Testposition 60 - Bishop vs Knight

[Uri Blass]

Bishop and knight are not "near equal" in an endgame with pawns on both wings. Bishop is clearly superior there as it can reach squares in one move that take a knight 2-3 (or more) moves to reach. I have an evaluation term for this, where I think the idea came from Fine/Kmoch/et. al.

That is an exaggeration based on one particular position. Pawns on both wings means nothing. Take for example the position from the famous Saidy vs Fischer (1964) game. Semi open position, pawns on both wings, no blocked pawns. Position is probably slightly better for black but Fischer won that game easily (partly because he is the Fischer).

[D]6k1/1p1n1ppp/p7/3p4/3P4/P7/1P3PPP/2B3K1 w - - 0 1

Your example is different because of the d4-d5 pawns cut the board in two. This is different as premise of pawns in both wings without central pawns present which is the context made by Dr. Hyatt.

The example is different but
karlo is right.

The bishop is not generally better when there are pawns in both wings.

if you talk about the case when there are pawns in both wings and the side with the bishop has no central blocked pawns that block the bishop to move then it is a different case.

[Kyodai]

Karlos example is not just different - but in fact totally different from the
basic position in Stoltz-Kashdan.

Of course due to the blocked pawns in the centre d4/d5.

(Block a few more pawns for white on the same color as the bishop and
the knight will be superior - of course... we have shown positions like that
also)

[Uri Blass]

Karlos example is not just different - but in fact totally different from the
basic position in Stoltz-Kashdan.

Of course due to the blocked pawns in the centre d4/d5.

(Block a few more pawns for white on the same color as the bishop and
the knight will be superior - of course... we have shown positions like that
also)

karlo responded to Bob's words and here are the words:

"Bishop and knight are not "near equal" in an endgame with pawns on both wings. Bishop is clearly superior there as it can reach squares in one move that take a knight 2-3 (or more) moves to reach."

Cases when there are pawns on both wings and there are also blocked pawns are common in chess so practically it is not correct to say that the bishop is better when there are pawns on both wings and chess players or programmers may do mistakes in their evaluation function because of believing in this principle.

Uri

[Lyudmil Tsvetkov]

Thanks for your contribution and analyse Lyudmil!

Right now I have limited time but will be back with this position in a couple of days.

Regarding 1-a5 - i was under the impression that white worsen his position by playing an early b4 - have a look at the previous analyse - black won all those variations.

Most problems we have with the Karsten Mueller Barrier - built up by keeping
the b-pawn on b3 - putting the Kingside pawns on f3-g3-h4 - the King
somewhere on the third rank - and the Knight on d4 (c2-e2).

Worked like a locked tin can which we eventually forced to open up by white
having to play f4 (zugzwang)
The first variation black won but just waiting one move for white with f4
and then using the Knight for tactics looked sufficient for a draw.

That's a summary where I ended up keeping the black bishop on d7 and
actually start with 1-f6 - the problem being to find the ultimate setup for
black.

But I will surely have a look att your ideas - thanks!

Hi Sune.
Again, I do not have time to analyse, and I actually nowadays never analyse, just play games. This was just my feeling based on intuition.

I would suggest looking at the following variations (if you actually are in the mood of analysing):
1...Kf8 2.b4 Ke7 3. Kf1 Kd6 4. Ke2 Kd5 5. Kd3 and now:
5...g5 6. Nf3
5...h5 6. Kc3 Ke4 7. Kc4 with the threat of Kc5, b5 and Nc6 and the black king must go back
5...f5 6. g3 g5 7. f3 (keeping under control e4) and white should hold

Of course, in many variations you can sacrifice the knight for a pawn on the king side, as the a pawn will not win.
I can (and possibly am) be absolutely wrong in my judgment, but my inner feeling suggests to me that after 1...Kf8 2. b4 should hold

Playing 1...a5 is also not an easy line to see a clear win for black.

The nasty thing about chess is that it all hinges onto very small little details.

Thanks again for contributing.

Lyudmil

[Kyodai]

Karlos example is not just different - but in fact totally different from the
basic position in Stoltz-Kashdan.

Of course due to the blocked pawns in the centre d4/d5.

(Block a few more pawns for white on the same color as the bishop and
the knight will be superior - of course... we have shown positions like that
also)

karlo responded to Bob's words and here are the words:

"Bishop and knight are not "near equal" in an endgame with pawns on both wings. Bishop is clearly superior there as it can reach squares in one move that take a knight 2-3 (or more) moves to reach."

Cases when there are pawns on both wings and there are also blocked pawns are common in chess so practically it is not correct to say that the bishop is better when there are pawns on both wings and chess players or programmers may do mistakes in their evaluation function because of believing in this principle.

Uri

Yes, but how did karlo describe his example...............>

"Take for example the position from the famous Saidy vs Fischer (1964) game. Semi open position, pawns on both wings, no blocked pawns.

NO BLOCKED PAWNS..... and still gives a position with d4/d5.....

[Lyudmil Tsvetkov]

Cases when there are pawns on both wings and there are also blocked pawns are common in chess so practically it is not correct to say that the bishop is better when there are pawns on both wings and chess players or programmers may do mistakes in their evaluation function because of believing in this principle.

Uri

Hi Uri.
Just for the sake of precision.
I think that whenever there are pawns on both wings, the bishop will be superior, simply because it will move quicker than the knight for the purposes of attacking enemy pawns, defending own pawns, etc. Having pawns in the center (still belonging to one of the sides) will not change things very much, unless a central pawn is the only available pawn on that side, as then play will be conducted almost as being conducted on one and the same side. Pawns being blocked is another thing, favouring the knight in the general case, and blocked pawns for the bishop side on squares the colour of the bishop is still another thing, also favouring the knight. The number of pawns overall would also be of importance, simply because bishops become more mobile relative to knights with decreasing number of pawns. The ability to construct a fortress through which the enemy king can not pass is still another thing.
But you know that certainly better than me, I am posting this just for the sake of precision.

Lyudmil

[peter]

The first variation black won but just waiting one move for white with f4
and then using the Knight for tactics looked sufficient for a draw.

That's a summary where I ended up keeping the black bishop on d7 and
actually start with 1-f6 - the problem being to find the ultimate setup for
black.

Hi again, Sune!

I did try your f6 as 1... and as 5... but didn't find a way do avoid White's fortress for sure.
Here's one more example of a win against Houdini with 6MOB, but there might be still some suboptimal white moves in the variant, beginning e.g. with 7.f4?!

1... Kf8 2. Kf1 Ke7 3. Ke2 Kd6 4. Kd3 Kd5 5. h4 h5 6. g3 f6 7. f4 Be8 8. Ke3 a6
9. Nf5 g6 10. Nd4 Kc5 11. Kd3 Kb4 12. Kc2 Bd7 13. Kb2 Bh3 14. Kc2 Bg2 15. Kb2
Kc5 16. Kc3 Kd5 17. f5 Be4 18. fxg6 Bxg6 19. b4 Ke4 20. Ne6 Kf3 21. Nc5 Kxg3
22. Nxa6 Kxh4 23. Nc5 Be8 24. Ne4 f5 25. Nd6 Bg6 26. Kd4 f4 27. b5 f3 28. b6 f2
29. b7 f1=Q 30. b8=Q Kh3 0-1

The problem is, engines are not good at fortresses, neither at building nor at avoding them, so they can be trapped by such positions always.
So at least we'll not be able to proove Black's win by engines' evals nor by variants winning against unsupported engines.
Regardless of this one more winning variant I'm more and more convinced that White has the much more difficult ending to play but should be able to hold the draw at best defence,

[peter]

Here's one more example of a win against Houdini with 6MOB, but there might be still some suboptimal white moves in the variant, beginning e.g. with 7.f4?!


1... Kf8 2. Kf1 Ke7 3. Ke2 Kd6 4. Kd3 Kd5 5. h4 h5 6. g3 f6 7. f4 Be8 8. Ke3 a6
9. Nf5 g6 10. Nd4 Kc5 11. Kd3 Kb4 12. Kc2 Bd7 13. Kb2 Bh3 14. Kc2 Bg2 15. Kb2
Kc5 16. Kc3 Kd5 17. f5 Be4 18. fxg6 Bxg6 19. b4 Ke4 20. Ne6 Kf3 21. Nc5 Kxg3
22. Nxa6 Kxh4 23. Nc5 Be8 24. Ne4 f5 25. Nd6 Bg6 26. Kd4 f4 27. b5 f3 28. b6 f2
29. b7 f1=Q 30. b8=Q Kh3 0-1

but one more time Stockfish is best at backward solving this variant, keeping the winning eval in hash.
Here is Houdini 3 with 6 MOB, 6 cores and 4GB hash after backward solving the variant up to depth 27 at each ply after 7.f4 (at least its learning .dat- file has grown up to about 1MB so far )


8/p2b2p1/5p2/3k3p/3N1P1P/1P1K2P1/8/8 b - - 0 1

Analysis by Houdini 3 Pro x640:

7...g6 8.Ke3 Kc5 9.Ke4 Kb4 10.Kd3 a6 11.Kc2 Lg4 12.Kd2 Lc8 13.Kc2 Ld7 14.Kd2 Lh3 15.f5 Lxf5 16.Sxf5 gxf5 17.Kc2 a5 18.Kb2 Kb5 19.Ka3
=/+ (-0.65) Tiefe: 30/55 00:01:46 1345mN, tb=5



and here Stockfish using Gaviota's 5MOB and same terms as for everything else:


8/p2b2p1/5p2/3k3p/3N1P1P/1P1K2P1/8/8 b - - 0 1

Analysis by Stockfish 2.2.1 ofry 64bit SSE4.2T:

7...Le8 8.Ke3 a6 9.Kd3 Lg6+ 10.f5 Lh7 11.Ke3 Lg8 12.b4 Lf7 13.b5 axb5 14.Sxb5 Ke5 15.Sd4 Ld5 16.Kd3 La2 17.Ke3 Lc4 18.Sc2 Lb5 19.Sd4 Ld7 20.Sf3+ Kxf5 21.Sd4+ Kg4 22.Kf2 Lf5 23.Sb5 Le6 24.Sd4 Ld7 25.Se2 Lc6 26.Sd4 Ld5 27.Se2 Lc4 28.Sd4 Ld5
-+ (-1.96) Tiefe: 44/3 00:03:02 89583kN


And here the both of them just one ply backward again, just before 7.f4


8/p2b2p1/5p2/3k3p/3N3P/1P1K2P1/5P2/8 w - - 0 1

Analysis by Houdini 3 Pro x640:

7.Sf3 Lb5+ 8.Ke3 Le8 9.Kd3 Ld7 10.Sd4 a6 11.Sf3 Lf5+ 12.Kc3 Lg6 13.b4 Le8 14.Sd4 Ld7 15.Sb3 Lc6 16.Sd4 Ld7 17.Sb3
= (-0.27) Tiefe: 29/57 00:05:47 4381mN


8/p2b2p1/5p2/3k3p/3N3P/1P1K2P1/5P2/8 w - - 0 1

Analysis by Stockfish 2.2.1 ofry 64bit SSE4.2T:

7.Ke3 Kc5 8.Kd3 a6 9.Kc3 Kd5 10.Kd3 Lg4 11.Ke3 Kc5 12.f3 Ld7 13.Kd3 g5 14.f4 gxh4 15.gxh4 a5 16.Ke3 Kd5 17.Se2 Lf5 18.Sg3 Lg6 19.Se2 Kc5 20.Sd4 Le8 21.Se2 Lf7 22.Sd4 Ld5 23.Kd3 Kb4 24.Kc2 Le4+ 25.Kd2 Kc5 26.Ke3 Lg6 27.Se2 Lf7
=/+ (-0.49) Tiefe: 40/53 00:04:21 207mN, tb=3

[Kyodai]

Thanks! here some variations aimed for the Karsten Barrier



[pgn][Event "Bishop Superiority VIII"]
[Site "The Hague"]
[Date "1928.??.??"]
[Round "?"]
[White "Stoltz"]
[Black "Kashdan"]
[Result "0-1"]
[Annotator "Larsson"]
[SetUp "1"]
[FEN "6k1/p2b1ppp/8/8/3N4/1P5P/5PP1/6K1 b - - 0 1"]
[PlyCount "59"]
[EventDate "1928.??.??"]

{This position is from Stoltz-Kashdan 1928. A celebrated ending where the
existence of pawns on both wings on an open board emphasizes the bishop's
superiority over the knight.} 1... Kf8 2. Kf1 Ke7 3. Ke2 Kd6 4. Kd3 Kd5 5. h4
h5 (5... Bc8 {Black's bishop helps push the white king backwards, enhancing
the scope of his own king.} 6. f3 Ba6+ 7. Ke3 Kc5 8. Nc2 Bf1 (8... f6 9. Kd2 g6
10. Ke3 Bc8 11. Kd2 h5 12. Ke3 Bd7 13. g3 a5 14. Nd4 Bc8 15. Kd3 Kd5 16. Kc3
Bd7) 9. g3 Ba6 10. Nd4 Bb7 11. Kd3 Kb4 12. Kc2 Bd5 13. Kb2 g6 14. Kc2 a6 15.
Kb2 {Black is still far away from victory according to Averbakh in his famous
classic series on the endgame.} Kc5 16. Kc3 Kd6 17. Kd3 f6 18. Kc3 h6 19. Kd3
h5 20. Kc3 Ke5 21. Kd3 Bb7 22. Ke3 Bc8 23. Kd3 Bd7 24. Ke3 Kd5 25. Kd3 g5 $17 {
compare this position with the one after 7.f3 Be8 8.Ke3 g5 9.Kd3 a6 in the
Hauptvariante} 26. Ke3 Be8 27. Kd3 Bg6+ 28. Ke3 gxh4 29. gxh4 Kc5 30. f4 Bb1
31. Ne6+ Kb4 32. Kd2 Kxb3 33. Nc5+ Kc4 34. Nxa6 Kd4 35. Nc7 Bg6 $19) 6. g3 f6
7. f4 (7. f3 Be8 8. Ke3 (8. Kc3 g5 9. Ne2 a6 10. Kd2 Bb5 11. Ng1 gxh4 12. gxh4
Kd4 13. Nh3 Bc6 14. Ng1 Bd5 $19) 8... g5 9. Kd3 a6 {compare this variation
with the Averbakh's line in A1 after 25.-g5} 10. Kc3 (10. Ke3 Bg6 11. Ne2 Kc5
12. Kd2 Kb4 13. Nd4 Bh7 14. f4 gxh4 15. gxh4 Be4 16. Ke3 Bd5 $19) (10. f4 gxh4
11. gxh4 Bg6+ 12. Ke3 Kc5 13. Ne6+ Kb4 14. Kd2 a5 15. Nd4 Be4 16. Ke3 Bd5 17.
Kd3 Bxb3 18. Nc6+ Ka4 19. Kc3 Be6 $19) 10... gxh4 11. gxh4 Ke5 $19) 7... Be8 8.
Ke3 a6 9. Nf5 g6 10. Nd4 Kc5 11. Kd3 Kb4 12. Kc2 Bd7 13. Kb2 Bh3 14. Kc2 Bg2
15. Kb2 Kc5 16. Kc3 Kd5 17. f5 Be4 18. fxg6 Bxg6 19. b4 Ke4 20. Ne6 Kf3 21. Nc5
Kxg3 22. Nxa6 Kxh4 23. Nc5 Be8 24. Ne4 f5 25. Nd6 Bg6 26. Kd4 f4 27. b5 f3 28.
b6 f2 29. b7 f1=Q 30. b8=Q Kh3 0-1[/pgn]

[peter]

Just a quick reply, the thing's getting better and better.

Thanks for taking my variant into your .pgn, but main variant should be from your 7.f3!? or from 7.Sf3!? instead of f4?

E.g. what about

1... Kf8 2. Kf1 Ke7 3. Ke2 Kd6 4. Kd3 Kd5 5. h4 h5 6. g3 f6 7. f3 Be8 8. Ke3 g5
9. f4 g4 10. Kd3 a6 11. f5 Bd7 12. Ke3 a5 13. Kd3 Ke5 14. Ke3 Bxf5 15. Nc6+
neither this nor 13....Kc5 works for me, even if 9.f4 instead of your 9.Kd3 seems to be loss of a tempo, that might save from zugzwang again.

As for your outplaying of Karsten's line, I miss the thematic try ...b4, latest at 22..., but even 23...Ne2 instead of 23...Kd3 might hold, don't you think so?

Maybe we should change to email not to overcharge this thread here?
If you want to, give me an adress by private message, I don't think, we'd be through too soon.