Double and Triple pawns are no liability

[OliverBr]

Just have a look at this position between Fruit 2.1 (white) and OliThink 5.11.9 (black) at move 50:

[d]3r4/8/1p1r1pk1/1p3p1p/1p1PbP1P/1P4P1/PN1R4/K2R4 b - - 1 50

We have equal materials with 6 pawns on each site.
White has a so called "healthy pawn structure" and Black has a double pawn AND a triple pawn.

Now comes the surprise for many: White is totally lost (Stockfish says +4 for black) being the decisive move that one that created the triple pawn for black.
How is this possible?

Because this "healthy pawn structure" doesn't matter.
Of course all strong engines do not care anymore and many expert do no use this phrase anymore or less frequent, but still it is heard often.

The first engine I remember doing this was Rybka and it surprised many. It created double pawns like a factory and won.

Look what happened in this very game about 24 moves later:

[d]8/8/1p3p2/1p3p2/1p2b2p/1k6/5N2/5K2 w - - 0 7

You see it right: White has lost all (!) its pawns, while black still has all its pawns, including the doubles and triples.
Black actually ate all the white pawns pac-man style after breaking through the terribly weak white d-pawn.

FYI: The complete game against the fantastic "old school" engine Fruit:

[pgn][Event "Computer Chess Game"]
[Site "Olivers-MacBook-3.local"]
[Date "2025.10.05"]
[Round "-"]
[White "Fruit 2.1"]
[Black "OliThink 5.11.9"]
[Result "0-1"]
[TimeControl "40/120"]
[Annotator "1. +0.22 1... -0.15"]

1. Nf3 {+0.22/13} d5 {-0.15/22 2.1} 2. d4 {+0.15/13 4} Nf6 {-0.10/20 2.3}
3. c4 {+0.21/12 3} e6 {-0.21/19 2.2} 4. Nc3 {+0.15/11 4} a6 {-0.13/20 2.9}
5. cxd5 {+0.52/11 4} exd5 {-0.31/23 2.3} 6. Bg5 {+0.51/12 3} h6
{-0.35/23 2.5} 7. Bxf6 {+0.98/12 2.7} gxf6 {-0.27/22 2.0} 8. Qb3
{+0.92/13 3} c6 {-0.33/21 2.0} 9. e4 {+0.95/11 2.6} dxe4 {-0.26/20 2.4} 10.
Nxe4 {+1.08/10 2.2} f5 {-0.48/20 2.8} 11. Nc3 {+0.93/11 4} Qf6
{-0.50/20 2.7} 12. O-O-O {+1.36/12 2.7} Bd6 {-0.39/19 2.3} 13. Re1+
{+1.32/10 1.8} Be7 {-0.59/20 4} 14. Kb1 {+1.49/10 1.9} O-O {-0.60/20 4} 15.
Bd3 {+1.34/10 2.7} a5 {-0.61/20 2.4} 16. Ne5 {+1.53/10 2.8} Na6
{-0.61/19 2.1} 17. Bc4 {+1.38/9 5} Bd6 {-0.35/19 2.1} 18. Bxa6 {+0.99/9 4}
Rxa6 {-0.19/22 2.1} 19. Na4 {+0.89/10 2.5} Bb4 {+0.25/19 3} 20. Re2
{+0.72/10 1.7} Be6 {+0.45/21 3} 21. Qg3+ {+0.77/10 2.2} Kh7 {+0.35/21 2.2}
22. f4 {+0.45/10 2.2} Rg8 {+0.38/21 2.7} 23. Qf2 {+0.52/11 2.8} Qd8
{+0.78/20 3} 24. Ka1 {+0.41/11 4} Bd5 {+0.63/20 2.2} 25. Rg1 {+0.31/12 3}
f6 {+0.69/21 2.5} 26. Nd3 {+0.35/11 2.4} b6 {+1.01/21 2.1} 27. Nc3
{+0.23/11 4} Bc4 {+0.91/21 11} 28. Rd1 {+0.20/11 1.8} Ra8 {+0.80/18 1.9}
29. g3 {+0.29/10 3} Bf7 {+0.84/19 3} 30. Qf3 {+0.29/10 2.0} Qd6
{+0.87/18 2.1} 31. Na4 {+0.45/10 2.2} Rgb8 {+0.77/19 2.0} 32. h3
{+0.40/10 2.0} Ra7 {+0.78/18 2.1} 33. Re3 {+0.33/10 3} Rd7 {+0.71/20 2.3}
34. b3 {+0.22/11 2.5} Bd5 {+0.85/22 5} 35. Qe2 {+0.26/11 2.8} Rg7
{+0.67/22 10} 36. h4 {+0.33/11 2.3} Ra7 {+0.94/21 4} 37. Rc1 {+0.20/11 5}
Rc7 {+0.95/19 1.6} 38. Nxb4 {+0.22/11 4} axb4 {+1.14/22 2.9} 39. Qf2
{+0.17/12 2.8} Be4 {+1.25/20 2.2} 40. Re2 {+0.12/12 2.0} Ra7 {+1.28/18 2.6}
41. Rd2 {+0.12/12 4} Qd5 {+1.32/20 1.8} 42. Qe3 {+0.07/12 4} h5
{+1.36/24 1.9} 43. Qe2 {+0.14/12 4} Kg6 {+1.32/24 1.8} 44. Qc4 {+0.15/12 3}
Qa5 {+1.45/22 2.0} 45. Qf1 {+0.14/12 2.2} Rd8 {+1.49/24 1.9} 46. Qf2
{+0.13/12 2.7} Rad7 {+1.46/26 2.6} 47. Rcd1 {+0.05/12 2.9} Rd6
{+1.46/27 2.3} 48. Qe2 {-0.12/12 3} Qb5 {+1.66/23 2.3} 49. Qxb5
{-0.37/14 2.3} cxb5 {+2.45/30 3} 50. Nb2 {-0.53/15 1.7} Rd5 {+2.30/31 2.0}
51. Rc1 {-0.26/15 2.4} R8d6 {+2.34/31 3} 52. Rdd1 {-0.34/16 2.3} Kf7
{+2.31/30 2.0} 53. Rc7+ {-0.38/16 3} Ke6 {+2.46/25 2.3} 54. Rcc1
{-0.34/16 2.0} Rxd4 {+3.19/28 2.0} 55. Rxd4 {-1.22/16 4} Rxd4
{+3.56/29 2.6} 56. Rd1 {-1.44/18 4} Kd5 {+4.10/28 2.6} 57. Re1 {-1.68/19 4}
Rd2 {+4.66/31 5} 58. Nd1 {-1.91/18 4} Kd4 {+4.68/30 2.1} 59. Ne3
{-2.33/18 4} Rf2 {+4.66/28 2.0} 60. Nf1 {-2.76/17 3} Rg2 {+5.15/28 2.2} 61.
Ne3 {-3.08/17 3} Rxg3 {+5.25/28 2.2} 62. Nf1 {-3.22/17 1.9} Rf3
{+5.31/28 5} 63. Rd1+ {-3.38/16 2.7} Bd3 {+5.95/24 2.4} 64. Nd2
{-3.64/18 2.2} Rxf4 {+6.26/26 2.5} 65. Kb2 {-3.81/17 2.4} Ke3 {+6.50/27 3}
66. Kc1 {-3.87/17 2.8} Rxh4 {+6.25/25 2.6} 67. Nf1+ {-4.37/16 2.8} Kd4
{+7.74/28 3} 68. Ng3 {-4.83/15 2.9} Kc3 {+9.80/29 8} 69. Rh1 {-5.10/17 1.9}
Rxh1+ {+17.39/28 2.1} 70. Nxh1 {-6.45/22 2.8} h4 {+18.83/32 2.9} 71. Nf2
{-6.87/21 4} Be4 {+18.78/30 3} 72. Kd1 {-9.56/21 13} Kb2 {+24.97/30 4} 73.
Ke2 {-14.09/19 2.0} Kxa2 {+24.04/28 2.2} 74. Kf1 {-18.60/22 6} Kxb3
{+32.24/28 5} 75. Ke2 {-21.47/18 1.0} Kc2 {+1000.15/25 2.6} 76. Ke1
{-21.40/18 1.0} b3 {+1000.11/26 2.1} 77. Nd1 {-21.40/16 1.0} h3
{+1000.10/26 2.4} 78. Ne3+ {-21.59/13 1.0} Kc1 {+1000.09/26 2.5} 79. Nf1
{-1000.08/11 0.8} b2 {+1000.08/30 2.9} 80. Nd2 {-1000.07/15 0.8} h2
{+1000.07/42 4} 81. Nb3+ {-1000.06/52 2.3} Kc2 {+1000.06/64 1.6} 82. Ke2
{-1000.05/53 3} Kxb3 {+1000.05/64 0.2} 83. Ke3 {-1000.04/51 3} b1=Q
{+1000.04/64 0.1} 84. Kf2 {-1000.03/48 2.3} f4 {+1000.03/64 0.1} 85. Ke2
{-1000.02/4 0.1} Qb2+ {+1000.02/64 0.1} 86. Kd1 {-1000.01/44 2.2} h1=Q#
{+1000.01/64 0.1}
{Xboard adjudication: Checkmate} 0-1
[/pgn]

[Ajedrecista]

Hello Oliver:

Thank you for sharing this game! I guess it is one of the rare cases of 'the exception that proves the rule'.

Once 51.- Rc1 is played, 51.- ..., Rxd4?? variant only draws after the trade of rooks. However, after 54.- Rcc1, the played move 54.- ..., Rxd4 is correct and wins after trading rooks. This must be because the black king is on g6 in the variant, too far of the action of the centre of the board and the white king is on time to help and defend; while in the game, the black king is on e6, much better placed.

Regards from Spain.

Ajedrecista.

[OliverBr]

[OliverBr]

Hello Oliver:

Hola Ajedrecista

Thank you for sharing this game! I guess it is one of the rare cases of 'the exception that proves the rule'.

Thank you for your opinion, but I do not thinl it is a rare case. I do often see double pawns in high level engines like Stockfish and still winning. I think that the "healthy pawn structure" was/is still overrated.

Once 51.- Rc1 is played, 51.- ..., Rxd4?? variant only draws after the trade of rooks. However, after 54.- Rcc1, the played move 54.- ..., Rxd4 is correct and wins after trading rooks. This must be because the black king is on g6 in the variant, too far of the action of the centre of the board and the white king is on time to help and defend; while in the game, the black king is on e6, much better placed.

Yes, there are a couple of reasons why white is lost. One is actually Zugzwang.
Regards from Shrimp-Country or was that England?

[BeyondCritics]

This funny position showcases, that piece activity is the most important factor and can compensate long term for any positional defect, even deep in the endgame. Similar positions were discussed in Euwes middle game encyclopedia and Suetin's strategy books, demonstrating the same tenet.

[abulmo2]

Repositioning all the white pawns on the white squares transforms the black win into a draw (according to stockfish):
[d]3r4/8/1p1r1pk1/1p3p1p/1p2b3/1P1P1P1P/PN1R2P1/K2R4 b - - 1 1
In the original position, more than the black pawn structure, I think this is the white position that is uncomfortable, with a lack of control of the white squares dominated by the black bishop, leading to some sorts of zugzwang.
I often heard some strong chess French players like Etienne Bacrot (2638 FIDE Elo and a former top ten player) explaining their games. They do not always avoid double or even triple pawns, but are more afraid by the holes created by a bad pawn structure. In your example, the black pawns control many squares, particularly the A, C, E & G columns; whereas the white pawns have some holes that leave the control of the white squares to the black bishop.

[BeyondCritics]

I do like the idea of moving around pieces, to get some information, but unfortunately it is not really supported within the "everything is a game" paradigm of current IDE's. Moving the pawns shows, that white has white square weaknesses.
Move the king to E3: Equal. Move the knight to E3: White is clearly better. Knight and king are hopelessly misplaced, fixing this would save white.

[Ajedrecista]

Hello Oliver:

Grabbing pawns Pac-Man style still leads to a lose in this joke problem by myself:

[pgn][Event ""]
[Site "Madrid (Spain)"]
[Round ""]
[Date "2025.10.07"]
[White ""]
[Black ""]
[SetUp "1"]
[FEN "8/1p6/bP6/P7/P7/P7/P1K5/kR6 b - - 0 1"]
[Result "1-0"]

1. ... Kxa2 2. Rb5 Kxa3 (2. ... Bxb5 3. axb5 Kxa3 4. a6 bxa6 5. bxa6 {(white wins)}) 3. Kc3 Kxa4 4. Rb1 Kxa5 5. Kd4 {(only winning move)} 5. ... Ka4 6. Kc5 Be2 7. Ra1+ Kb3 8. Ra7 Bf3 9. Kd6 Kc4 10. Kc7 {(white wins)} 1-0[/pgn]

I am not saying this is the optimal solution. Not all variants are in the PGN. Please do not ask about the legality and/or the feasibility of the position...

Regards from Spain.

Ajedrecista.

[syzygy]

I am not saying this is the optimal solution. Not all variants are in the PGN. Please do not ask about the legality and/or the feasibility of the position...

Looks perfectly legal to me

[Ajedrecista]

Hello Ronald:

I am not saying this is the optimal solution. Not all variants are in the PGN. Please do not ask about the legality and/or the feasibility of the position...

Looks perfectly legal to me

Sure. Driving most of the white pawns to a-file in Kuwait Immortal fashion; driving the black king to a1 square while the white rook is elsewhere, allowing it; and black bishop doing waiting moves until reaching the desired position. I guess this is a pseudo proof game, like pseudo code relates to code.

The former article gives a hint about a known problem with pawn grabbing being compulsory and losing (YACPDB #99163, PDB P1267725), which is cooked; and I found other related problem (KRPPPPPP vs. k) that I was not aware before (YACPDB #312367, PDB P1424551).

All being said, I do not want to hijack Oliver's thread.

Regards from Spain.

Ajedrecista.