The Sorcerer's Apprentice

[Lyudmil Tsvetkov]

PS. Please, do not tell me there are engines that would favour black.

The advantage is more than a pawn at the present depth, if Black plays Nd6. But I am trying to get it over 1.12 It has dropped to 1.08 again:


r1bqkb1r/ppp2pp1/5n1p/3P4/2nQP3/5N2/PPP2PPP/RNB1K2R b KQkq -

Engine: Sf20150403 MOD MP (512 MB)
by Tord Romstad, Marco Costalba and Joona Kiiski
.
.
.
35/39 10:18 +1.00 1...Nd6 2.O-O Ndxe4 3.Re1 Be7 4.Rxe4 Nxe4
5.Qxe4 O-O 6.Nc3 Re8 7.Bf4 Bg4 8.Qd3 Bf6
9.Nd2 Bxc3 10.bxc3 Bh5 11.Rc1 Bg6
12.Qf3 Rc8 13.c4 Qf6 14.h3 b6 (1.618.650.157) 2618
.
.
.
36/40 12:09 +0.99 1...Nd6 2.O-O Ndxe4 3.Re1 Be7 4.Rxe4 Nxe4
5.Qxe4 O-O 6.Bf4 Re8 7.Nc3 Bg4 8.Qd3 Bxf3
9.Qxf3 Bd6 10.Bxd6 Qxd6 11.Qd3 c6
12.g3 Rad8 13.Rd1 Qb4 14.b3 cxd5 (1.919.793.540) 2630
.
.
.
37/46 26:09 +1.12 1...Nd6 2.Nc3 Nfxe4 3.Nxe4 Qe7 4.O-O Nxe4
5.Be3 a6 6.Rae1 f5 7.Ne5 Qf6 8.Nd3 Qxd4
9.Bxd4 Kf7 10.f3 Nf6 11.Ne5+ Kg8
12.Ng6 Rh7 13.Nxf8 Kxf8 14.c4 Bd7 (4.129.012.497) 2629
.
.
.
38/46 31:56 +1.08 1...Nd6 2.Nc3 Nfxe4 3.Nxe4 Qe7 4.O-O Nxe4
5.Be3 a6 6.Rae1 f5 7.Ne5 Qf6 8.Nd3 Qxd4
9.Bxd4 Kf7 10.f3 Nf6 11.Ne5+ Kg8
12.Ng6 Rh7 13.Nxf8 Kxf8 14.c4 Bd7 (5.017.990.544) 2618

Could not be like that, after Ne4 the black king stays in the center and white has more than sufficient compensation, but you need depth here, and real depth, not just what SF sees.

I might also have posted a position from later in the game, where all tactics are excluded, but I am sure SF will still give winning score for black with a piece more.

[Lyudmil Tsvetkov]

[d]r2qk2r/pp2bpp1/1n3n1p/2pP4/2P1P1b1/3Q1N2/PP1N1PPP/R1B1K2R w KQkq - 4 13


for example, here is a quiet position, what is SF score?

I will be satisfied with any value over 400cps white edge.

[Eelco de Groot]

PS. Please, do not tell me there are engines that would favour black.

The advantage is more than a pawn at the present depth, if Black plays Nd6. But I am trying to get it over 1.12 It has dropped to 1.08 again:


r1bqkb1r/ppp2pp1/5n1p/3P4/2nQP3/5N2/PPP2PPP/RNB1K2R b KQkq -

Engine: Sf20150403 MOD MP (512 MB)
by Tord Romstad, Marco Costalba and Joona Kiiski
.
.
.
35/39 10:18 +1.00 1...Nd6 2.O-O Ndxe4 3.Re1 Be7 4.Rxe4 Nxe4
5.Qxe4 O-O 6.Nc3 Re8 7.Bf4 Bg4 8.Qd3 Bf6
9.Nd2 Bxc3 10.bxc3 Bh5 11.Rc1 Bg6
12.Qf3 Rc8 13.c4 Qf6 14.h3 b6 (1.618.650.157) 2618
.
.
.
36/40 12:09 +0.99 1...Nd6 2.O-O Ndxe4 3.Re1 Be7 4.Rxe4 Nxe4
5.Qxe4 O-O 6.Bf4 Re8 7.Nc3 Bg4 8.Qd3 Bxf3
9.Qxf3 Bd6 10.Bxd6 Qxd6 11.Qd3 c6
12.g3 Rad8 13.Rd1 Qb4 14.b3 cxd5 (1.919.793.540) 2630
.
.
.
37/46 26:09 +1.12 1...Nd6 2.Nc3 Nfxe4 3.Nxe4 Qe7 4.O-O Nxe4
5.Be3 a6 6.Rae1 f5 7.Ne5 Qf6 8.Nd3 Qxd4
9.Bxd4 Kf7 10.f3 Nf6 11.Ne5+ Kg8
12.Ng6 Rh7 13.Nxf8 Kxf8 14.c4 Bd7 (4.129.012.497) 2629
.
.
.
38/46 31:56 +1.08 1...Nd6 2.Nc3 Nfxe4 3.Nxe4 Qe7 4.O-O Nxe4
5.Be3 a6 6.Rae1 f5 7.Ne5 Qf6 8.Nd3 Qxd4
9.Bxd4 Kf7 10.f3 Nf6 11.Ne5+ Kg8
12.Ng6 Rh7 13.Nxf8 Kxf8 14.c4 Bd7 (5.017.990.544) 2618

Could not be like that, after Ne4 the black king stays in the center and white has more than sufficient compensation, but you need depth here, and real depth, not just what SF sees.

I might also have posted a position from later in the game, where all tactics are excluded, but I am sure SF will still give winning score for black with a piece more.

Hi Lyudmil,

I am just looking myself at this one position, so with the most shallow depth of 0 plies, I just see those two pawns in the center that White has, and one is going to fall again, immediately. So then there will just be a pawn on d5 and I would want to support that at least with c4, but if the pawn needs the queen there in the center to defend it, the center is vulnerable to attacks from all sides. And with one light piece missing, King attacks are already a lot harder too for White. I think he can forget about attacking and Black has not castled yet. It seems it could go to an endgame straight, if the only center pawn left on both sides is one too far extended d5 pawn. In the hypothetical ensuing ending, Black will have the Bishop pair and I see no White compensation for the Bishop pair. But as said, this is just a zero ply search from my a 0.005 Hz computer The compiler broke down years ago, there is just a Basic interpreter present.

Eelco

[zullil]

[d]r1bqkb1r/ppp2pp1/5n1p/3P4/2nQP3/5N2/PPP2PPP/RNB1K2R b KQkq - 1 9

This is a position arisen out of the opening.

How many engines recognise white is winning above?

PS. Please, do not tell me there are engines that would favour black.

The latest SF with MultiPV=2 suggests that White will work to draw from here:

Code: Select all

info depth 45 seldepth 66 multipv 1 score cp 130 nodes 187124782111 nps 22214152 hashfull 999 tbhits 0 time 8423674 pv c4d6 b1c3 f6e4 c3e4 d8e7 e1g1 d6e4 c1e3 a7a6 f1e1 f7f5 f3e5 e7f6 e5d3 e8f7 f2f3 f6d4 e3d4 e4f6 d3e5 f7g8 e5g6 h8h7 g6f8 g8f8 d4c5 f8g8 a1d1 c8d7 d5d6 c7d6 d1d6 a8c8 b2b4 c8e8 e1e8 d7e8 g1f2 e8a4 d6d8 g8f7 c2c4 f6d7 c5d4 g7g5 d8c8 f7g6 c8g8 g6f7 g8d8 f7e7

info depth 45 seldepth 66 multipv 2 score cp 63 nodes 187124782111 nps 22214152 hashfull 999 tbhits 0 time 8423674 pv c4b6 e1g1 c8g4 e4e5 c7c5 d4f4 g4f3 e5f6 f3d5 f4e5 e8d7 f1d1 d8f6 e5f6 g7f6 c2c4 d7c6 c4d5 b6d5 c1d2 f8g7 b1c3 h8d8 c3d5 d8d5 d2c3 d5d1 a1d1 b7b5 d1c1 a8d8 g1f1 h6h5 c1c2 d8d6 c3b4 d6d5 b4c3 c6b6 f1e2 d5d6 b2b3 g7h6 g2g3 f6f5 f2f4 h5h4 c3e5 d6g6 e2f3 h4g3 h2g3 a7a6 c2c1

[Lyudmil Tsvetkov]

Btw., as you probably very well know, Bronstein was not a Soviet, but an Ukranian, just like Boleslavski, Geller, Stein, etc.

[Lyudmil Tsvetkov]

[d]r1bqkb1r/ppp2pp1/5n1p/3P4/2nQP3/5N2/PPP2PPP/RNB1K2R b KQkq - 1 9

This is a position arisen out of the opening.

How many engines recognise white is winning above?

PS. Please, do not tell me there are engines that would favour black.

The latest SF with MultiPV=2 suggests that White will work to draw from here:

Code: Select all

info depth 45 seldepth 66 multipv 1 score cp 130 nodes 187124782111 nps 22214152 hashfull 999 tbhits 0 time 8423674 pv c4d6 b1c3 f6e4 c3e4 d8e7 e1g1 d6e4 c1e3 a7a6 f1e1 f7f5 f3e5 e7f6 e5d3 e8f7 f2f3 f6d4 e3d4 e4f6 d3e5 f7g8 e5g6 h8h7 g6f8 g8f8 d4c5 f8g8 a1d1 c8d7 d5d6 c7d6 d1d6 a8c8 b2b4 c8e8 e1e8 d7e8 g1f2 e8a4 d6d8 g8f7 c2c4 f6d7 c5d4 g7g5 d8c8 f7g6 c8g8 g6f7 g8d8 f7e7

info depth 45 seldepth 66 multipv 2 score cp 63 nodes 187124782111 nps 22214152 hashfull 999 tbhits 0 time 8423674 pv c4b6 e1g1 c8g4 e4e5 c7c5 d4f4 g4f3 e5f6 f3d5 f4e5 e8d7 f1d1 d8f6 e5f6 g7f6 c2c4 d7c6 c4d5 b6d5 c1d2 f8g7 b1c3 h8d8 c3d5 d8d5 d2c3 d5d1 a1d1 b7b5 d1c1 a8d8 g1f1 h6h5 c1c2 d8d6 c3b4 d6d5 b4c3 c6b6 f1e2 d5d6 b2b3 g7h6 g2g3 f6f5 f2f4 h5h4 c3e5 d6g6 e2f3 h4g3 h2g3 a7a6 c2c1

Meaning, that with very long search, white is winning.

Bronstein's opponent played a busted opening, the 2 Knights, and he also had the misfortune to meet one of the best experts of this opening.

[zullil]

[d]r1bqkb1r/ppp2pp1/5n1p/3P4/2nQP3/5N2/PPP2PPP/RNB1K2R b KQkq - 1 9

This is a position arisen out of the opening.

How many engines recognise white is winning above?

PS. Please, do not tell me there are engines that would favour black.

The latest SF with MultiPV=2 suggests that White will work to draw from here:

Code: Select all

info depth 45 seldepth 66 multipv 1 score cp 130 nodes 187124782111 nps 22214152 hashfull 999 tbhits 0 time 8423674 pv c4d6 b1c3 f6e4 c3e4 d8e7 e1g1 d6e4 c1e3 a7a6 f1e1 f7f5 f3e5 e7f6 e5d3 e8f7 f2f3 f6d4 e3d4 e4f6 d3e5 f7g8 e5g6 h8h7 g6f8 g8f8 d4c5 f8g8 a1d1 c8d7 d5d6 c7d6 d1d6 a8c8 b2b4 c8e8 e1e8 d7e8 g1f2 e8a4 d6d8 g8f7 c2c4 f6d7 c5d4 g7g5 d8c8 f7g6 c8g8 g6f7 g8d8 f7e7

info depth 45 seldepth 66 multipv 2 score cp 63 nodes 187124782111 nps 22214152 hashfull 999 tbhits 0 time 8423674 pv c4b6 e1g1 c8g4 e4e5 c7c5 d4f4 g4f3 e5f6 f3d5 f4e5 e8d7 f1d1 d8f6 e5f6 g7f6 c2c4 d7c6 c4d5 b6d5 c1d2 f8g7 b1c3 h8d8 c3d5 d8d5 d2c3 d5d1 a1d1 b7b5 d1c1 a8d8 g1f1 h6h5 c1c2 d8d6 c3b4 d6d5 b4c3 c6b6 f1e2 d5d6 b2b3 g7h6 g2g3 f6f5 f2f4 h5h4 c3e5 d6g6 e2f3 h4g3 h2g3 a7a6 c2c1

Meaning, that with very long search, white is winning.

No, the scores are from the engine's point of view (as per the UCI protocol), and thus favor Black here.

[zullil]

[d]r2qk2r/pp2bpp1/1n3n1p/2pP4/2P1P1b1/3Q1N2/PP1N1PPP/R1B1K2R w KQkq - 4 13


for example, here is a quiet position, what is SF score?

I will be satisfied with any value over 400cps white edge.

Code: Select all

info depth 42 seldepth 56 multipv 1 score cp 0 nodes 18649081346 nps 21788371 hashfull 999 tbhits 0 time 855919 pv e1g1 e8g8 b2b3 f6h5 c1b2 d8c7 e4e5 c7d7 h2h3 g4e6 d5d6 e6h3 g2h3 d7h3 d3e4 f7f5 e4e3 h3g4 g1h1 g4h3 h1g1

[Lyudmil Tsvetkov]

[d]r1bqkb1r/ppp2pp1/5n1p/3P4/2nQP3/5N2/PPP2PPP/RNB1K2R b KQkq - 1 9

This is a position arisen out of the opening.

How many engines recognise white is winning above?

PS. Please, do not tell me there are engines that would favour black.

The latest SF with MultiPV=2 suggests that White will work to draw from here:

Code: Select all

info depth 45 seldepth 66 multipv 1 score cp 130 nodes 187124782111 nps 22214152 hashfull 999 tbhits 0 time 8423674 pv c4d6 b1c3 f6e4 c3e4 d8e7 e1g1 d6e4 c1e3 a7a6 f1e1 f7f5 f3e5 e7f6 e5d3 e8f7 f2f3 f6d4 e3d4 e4f6 d3e5 f7g8 e5g6 h8h7 g6f8 g8f8 d4c5 f8g8 a1d1 c8d7 d5d6 c7d6 d1d6 a8c8 b2b4 c8e8 e1e8 d7e8 g1f2 e8a4 d6d8 g8f7 c2c4 f6d7 c5d4 g7g5 d8c8 f7g6 c8g8 g6f7 g8d8 f7e7

info depth 45 seldepth 66 multipv 2 score cp 63 nodes 187124782111 nps 22214152 hashfull 999 tbhits 0 time 8423674 pv c4b6 e1g1 c8g4 e4e5 c7c5 d4f4 g4f3 e5f6 f3d5 f4e5 e8d7 f1d1 d8f6 e5f6 g7f6 c2c4 d7c6 c4d5 b6d5 c1d2 f8g7 b1c3 h8d8 c3d5 d8d5 d2c3 d5d1 a1d1 b7b5 d1c1 a8d8 g1f1 h6h5 c1c2 d8d6 c3b4 d6d5 b4c3 c6b6 f1e2 d5d6 b2b3 g7h6 g2g3 f6f5 f2f4 h5h4 c3e5 d6g6 e2f3 h4g3 h2g3 a7a6 c2c1

Meaning, that with very long search, white is winning.

No, the scores are from the engine's point of view (as per the UCI protocol), and thus favor Black here.

Sorry, I did not observe those were for 2 different lines.

I do not believe neither the one, nor the second line.
In both white has the advantage.

Look very carefully at the original game, to see for how long into the game SF will still give black advantage, when from human point of view white is obviously winning.

If you want, we might concentrate for a while on the second position I posted, as this is too tactically complicated after Nd6 to easily decide.

[d]r2qk2r/pp2bpp1/1n3n1p/2pP4/2P1P1b1/3Q1N2/PP1N1PPP/R1B1K2R w KQkq - 0 1

What does SF say here?

[zullil]

[d]r2qk2r/pp2bpp1/1n3n1p/2pP4/2P1P1b1/3Q1N2/PP1N1PPP/R1B1K2R w KQkq - 0 1

What does SF say here?

Code: Select all

info depth 42 seldepth 56 multipv 1 score cp 0 nodes 18649081346 nps 21788371 hashfull 999 tbhits 0 time 855919 pv e1g1 e8g8 b2b3 f6h5 c1b2 d8c7 e4e5 c7d7 h2h3 g4e6 d5d6 e6h3 g2h3 d7h3 d3e4 f7f5 e4e3 h3g4 g1h1 g4h3 h1g1