Page 1 of 1

test position. who is right the GM or the computer?

Posted: Mon Jan 27, 2014 2:09 pm
by Uri Blass
[D]5rk1/2q3b1/1rbp2p1/2p1p2p/2PnPp2/R2Q1N1P/P1B2PP1/2B1R1K1 b - - 0 1

I read a book of GM John Nunn "understanding chess middlegames" from 2011

GM John nunn claims that 30...Nxc2 is a strong move and suggests 31.Nxd4 against 30...Qd7

Stockfish does not like 30...Nxc2 and also does not like 31.Nxd4
Now the question is if stockfish is right or GM nunn is right.

The position is from the world correspondence championship 1996-2001 and black was Joop J. van Oosterom.

Re: test position. who is right the GM or the computer?

Posted: Mon Jan 27, 2014 4:43 pm
by AdminX
Hi Uri,

I could be wrong here but based on the principle that it is normally a good idea to maintain the tension as long as possible, I think Stockfish may be correct here. Then again if you are going to violate general principles you should always have a good reason. I am sure GM John Nunn has his own theories as he stated in his book, however I still think Stockfish is correct here. This might be a good position to run a little Monte Carlo analysis on.

Re: test position. who is right the GM or the computer?

Posted: Mon Jan 27, 2014 5:33 pm
by michiguel
Uri Blass wrote:[D]5rk1/2q3b1/1rbp2p1/2p1p2p/2PnPp2/R2Q1N1P/P1B2PP1/2B1R1K1 b - - 0 1

I read a book of GM John Nunn "understanding chess middlegames" from 2011

GM John nunn claims that 30...Nxc2 is a strong move and suggests 31.Nxd4 against 30...Qd7

Stockfish does not like 30...Nxc2 and also does not like 31.Nxd4
Now the question is if stockfish is right or GM nunn is right.

The position is from the world correspondence championship 1996-2001 and black was Joop J. van Oosterom.
To me, sounds like Nunn is right. Nd4 is going no matter what, so the question is, what does white prefer to keep, the N or Bc2? In this position, Nf3 does not seem to have good squares, and Bc2 is a "bad" bishop. However, in certain positions, "bad" bishops are the ones that keep everything together (see the famous Fischer Petrosian game) and become outstanding defenders. Black could be attacking in the K side, g5-g4 and Bc2 could redirect to d1 to fight against it. Nf3 seems a bit hopeless in that regard. What is Nunn rationale to suggest Nc2 so we can evaluate this better?

To summarize, as white, I prefer to keep Bc2 rather than the Nf3.

Considering how shaky is SF in King indian defense setups, I would be very careful and skeptical of its suggestions. Maybe there is a deep tactical justification for its preference, but the truth of these positions could very well fall behind horizon.

Miguel
PS: Gaviota prefers Nxc2 and Nxd4 against Qd7, so I have to trust that :-)

Re: test position. who is right the GM or the computer?

Posted: Mon Jan 27, 2014 6:05 pm
by zullil
Uri Blass wrote:[D]5rk1/2q3b1/1rbp2p1/2p1p2p/2PnPp2/R2Q1N1P/P1B2PP1/2B1R1K1 b - - 0 1

I read a book of GM John Nunn "understanding chess middlegames" from 2011

GM John nunn claims that 30...Nxc2 is a strong move and suggests 31.Nxd4 against 30...Qd7

Stockfish does not like 30...Nxc2 and also does not like 31.Nxd4
Now the question is if stockfish is right or GM nunn is right.


The position is from the world correspondence championship 1996-2001 and black was Joop J. van Oosterom.
Latest Stockfish (16 threads, 8 GB hash):

Code: Select all

info depth 46 seldepth 61 score cp 106 nodes 135288908681 nps 19290321 time 7013305 multipv 1 pv g7f6 f3d4 c5d4 a3b3 b6b8 b3b8 f8b8 c1d2 c6e8 a2a4 e8f7 c2b3 g8g7 a4a5 f6d8 e1a1 c7c6 a1a4 b8a8 d2b4 a8a7 a4a1 f7e6 d3d2 c6a6 d2d3 a7b7 b4d2 a6c6 a5a6

Re: test position. who is right the GM or the computer?

Posted: Mon Jan 27, 2014 6:28 pm
by beachknight
Quick analysis of the position with SF GTB DD:

20:
5rk1/2q3b1/1rbp2p1/2p1p2p/2PnPp2/R2Q1N1P/P1B2PP1/2B1R1K1 b - - 0 1

Analysis by Stockfish PA_GTB DD 64 3cpu:

1...Bf6 2.Bd2 Nxc2 3.Qxc2 Be7 4.Qd3
-/+ (-1.17 ++) Depth: 33/51 00:14:09 1902MN
(Chess Engines Testing, 27.01.2014)

Re: test position. who is right the GM or the computer?

Posted: Mon Jan 27, 2014 6:31 pm
by beachknight
At depth 46, latest SF (without egtbs, I suppose) still favours Bf6.
Best,

Re: test position. who is right the GM or the computer?

Posted: Mon Jan 27, 2014 6:41 pm
by beachknight
After 30. ... Qd7, no exchange is in sight so far:

20:
5rk1/3q2b1/1rbp2p1/2p1p2p/2PnPp2/R2Q1N1P/P1B2PP1/2B1R1K1 w - - 0 1

Analysis by Stockfish PA_GTB DD 64:

2.Bd2 Rbb8 3.Rb1 Nxc2 4.Qxc2 Rxb1+ 5.Qxb1 Bf6 6.Ra6 Ra8 7.Rxa8+ Bxa8 8.h4 Kf7 9.Ng5+ Bxg5 10.hxg5 Qb7 11.Qxb7+ Bxb7 12.f3 Ba6 13.a4 Bxc4 14.Kf2 Ke6 15.g3 fxg3+ 16.Kxg3 Bd3 17.Be1 Bc2 18.a5 Bd3 19.Kf2 Kd7 20.Ke3 Bc4 21.Bg3 Kc6 22.f4 exf4+ 23.Bxf4 h4
-/+ (-1.02) Depth: 35/50 00:14:00 1887MN, tb=2
(Chess Engines Testing, 27.01.2014)

Re: test position. who is right the GM or the computer?

Posted: Mon Jan 27, 2014 8:07 pm
by zullil
zullil wrote:
Uri Blass wrote:[D]5rk1/2q3b1/1rbp2p1/2p1p2p/2PnPp2/R2Q1N1P/P1B2PP1/2B1R1K1 b - - 0 1

I read a book of GM John Nunn "understanding chess middlegames" from 2011

GM John nunn claims that 30...Nxc2 is a strong move and suggests 31.Nxd4 against 30...Qd7

Stockfish does not like 30...Nxc2 and also does not like 31.Nxd4
Now the question is if stockfish is right or GM nunn is right.


The position is from the world correspondence championship 1996-2001 and black was Joop J. van Oosterom.
Latest Stockfish (16 threads, 8 GB hash):

Code: Select all

info depth 46 seldepth 61 score cp 106 nodes 135288908681 nps 19290321 time 7013305 multipv 1 pv g7f6 f3d4 c5d4 a3b3 b6b8 b3b8 f8b8 c1d2 c6e8 a2a4 e8f7 c2b3 g8g7 a4a5 f6d8 e1a1 c7c6 a1a4 b8a8 d2b4 a8a7 a4a1 f7e6 d3d2 c6a6 d2d3 a7b7 b4d2 a6c6 a5a6
g7f6 failed low, but SF hasn't resolved depth 47 in more than an hour (at almost 20M nps).

Code: Select all

info depth 47 currmove g7f6 currmovenumber 1
info depth 47 currmove d4c2 currmovenumber 2
info depth 47 seldepth 61 score cp 114 lowerbound nodes 181676424198 nps 19532228 time 9301367 multipv 1 pv c7d7 c1d2 g8h8 a3a5 g7f6 f3d4 c5d4 c4c5
info depth 47 currmove c7d7 currmovenumber 1
info depth 47 seldepth 64 score cp 122 lowerbound nodes 224812829957 nps 19595895 time 11472445 multipv 1 pv c7d7 a3a5 g8h8 c1a3 b6b8 f3d4 c5d4 c4c5 f4f3 c5d6 d7d8 a5a7 c6b5 d3d2 f3g2 g1g2 d8h4
info depth 47 currmove c7d7 currmovenumber 1
OK, finished depth 47:

Code: Select all

info depth 47 seldepth 86 score cp 114 nodes 332312554173 nps 19662317 time 16900986 multipv 1 pv c7d7 a3a5 g8h8 c1d2 g7f6 f3d4 c5d4 d3a3 f6e7 a5a7 b6b7 a7b7 c6b7 e1b1 f8a8 a3b3 b7c6 b1a1 c6a4 b3d3 h8g7 c2a4 a8a4 a2a3 d7c6 d2b4 a4a8 b4d2 g7f6 d2b4 a8c8 a1c1 f6g7 d3c2