It seems that chess is still not a draw between top engines

[Uri Blass]

This game from mark young's match when Dragon at time control 120 minutes/40 moves beat Stockfish at fisher time control 3+2
It does not seem that the opening position is unbalanced and it suggest that it is possible to beat stockfish from the opening position when stockfish play without book.
source is this thread that has also other games when stockfish lost.
https://prodeo.actieforum.com/t896-drag ... -3m2s-live

[pgn][Event "Chess Tournament"]
[Site "*"]
[Date "2022.08.19"]
[Round "1"]
[White "Dragon 3.1 by Komodo Chess 64-bit"]
[Black "Stockfish 240722"]
[Result "1-0"]
[TimeControl "40/600+5"]
[Time "22"]
[Board "82"]
[Termination "mate"]
[FEN "rn1q1rk1/pbp1bpp1/1p2pn1p/3p4/2PP3B/2N1PN2/PP2BPPP/R2QK2R w KQ - c0 Gms: 406; c1 Score: 55.542%; c2 Draws: 54.926%; c3 Wins: 28.079%; c4 Losses: 16.995%;"]
[SetUp "1"]

1.Bxf6 Bxf6 2.cxd5 exd5 3.O-O Nd7 4.Qb3 c6
5.Rad1 g6 6.Rfe1 Bg7 7.a4 Qc7 8.Rc1 Qd8
9.a5 Rb8 10.Qa3 Re8 11.a6 Ba8 12.b4 Bf8
13.Na4 Qf6 14.Rc2 Qe6 15.Rec1 Qd6 16.Rb1 Qe6
17.Bf1 Bd6 18.Rcc1 Qe7 19.g3 Nf6 20.Nd2 Red8
21.Bg2 Rbc8 22.Qb2 Kh7 23.Rc2 Kg7 24.Qb3 Rb8
25.Qa3 Rbc8 26.Qb2 Kh7 27.Qb3 Rb8 28.Qa3 Rbc8
29.Qb2 Kg7 30.Rc3 Qe6 31.Qa2 Kg8 32.Qb3 Rb8
33.Rc2 Qe7 34.Qb2 Rbc8 35.Nf3 Rb8 36.Qa3 Rbc8
37.Qb3 Kg7 38.Qb2 Rb8 39.Rcc1 Kg8 40.Qa3 Ne4
41.Rb3 Re8 42.Qb2 Qf6 43.Ne1 Qe7 44.h3 Qe6
45.Qe2 Qe7 46.Nd3 h5 47.Nc3 h4 48.g4 Rbc8
49.Rb2 Kg7 50.Qd1 Bb8 51.Nb1 Qd6 52.Nd2 Re7
53.Nf4 Rh8 54.Nf3 f6 55.Qd3 g5 56.Ne2 Rhe8
57.Nc3 Qe6 58.Qb1 Nd6 59.Re2 Rf8 60.Nh2 Rfe8
61.Nf1 f5 62.Nh2 b5 63.gxf5 Nxf5 64.Rce1 Bxh2+
65.Kxh2 Kh8 66.Na2 g4 67.hxg4 Nh6 68.Bh3 Rg8
69.g5 Qd6+ 70.Kh1 Rxg5 71.Rg1 Rxg1+ 72.Qxg1 Rg7
73.Qd1 Ng4 74.Bxg4 Rxg4 75.Re1 Qe6 76.Qf3 Rg6
77.Kh2 Qd6+ 78.Qf4 Kg7 79.Qxd6 Rxd6 80.Nc3 Rf6
81.f4 Kf7 82.Kh3 Rh6 83.e4 Rh8 84.e5 Ke6
85.Rg1 Kf5 86.Rg4 Rh7 87.Rxh4 Rf7 88.Kg3 Rg7+
89.Kf3 Kg6 90.Rg4+ Kf7 91.Rxg7+ Kxg7 92.Ke3 Kg6
93.Ne2 Kf5 94.Ng3+ Kg4 95.f5 Kg5 96.f6 Kg6
97.Ne2 c5 98.dxc5 Bc6 99.Nd4 Bd7 100.e6 Bxe6
101.Nxe6 Kxf6 102.Nd4 Kf7 103.c6 Ke7 104.Kf4 Kd6
105.Kf5 Ke7 106.Ke5 Kd8 107.Kd6 Ke8 108.c7 Kf8
109.c8=Q+ Kg7 110.Qe6 Kh8 111.Nf5 d4 112.Qh6+ Kg8
113.Qg7# 1-0[/pgn]

I wonder if white is winning after 11...Ba8
Engines do not see the position as a big advantage for white but it seems that black bishop a8 is out of the game.

Of course even if the black bishop is out of the game it does not prove that white is winning because the position may be a type of fortress when white cannot break the fortress.

[xr_a_y]

recent TCEC testing on the subject (from starting position) : https://tcec-chess.com/#div=ch&game=1&season=23

[Ajedrecista]

Hello Uri:

[...]

It does not seem that the opening position is unbalanced [...]

The opening has a reputation of being solid: the Tartakower Variation of the QGD, also known as the Tartakower–Makogonov–Bondarevsky System (ECO D58). The game that you have posted starts with the following eight moves (transpositions may apply):

1. d4 d5 2. c4 e6 3. Nc3 Nf6 4. Bg5 Be7 5. Nf3 h6 6. Bh4 O-O 7. e3 b6 8. Be2 Bb7

According to the data provided by the PGN at ProDeo forum, there were 406 games (W: 114; D: 223; B: 69) and a white performance of 55.5% more less, pretty standard IMHO.

I only found two games at Lichess Masters database until 14. ... Bg7 (6. ... Bg7 in your PGN), so 15. a4 could be a novelty:

Fish ½—½ Baramidze. (2003).
Gerbich ½—½ Trofimov. (2020).

A fairly old source like the First Edition of the Volume D of the Encyclopedia of Chess Openings (year 1976) has many mainlines on ECO D58. The mainline listed as 11 follows the game until 11. O-O, then deviates with 11. ... Qe7, but note 64 mentions 11. ... Nd7 and a different continuation from Furman—Klovans, SSSR 1972: 12. b4 Re8 13. b5 Nc5 14. Nd2 Ne6 15. Qb3 Qd6 16. a4 → white stands slightly better.

By the way, the same source gives a slighlty different line that in your PGN in the same mainline 11, just changing 11. ... Nd7 by 11. ... Qe7 (as seen above), then 12. Qb3 [{note 65} Karasev—Klovans, SSSR 1974: 12. Rc1 Nd7 13. Qa4 a6, etc. → the game is even] c6 {note 66 with other continuations} 13. Rfe1 Bc8 14. Rac1 {note 67 with other continuations}, etc. → white stands slightly better. This is, the white rooks on c1 and e1 (like in the game of the thread), the white queen also on b3 and the black pawn also on c6. With the great difference of black bringing the bishop back to c8, which was a concern by you.

------------------------

[...] but it seems that black bishop a8 is out of the game.

Of course even if the black bishop is out of the game it does not prove that white is winning because the position may be a type of fortress when white cannot break the fortress.

A bishop on a8 does not need to be out of the game always, as in the last game of the WCC 2010, where Anand won with black against Topalov and played 28. ... Ba8 until it was captured with 47. Kxg2.

Topalov 0—1 Anand. WCC 2010, Sofia (Bulgaria).

[pgn][Event "Anand - Topalov World Championship Match"]
[Site "Sofia BUL"]
[Date "2010.05.11"]
[EventDate "2010.04.26"]
[Round "12"]
[Result "0-1"]
[White "Veselin Topalov"]
[Black "Viswanathan Anand"]
[ECO "D56"]
[WhiteElo "2805"]
[BlackElo "2787"]
[PlyCount "112"]

1.d4 d5 2.c4 e6 3.Nf3 Nf6 4.Nc3 Be7 5.Bg5 h6 6.Bh4 O-O 7.e3
Ne4 8.Bxe7 Qxe7 9.Rc1 c6 10.Be2 Nxc3 11.Rxc3 dxc4 12.Bxc4 Nd7
13.O-O b6 14.Bd3 c5 15.Be4 Rb8 16.Qc2 Nf6 17.dxc5 Nxe4 18.Qxe4
bxc5 19.Qc2 Bb7 20.Nd2 Rfd8 21.f3 Ba6 22.Rf2 Rd7 23.g3 Rbd8
24.Kg2 Bd3 25.Qc1 Ba6 26.Ra3 Bb7 27.Nb3 Rc7 28.Na5 Ba8 29.Nc4
e5 30.e4 f5 31.exf5 e4 32.fxe4 Qxe4+ 33.Kh3 Rd4 34.Ne3 Qe8
35.g4 h5 36.Kh4 g5+ 37.fxg6 Qxg6 38.Qf1 Rxg4+ 39.Kh3 Re7
40.Rf8+ Kg7 41.Nf5+ Kh7 42.Rg3 Rxg3+ 43.hxg3 Qg4+ 44.Kh2 Re2+
45.Kg1 Rg2+ 46.Qxg2 Bxg2 47.Kxg2 Qe2+ 48.Kh3 c4 49.a4 a5
50.Rf6 Kg8 51.Nh6+ Kg7 52.Rb6 Qe4 53.Kh2 Kh7 54.Rd6 Qe5 55.Nf7
Qxb2+ 56.Kh3 Qg7 0-1[/pgn]

I know that it is a different case because the black pawn on c6 remains there until the very end of the game plus the white pawn on a6.

Regards from Spain.

Ajedrecista.

[Uri Blass]

Hello Uri:

[...]

It does not seem that the opening position is unbalanced [...]

The opening has a reputation of being solid: the Tartakower Variation of the QGD, also known as the Tartakower–Makogonov–Bondarevsky System (ECO D58). The game that you have posted starts with the following eight moves (transpositions may apply):

1. d4 d5 2. c4 e6 3. Nc3 Nf6 4. Bg5 Be7 5. Nf3 h6 6. Bh4 O-O 7. e3 b6 8. Be2 Bb7

According to the data provided by the PGN at ProDeo forum, there were 406 games (W: 114; D: 223; B: 69) and a white performance of 55.5% more less, pretty standard IMHO.

I only found two games at Lichess Masters database until 14. ... Bg7 (6. ... Bg7 in your PGN), so 15. a4 could be a novelty:

Fish ½—½ Baramidze. (2003).
Gerbich ½—½ Trofimov. (2020).

A fairly old source like the First Edition of the Volume D of the Encyclopedia of Chess Openings (year 1976) has many mainlines on ECO D58. The mainline listed as 11 follows the game until 11. O-O, then deviates with 11. ... Qe7, but note 64 mentions 11. ... Nd7 and a different continuation from Furman—Klovans, SSSR 1972: 12. b4 Re8 13. b5 Nc5 14. Nd2 Ne6 15. Qb3 Qd6 16. a4 → white stands slightly better.

By the way, the same source gives a slighlty different line that in your PGN in the same mainline 11, just changing 11. ... Nd7 by 11. ... Qe7 (as seen above), then 12. Qb3 [{note 65} Karasev—Klovans, SSSR 1974: 12. Rc1 Nd7 13. Qa4 a6, etc. → the game is even] c6 {note 66 with other continuations} 13. Rfe1 Bc8 14. Rac1 {note 67 with other continuations}, etc. → white stands slightly better. This is, the white rooks on c1 and e1 (like in the game of the thread), the white queen also on b3 and the black pawn also on c6. With the great difference of black bringing the bishop back to c8, which was a concern by you.

------------------------

[...] but it seems that black bishop a8 is out of the game.

Of course even if the black bishop is out of the game it does not prove that white is winning because the position may be a type of fortress when white cannot break the fortress.

A bishop on a8 does not need to be out of the game always, as in the last game of the WCC 2010, where Anand won with black against Topalov and played 28. ... Ba8 until it was captured with 47. Kxg2.

Topalov 0—1 Anand. WCC 2010, Sofia (Bulgaria).

[pgn][Event "Anand - Topalov World Championship Match"]
[Site "Sofia BUL"]
[Date "2010.05.11"]
[EventDate "2010.04.26"]
[Round "12"]
[Result "0-1"]
[White "Veselin Topalov"]
[Black "Viswanathan Anand"]
[ECO "D56"]
[WhiteElo "2805"]
[BlackElo "2787"]
[PlyCount "112"]

1.d4 d5 2.c4 e6 3.Nf3 Nf6 4.Nc3 Be7 5.Bg5 h6 6.Bh4 O-O 7.e3
Ne4 8.Bxe7 Qxe7 9.Rc1 c6 10.Be2 Nxc3 11.Rxc3 dxc4 12.Bxc4 Nd7
13.O-O b6 14.Bd3 c5 15.Be4 Rb8 16.Qc2 Nf6 17.dxc5 Nxe4 18.Qxe4
bxc5 19.Qc2 Bb7 20.Nd2 Rfd8 21.f3 Ba6 22.Rf2 Rd7 23.g3 Rbd8
24.Kg2 Bd3 25.Qc1 Ba6 26.Ra3 Bb7 27.Nb3 Rc7 28.Na5 Ba8 29.Nc4
e5 30.e4 f5 31.exf5 e4 32.fxe4 Qxe4+ 33.Kh3 Rd4 34.Ne3 Qe8
35.g4 h5 36.Kh4 g5+ 37.fxg6 Qxg6 38.Qf1 Rxg4+ 39.Kh3 Re7
40.Rf8+ Kg7 41.Nf5+ Kh7 42.Rg3 Rxg3+ 43.hxg3 Qg4+ 44.Kh2 Re2+
45.Kg1 Rg2+ 46.Qxg2 Bxg2 47.Kxg2 Qe2+ 48.Kh3 c4 49.a4 a5
50.Rf6 Kg8 51.Nh6+ Kg7 52.Rb6 Qe4 53.Kh2 Kh7 54.Rd6 Qe5 55.Nf7
Qxb2+ 56.Kh3 Qg7 0-1[/pgn]

I know that it is a different case because the black pawn on c6 remains there until the very end of the game plus the white pawn on a6.

Regards from Spain.

Ajedrecista.

Yes it is clearly a different case because there was a pawn at c6 and I thought that even after a move like c5 white can reply b5 and keep the bishop out of game.

In any case here is another evidence that stockfish is stupid and it gives an evaluation of more than +3 but choose to close the position to get a draw.
I thought their smart NN has better positional understanding than me in every position because it even understand some non trivial draws in endgames but it seems that it still have some things that humans understand and the net does not understand.

https://tcec-chess.com/#div=1de&game=2&season=23

Note that initially I thought that white is winning because lc0 could beat stockfish with white but it seems that stockfish wanted to lose more than lc0 wanted to win and it is not clear why stockfish helped lc0 to open the position and blundered by 35...Bxa6 that lc0 did not expect in the following game:

https://tcec-chess.com/#div=1de&game=1&season=23