Not logged inRybka Chess Community Forum
Up Topic The Rybka Lounge / Test Positions / Speaking of Abillo positions, can any engine solve this one
- - By dcorbit (***) Date 2010-11-17 10:36
Fig. 6.1: Korchnoi vs. Fischer: Black to play

FEN:
r1bq2k1/p5bp/3p1nnr/P1pPp3/2P1Pp2/1NN2R1P/4B1P1/1R1QB1K1 b - - bm Nh8;


1. ... Nh8

Reviewer Notes:
It's quite remarkable that Fischer was able to produce such a move in a few seconds. Modern chess programs, however, not only do not select that move in a few seconds, but neither do they when given many hours, preferring other alternatives.
Parent - - By HarshQueen (**) Date 2010-11-17 13:06

>1. ... Nh8


>Reviewer Notes:


>It's quite remarkable that Fischer was able to produce such a move in a few seconds. Modern chess programs, however, not only do not select that move in a few seconds, but neither do they when given many hours, preferring other alternatives.


r1bq2k1/p5bp/3p1nnr/P1pPp3/2P1Pp2/1NN2R1P/4B1P1/1R1QB1K1 b - - 0 1


Prohibited w32 2_CPU:

1...Nh8 2.Nd2 Nf7 3.Qc2 Qxa5 4.Rf1 Rg6 5.Kh1 a6 6.Nf3 Qc7 7.Nh4 Rg5 8.Nf3
  =  (0.00)   Depth: 11/24   00:00:00  242kN
1...Nh8 2.Nd2 Nf7 3.Qc2 Qxa5 4.Rf1 Rg6 5.Kh1 a6 6.Nf3 Qc7 7.Nh4 Rg5 8.Nf3
  =  (0.00)   Depth: 12/25   00:00:00  254kN
1...Nh8 2.Nd2 Nf7 3.Qc2 Qxa5 4.Rf1 Rg6 5.Kh1 a6 6.Nf3 Qc7 7.Nh4 Rg5 8.Nf3
  =  (0.00)   Depth: 13/25   00:00:00  280kN
1...Nh8 2.Nd2 Nf7 3.Qc2 Qxa5 4.Rf1 Rg6 5.Kh1 a6 6.Nf3 Qc7 7.Nh4 Rg5 8.Nf3 Rg6 9.Nh4
  =  (0.00)   Depth: 14/25   00:00:00  320kN
1...Nh8 2.Nd2 Nf7 3.Qc2 Qxa5 4.Rf1 Rg6 5.Kh1 a6 6.Nf3 Qc7 7.Nh4 Rg5 8.Nf3 Rg6 9.Nh4
  =  (0.00)   Depth: 15/29   00:00:00  407kN
1...Nh8 2.Nd2 Nf7 3.Qc2 Qxa5 4.Rf1 Rg6 5.Kh1 a6 6.Nf3 Qc7 7.Nh4 Rg5 8.Nf3 Rg6 9.Nh4
  =  (0.00)   Depth: 16/30   00:00:00  654kN
1...Nh8 2.Nd2 Nf7 3.Qc2 Qxa5 4.Rf1 Rg6 5.Kh1 a6 6.Nf3 Qc7 7.Nh4 Rg5 8.Nf3 Rg6 9.Nh4
  =  (0.00)   Depth: 17/33   00:00:00  964kN
1...Nh8 2.Nd2 Nf7 3.Qc2 Qxa5 4.Rf1 Rg6 5.Kh1 a6 6.Nf3 Qc7 7.Nh4 Rg5 8.Nf3 Rg6 9.Nh4
  =  (0.00)   Depth: 18/33   00:00:00  1703kN
1...Nh8 2.Nd2 Nf7 3.Qc2 Qxa5 4.Rf1 Rg6 5.Kh1 a6 6.Nf3 Qc7 7.Nh4 Rg5 8.Nf3 Rg6 9.Nh4
  =  (0.00)   Depth: 19/41   00:00:02  3699kN
1...Nh8 2.Nd2 Nf7 3.Qc2 Qxa5 4.Rf1 Rg6 5.Kh1 a6 6.Nf3 Qc7 7.Nh4 Rg5 8.Nf3 Rg6 9.Nh4
  =  (0.00)   Depth: 20/41   00:00:03  7617kN
1...Nh8 2.Nd2 Nf7 3.Qc2 Qxa5 4.Rf1 Rg6 5.Kh1 a6 6.Nf3 Nh5 7.Na4 Ng3+ 8.Bxg3 Rxg3 9.Nb6 Rb8 10.Nxc8 Rxc8 11.Ra1 Qb6 12.Rfb1
Qa7 13.Nh4 Nh6 14.Nf3 Nf7 15.Nh4
  =  (0.00)   Depth: 21/43   00:00:10  20549kN
Parent - By HarshQueen (**) Date 2010-11-17 13:18
[White "Kortschnoj, Viktor"]
[Black "Fischer, Robert James"]
[Result "0-1"]
[ECO "E97"]

1. d4 Nf6 2. c4 g6 3. Nc3 Bg7 4. e4 d6 5. Be2 O-O 6. Nf3 e5 7. O-O Nc6 8. d5
Ne7 9. Nd2 c5 10. a3 Ne8 11. b4 b6 12. Rb1 f5 13. f3 f4 14. a4 g5 15. a5 Rf6
16. bxc5 bxc5 17. Nb3 Rg6 18. Bd2 Nf6 19. Kh1 g4 20. fxg4 Nxg4 21. Rf3 Rh6 22.
h3 Ng6 23. Kg1 Nf6 24. Be1 Nh8 25. Rd3 Nf7 26. Bf3 Ng5 27. Qe2 Rg6 28. Kf1 Nxh3
29. gxh3 {???} Bxh3+ 30. Kf2 Ng4+ 31. Bxg4 Bxg4 0-1
Parent - - By dcorbit (***) Date 2010-11-17 20:46
The score certainly is not very convincing (0.0).  I am not convinced that Nh8 is better than Bf8, for instance:

r1bq2k1/p5bp/3p1nnr/P1pPp3/2P1Pp2/1NN2R1P/4B1P1/1R1QB1K1 b - - acd 26/73; acn 3223447977; acs 746; bm Bf8; ce 8;
pv Bf8 Nd2 Qxa5 Qa4 Qd8 Qc6 Bd7 Qb7 Nh4 Rd3 a6 Bd1 Qc8 Qxc8 Rxc8 Rb7 Ra8 Na4 Bc8 Rb1 Nd7 Bg4 Ra7 Nf3 Nxf3+ Rxf3 a5 Bf5 Rh5 h4 Kg7;
Parent - - By HarshQueen (**) Date 2010-11-18 07:57 Edited 2010-11-18 08:06

>The score certainly is not very convincing (0.0).  I am not convinced that Nh8 is better than Bf8, for instance:


Your right, but  *dcorbit wrote: Modern chess programs, however, not only do not select that move in a few seconds, but neither do they when given many hours, preferring other alternatives

From my engine collection i found only this engine with the move Nh8!? This engine i use in my corr. games and
i have good results, very fine results indeed.
Parent - By HarshQueen (**) Date 2010-11-19 08:18
Maybe anyone can try it with other engines.
Parent - - By Dadi Jonsson (Silver) Date 2010-11-17 23:00
Interesting position. It's easy enough to spot the Nh8-Nf7-Ng5 maneuver. Any decent chess player would consider it (and some players will probably think that your position is too easy because they can see this maneuver). However, the hard part is understanding the position and being able to tell if Nh8 is actually better than other moves. It's definitely one of the best moves, but is it the best one? If an engine comes up with that move with around 0.00 evaluation, it hasn't told us much if we don't have any other information.

Here is what quick analysis with Rybka 4 gives with MultiPV set to two:

[-0.14]  d=20  1...Nh4 2.Rd3 Bf8 3.Nd2 Qxa5 4.Bf2 Qc7 5.Qe1 Ng6 6.Nf3 Rb8 7.Rxb8 Qxb8 8.Rd2 Qb7 9.Kf1 Ne7 10.Qa1 Rg6 11.Rb2 Qc7 12.Bd3 (0:20:06) 59484kN
[-0.11]  d=20  1...Bf8 2.Rf2 Nh4 3.Nd2 Rg6 4.Kh1 Be7 5.Qa4 Bd7 6.Qa2 Rb8 7.Rxb8 (0:24:15) 75300kN


I wouldn't rule out the possibility that one of these moves is just as good as Nh8, but much more analysis is needed.
Parent - - By dcorbit (***) Date 2010-11-18 19:55
Louis Zulli gave this 40 ply stockfish analysis also in CCC:

42    +0.97 14:09:34  438410M Nh4 Bxh4 Rxh4 Rf2 Rh6 Bf3 Rg6 Kf1 Ba6 Nb5 Bh6
                              Rfb2 Qe7 Nc1 Kh8 Qa4 Rag8 Nc3 Qc7 Nd3 Bg5 Nf2

So Nh4 is another solid alternative.
Parent - By Lazy Frank (****) Date 2011-03-21 11:48
Mdaaa ......

This analysis is totally wrong, looks like chess play one engine and opponent only follows ....
Bad algorithm example.
Parent - - By Dadi Jonsson (Silver) Date 2010-11-20 20:13 Edited 2010-11-20 20:20
I analyzed the position with IDeA. 1...Nh4 looks like the best move although Stockfish may be a bit generous with it's +1 evaluation. 1...Bf8 is not far behind and these two moves often transpose into the same variations. However, 1...Nh8 stayed at 0.00 after 2.Nd2. Korchnoi's 2.Rd3 isn't bad either, but after 2...Nf7 his 3.Bf3 doesn't look good and 3.Nd2 was better. Of course we have to remember that this was a blitz game between the masters.

After 1...Nh4 2.Bxh4 (2.Rd3 also comes into consideration, but Black could answer with 2...Bf8) 2...Rxh4 as seen in the Stockfish PV in Louis Zulli's analysis,

r1bq2k1/p5bp/3p1n2/P1pPp3/2P1Pp1r/1NN2R1P/4B1P1/1R1Q2K1 w - - 0 3


3.Nd2 scores higher than 3.Rf2 which was suggested by Stockfish (-0.26 vs. -0.37).

Here is a screenshot of the IDeA tree in the position you posted.
Parent - By TheHug (Bronze) Date 2010-11-20 21:34
This is good AN work. One also has to understand how difficult it is to understand a King's Indian position. Engines have difficultly in understanding it.
Parent - By dcorbit (***) Date 2010-11-22 20:11
It is amazing what computer analysis can do today.
With a clogged middle like that where all the best moves are non-captures, a few years ago computers would flail helplessly at positions like this.

As for Stockfish's analysis:
Stockfish is always over-optimistic in evaluations until the endgame where it levels out.
You can more or less divide the evaluation by two in the middle game and by four in the early opening.  (IOW, you will see evaluations of +0.5 for the slightest edge in an opening position).
Parent - By Master Om (Bronze) Date 2010-12-02 10:12
What were your analysis Quality  IDeA settings ?
Up Topic The Rybka Lounge / Test Positions / Speaking of Abillo positions, can any engine solve this one

Powered by mwForum 2.27.4 © 1999-2012 Markus Wichitill