What is the outcome of this unusual endgame

[hgm]

I am not sure how drawish KNNPKB is. It is true you have no mating potential with the two Knights, so that superficially it isn't any better than KNPKB. But the two Knights give you a lot more possibility to screen the Pawn from the Bishop, when it has to pass through a square of the Bishop's shade. So it is a lot harder to make the drawing B for P sac. And with KNNPPKBB the defender has the problem that a B-for-N trade will be fatal, and he has two Bishops that could be targets for Knight tactics, forcing the Bishops to stay at safe distance from their King.

[mwyoung]

I have not looked at this position with an engine. But I believe that at a glance, most strong players would evaluate this as winning for black. Black is up two pawns, and all four pawns are connected. It seems obviously winning.

I also thought this was winning for black as a player. But some here said this was a draw with strong engines and hardware.
That should not matter, unless the human has the strong side. Then I think the engine could hold a draw against most humans.
As the win is not easy, or straight forward.

[yurikvelo]

D=81, 167 555 MN, 2232 MTBHits
-3,45 18.Be1 Kg5 19. Bc4 Nf5 20. Ba6 Kh5 21. Ba5 Kh4 22. Bd8 Nd5

further deepening:
[pgn][SetUp "1"]
[FEN "8/5pk1/4pnp1/3n3p/8/3B4/3B1PP1/6K1 w - - 0 1"]


1. Bc1 Kf8 2. Ba3+ Ke8 3. Kh2 Kd7 4. Kg3 e5 5. Bc1 Ke7 6. f3 Nd7 7. Bb5 f6
8. Kh3 Nf4+ 9. Kh2 h4 10. Ba3+ Ke6 11. Bb4 Nb6 12. Ba6 Kf5 13. Bd2 Kg5 14.
g3 hxg3+ 15. Kxg3 Nd5 16. Bb5 Kh5 17. Kf2 Ne7 18. Be1 Kg5 19. Bc4 Nf5 20.
Ba6 Kh5 21. Ba5 Kh4 22. Bd8 Nd5 23. Bc4 Nfe7 24. Bb5 g5 25. Bd7 Ng6 26. Kg2
Ngf4+ 27. Kh2 Nd3 28. Ba5 N3b4 29. Be6 Nf4 30. Bf5 Nc6 31. Be1+ Kh5 32. Kg3
Nd4 33. Bd7 f5 34. Be8+ Kh6 35. Ba5 Nd5 36. Bf7 f4+ 37. Kf2 Ne3 38. Bc7 g4
39. fxg4 Nxg4+ 40. Kg1 Kg5 41. Bd5 Kf5 42. Bb7 Nf6 43. Bg2 Ke6 44. Bh3+ Nf5
45. Bb8 Nd5 46. Bg2 Nb4 47. Kf2 Nd4 48. Bh3+ Ke7 49. Bg2 Ndc6 50. Bc7 Ke6
51. Bb6 Nd4 52. Be4 Nd5 53. Ba7 Nb5 54. Bb8 Nb4 55. Bb7 Nd4 56. Bc7 Kd7 57.
Ba5 Ndc6 58. Bb6 e4 59. Ba8 Kd6 60. Kg2 Nd5 61. Bg1 Ne5 62. Bd4 e3 63. Kf1
Ke6 64. Bb7 Nf3 65. Ba7 Nd2+ 66. Ke1 Nc3 67. Bb6 Kf5 68. Bd4 Nb5 69. Bb6
Nc4 70. Bc5 Ne5 71. Kf1 Nc3 72. Ba6 Ke4 73. Ba7 Nd5 74. Bb7 Nf3 75. Bb6
Nd2+ 76. Ke1 Nc4 77. Bc7 f3 78. Bg3 Nb2 79. Kf1 Nd3 80. Kg1 e2 81. Bf2
e1=Q+ 82. Bxe1 Nxe1 83. Ba6 Nd3 84. Bb7 Kd4 85. Bxd5 Kxd5 86. Kf1 Ke4 87.
Kg1 Ke3 88. Kf1 f2 89. Kg2 Ke2 90. Kh3 f1=Q+ 91. Kg4 Qf4+ 92. Kh5 Ke3 93.
Kg6 Ne5+ 94. Kh7 Qf7+ 95. Kh8 Ng6# *
[/pgn]

for better analysis KNNPvKBB should be obtained (15.8 Gb WDL and 3.2 Gb DTZ)
KNNPvKBB contain very many win positions which obey 50-move rule. So there may be a chance if 50-move rule apply, it is possible to find cursed draw

[Nordlandia]

Against a pair of knights the bishops will almost certainly dominate - so much so that the rule of thumb 'B+B = N+N+P' has more than a ring of truth to it.

It may be winning for black.

[mwyoung]

Against a pair of knights the bishops will almost certainly dominate - so much so that the rule of thumb 'B+B = N+N+P' has more than a ring of truth to it.

It may be winning for black.

You clearly do not know chess theory. Knights are not weaker when the pawns on the same side of the board.

[d]8/5pk1/4pnp1/3n3p/8/3B4/3B1PP1/6K1 w - - 0 1

Look at the position. All the action is on 1/2 the board. All of blacks pawns are connected with a 2 pawn advantage.
What did you think was going to happen. All you know is bishops are better then knights. And a pair of bishops seals the deal.

[mwyoung]

Shoot out:
----- Time = Rapid 30min -----
Stockfish 151120 0-1
Dragon by Komodo Chess 64-bit 0-1
Lc0 v0.26.3 0-1

All shootouts in favor of black. White wins 0%, Draws 0%

[ernest]

D=81, 167 555 MN, 2232 MTBHits
-3,45 18.Be1 Kg5 19. Bc4 Nf5 20. Ba6 Kh5 21. Ba5 Kh4 22. Bd8 Nd5

further deepening:
[pgn][SetUp "1"]
[FEN "8/5pk1/4pnp1/3n3p/8/3B4/3B1PP1/6K1 w - - 0 1"]


1. Bc1 Kf8 2. Ba3+ Ke8 3. Kh2 Kd7 4. Kg3 e5 5. Bc1 Ke7 6. f3 Nd7 7. Bb5 f6
8. Kh3 Nf4+ 9. Kh2 h4 10. Ba3+ Ke6 11. Bb4 Nb6 12. Ba6 Kf5 13. Bd2 Kg5 14.
g3 hxg3+ 15. Kxg3 Nd5 16. Bb5 Kh5 17. Kf2 Ne7 18. Be1 Kg5 19. Bc4 Nf5 20.
Ba6 Kh5 21. Ba5 Kh4 22. Bd8 Nd5 23. Bc4 Nfe7 24. Bb5 g5 25. Bd7 Ng6 26. Kg2
Ngf4+ 27. Kh2 Nd3 28. Ba5 N3b4 29. Be6 Nf4 30. Bf5 Nc6 31. Be1+ Kh5 32. Kg3
Nd4 33. Bd7 f5 34. Be8+ Kh6 35. Ba5 Nd5 36. Bf7 f4+ 37. Kf2 Ne3 38. Bc7 g4
39. fxg4 Nxg4+ 40. Kg1 Kg5 41. Bd5 Kf5 42. Bb7 Nf6 43. Bg2 Ke6 44. Bh3+ Nf5
45. Bb8 Nd5 46. Bg2 Nb4 47. Kf2 Nd4 48. Bh3+ Ke7 49. Bg2 Ndc6 50. Bc7 Ke6
51. Bb6 Nd4 52. Be4 Nd5 53. Ba7 Nb5 54. Bb8 Nb4 55. Bb7 Nd4 56. Bc7 Kd7 57.
Ba5 Ndc6 58. Bb6 e4 59. Ba8 Kd6 60. Kg2 Nd5 61. Bg1 Ne5 62. Bd4 e3 63. Kf1
Ke6 64. Bb7 Nf3 65. Ba7 Nd2+ 66. Ke1 Nc3 67. Bb6 Kf5 68. Bd4 Nb5 69. Bb6
Nc4 70. Bc5 Ne5 71. Kf1 Nc3 72. Ba6 Ke4 73. Ba7 Nd5 74. Bb7 Nf3 75. Bb6
Nd2+ 76. Ke1 Nc4 77. Bc7 f3 78. Bg3 Nb2 79. Kf1 Nd3 80. Kg1 e2 81. Bf2
e1=Q+ 82. Bxe1 Nxe1 83. Ba6 Nd3 84. Bb7 Kd4 85. Bxd5 Kxd5 86. Kf1 Ke4 87.
Kg1 Ke3 88. Kf1 f2 89. Kg2 Ke2 90. Kh3 f1=Q+ 91. Kg4 Qf4+ 92. Kh5 Ke3 93.
Kg6 Ne5+ 94. Kh7 Qf7+ 95. Kh8 Ng6# *
[/pgn]

for better analysis KNNPvKBB should be obtained (15.8 Gb WDL and 3.2 Gb DTZ)
KNNPvKBB contain very many win positions which obey 50-move rule. So there may be a chance if 50-move rule apply, it is possible to find cursed draw

Interesting to note that after 39. ...Nxg4+ the only 2 Pawns that remain are the 2 Black Pawns,
but while the sacrifice of the first White Bishop against Black Pawn is easy, the second sacrifice Bishop-Pawn (ensuring the Draw) can be prevented ! (at least so it seems...)

[hgm]

That was also my intuition: with two Knights it must be possible to 'lock out' the Bishop, and create a safe path for the Pawn to advance. While the Pawn is on the other shade as the Bishop it is absolutely safe, and you can prepare the next advance at leisure. Also, you don't have to fear B-for-N trades in KNNPKB, while in KNPKB such trades easily leave you with a drawn KPK. So my engines usually do not discount end-games with 2 Pawns, even when they are supported only by 2 Knights with 2 defending minors to sacrifice for the Pawns.

[Karlo Bala]

Against a pair of knights the bishops will almost certainly dominate - so much so that the rule of thumb 'B+B = N+N+P' has more than a ring of truth to it.

It may be winning for black.

You clearly do not know chess theory. Knights are not weaker when the pawns on the same side of the board.

A very bold statement.

If that's true then we wouldn't even debate if NNPPPP vs BBPP is a win or a draw. Well, it is most probably a win but that only suggests how BB duo is strong regardless of pawns. BBPP vs NNPPP is a draw, while BBPPP vs NNPP is an easy win even if all pawns are on the same side.

[Nordlandia]

Is adjusting the pawns to the first rank helping white.

[d]5k2/4pppp/3n2n1/8/8/1B6/1B3PP1/5K2 w - - 0 1

Cfish dev -1.5ish

----

4x8 chess

I suspect this is winning for black