Stockfish entombs a bishop

Discussion of anything and everything relating to chess playing software and machines.

Moderators: bob, hgm, Harvey Williamson

Forum rules
This textbox is used to restore diagrams posted with the [d] tag before the upgrade.
User avatar
Evert
Posts: 2929
Joined: Fri Jan 21, 2011 11:42 pm
Location: NL
Contact:

Re: Stockfish entombs a bishop

Post by Evert » Sun Mar 02, 2014 1:13 pm

I'm not an IM, but I'd say that black's bishop is bad whether black plays h5 or not...

Michel
Posts: 2102
Joined: Sun Sep 28, 2008 11:50 pm

Re: Stockfish entombs a bishop

Post by Michel » Sun Mar 02, 2014 5:08 pm

The point is that black can exchange the bad bishop against the knight I think.

IGarcia
Posts: 543
Joined: Mon Jul 05, 2010 8:27 pm

Re: Stockfish entombs a bishop

Post by IGarcia » Sun Mar 02, 2014 6:26 pm

modolief wrote:Here's a position from a 15 minute game played by a IM Chessexplained on ICC. He's analyzing after the game with Stockfish. At 32:45 he points out that Stockfish's choice of move for the position, 17..h5 for black, is obviously bad.

http://www.youtube.com/watch?v=uIkhZInjkbE

Image[/img]
Besides it is a move I probably do not find while playing... h5 seems a good move as my understand, because I would like to take the Knight at g6. It also keeps the h file closed and gains some space.

yanquis1972
Posts: 1766
Joined: Tue Jun 02, 2009 10:14 pm

Re: Stockfish entombs a bishop

Post by yanquis1972 » Sun Mar 02, 2014 9:09 pm

i dont understand why engines dont immediately think to move Bxg5 & stick w/ it, but both houdini & SF require some depth to see it is inferior. maybe one of the authors can explain it.

carldaman
Posts: 1962
Joined: Sat Jun 02, 2012 12:13 am

Re: Stockfish entombs a bishop

Post by carldaman » Sun Mar 02, 2014 10:22 pm

yanquis1972 wrote:i dont understand why engines dont immediately think to move Bxg5 & stick w/ it, but both houdini & SF require some depth to see it is inferior. maybe one of the authors can explain it.
If Black plays ...Bxg5, then the h-pawn, still on h7, would be backward and weak after White's hxg5.

CL

IGarcia
Posts: 543
Joined: Mon Jul 05, 2010 8:27 pm

Re: Stockfish entombs a bishop

Post by IGarcia » Mon Mar 03, 2014 1:40 am

carldaman wrote:
yanquis1972 wrote:i dont understand why engines dont immediately think to move Bxg5 & stick w/ it, but both houdini & SF require some depth to see it is inferior. maybe one of the authors can explain it.
If Black plays ...Bxg5, then the h-pawn, still on h7, would be backward and weak after White's hxg5.

CL
Yes, and is not possible to play h5 because now the g pawn takes e.p. and white has a passed pawn with a rook behind!


So h5 prepares the capture, closes the column, win some space, very nice. Still I prefer to play here as white.

User avatar
Eelco de Groot
Posts: 4249
Joined: Sun Mar 12, 2006 1:40 am
Location: Groningen

Re: Stockfish entombs a bishop

Post by Eelco de Groot » Mon Mar 03, 2014 7:04 pm

Sometimes computerplay can be very counterintuitive. In this case, at least from static eval the computer has no chance to see the pattern that the bishop is going to be locked up, so it is possible that 17... Bxg5 would get a higher evaluation than 17... h5 if it did see. But then for instance in the following PV after 17... Bxg5:


[FEN "r3n1k1/1p2pp1p/1p2bbp1/3p2N1/3P3P/P1N1P3/1P3PP1/4KB1R b K -"]

1... Bxg5 2. hxg5 Kg7 3. Be2 f6 4. gxf6+ Nxf6 5. Kd2 h5
6. f3 Ne8 7. e4 dxe4 8. fxe4 Rd8 9. d5 Bf7 10. Ke3 Nf6
11. Rd1 e6 12. dxe6 Rxd1 13. Bxd1 Bxe6 14. Kd4 Nd7 *

the move 3... f6 seems strange again. The IM would have a field day explaining that this is obviously bad again because Black compromises its king shelter, without much need because only the rook on h1 is a threat. One piece is not a threat, but why does the king not just stay put. But I did a very deep search and at the end 3... f6 is still on top, in both PVs of the two best moves.


[D]r3n3/1p2ppkp/1p2b1p1/3p2P1/3P4/P1N1P3/1P2BPP1/4K2R b K -

Engine: Scid Serpent 20140216_009 MP (512 MB)
by Tord Romstad, Marco Costalba and Joona Kiiski

45 870:19 -0.94 3...f6 {!? Stupid computermove} 4.gxf6+ Nxf6 5.Kd2 h5 6.f3 Bf5
7.Bd1 Rc8 8.Re1 Bd7 9.e4 dxe4
10.fxe4 e6 11.Rh1 Bc6 12.Kd3 Rf8
13.Be2 Ra8 14.g3 Rf8 15.Rf1 Rd8
16.Ke3 Rf8 (171.281.380.307) 3279

44 870:19 -0.94 3...Ra5 4.Bf3 f6 5.gxf6+ Nxf6 6.Kd2 Ra8
7.Be2 h5 8.f3 Bf5 9.Bd1 Rc8 10.Re1 Bd7
11.e4 dxe4 12.fxe4 e6 13.Rh1 Bc6
14.Kd3 Rf8 15.Be2 Ra8 16.g3 Rf8 (171.281.380.307) 3279

best move: f7-f6 time: 870:20.632 min n/s: 3.279.997 nodes: 171.281.380.307

After such a deep search I give up, Serpent should now see f6 is bad but does not, obviously the IM must be wrong.

Eelco
Debugging is twice as hard as writing the code in the first
place. Therefore, if you write the code as cleverly as possible, you
are, by definition, not smart enough to debug it.
-- Brian W. Kernighan

Post Reply