.439.631 2.831.000 0.00 g3g4 Kd8c7 Ke1d2 Bc8d7 Bg6e4 Bd7c6 Kd2c3 Kc7d6 Kc3c2 Bc6d5 Kc2c3 Kd6c6 Kc3c2 Kc6d6 Kc2c3
36/60 08:22 1.427.804.581 2.842.000 0.00 g3g4 Kd8c7 Ke1d2 Bc8d7 Bg6e4 Bd7c6 Kd2c3 Kc7d6 Kc3c2 Bc6d5 Kc2c3 Kd6c6 Kc3c2 Kc6d6 Kc2c3
37/62 12:47 2.216.794,444 2.889.000 0.00 g3g4 Kd8c7 Ke1d2 Bc8d7 Bg6e4 Bd7c6 Kd2c3 Kc7d6 Kc3c2 Bc6d5 Kc2c3 Kd6c6 Kc3c2 Kc6d6 Kc2c3