This is a compilation of a large set of positions posted by Sune Larsson at the CCC
computer chess forum. Try your silicon and find out that there is still hope for human kind as these
are the strategic kind of positions computers still have a hard time. You even may find out you
will do better than the silicon.
Enjoy!
CM8
Larsson
Bishop ending
2001
This position arose from a game of mine vs CM8K. Black is on the move and has an easy draw. How do your programs evaluate this position?
White to move and make it for the draw. Normal play like 1.Rf5? Kg6! 2.Rxe5 Rh8! followed by Rh2 gives the full point to black. 1.Re6? makes no difference after 1.-Rh8!. But the pawnchain carries some hidden secrets...To be honest, this is more a test of one's own limitations of the mind. I think it's too difficult for any chessprogram (but who knows?). Then again, it's possible to test the programs when 1/3, 1/2 or 2/3 of the solution is performed. A human can solve it by trial and error. If normal play doesn't work, one is forced to think in new patterns...
Here is an example of a definitive no-no. This position happened in Winter(!) - Capablanca, Hastings 1922. Here white was a bit careless and played 1.Nd5? after 1.-g5! 2.Nxf6+ Qxf6 3.Bg3 Bg4 4.h3 Bxf3 5.Qxf3 Qxf3 6.gxf3, the proud bishop was reduced tp a simple pawn. This preacher never saw the daylight anymore... Test1: 1.Nd5? is a clear no-no and should be evaluated with penalty points. Test2: If 1.Nd5? is faced, your program should give definite +scores to the line above. And choose this path in play.
Only two moves are required to immobilize the black knight and pave the way for the victorious march of the white king to the deserted queenside: 1.g5! Bc8 2.g4! and black's position is ripe for resignation. Test: The above move order should be rewarded with big +scores for white.
True inspiration by Amelung. White is on the edge of losing and only a real desperado can save the day. Go for it farmer boy!! Test: Some programs get it other's don't...
This is old Swedish chess history. White is in deep trouble. Knight moves to a1 or c2 will just lose the lonely one. But the knight has a braver possibility! 1.Na5!! actually saved the draw for white. After 1.-bxa5 2.bxa5 black has the wrong bishop and white's king can just make it to a1, if black goes for both the white a-pawns. Test: This is a difficult one for chess programs and presupposes extended knowledge about a/h pawn and the wrong bishop.
This is from Alekhine-Johner, Zurich 1934. A pawn down, black is about to manouvre his knight to e5, with excellent prospects. On the other hand the white bishop is in the need of more space, free lines to feel alive again. So Alekhine took a deep breath, played 1.e5! and said: "I need some more air." Well it could have been that way anyhow...;) Test: A few of the positions sent were rather easy, so for a change some real stuff to chew on...If your program plays 1.e5! you really have something going...
(The end could be 4.-Nh7 5.Bd5+ Kh8 6.Qg6 Qd8 7.d7)
1-0
Larsen
Study
The Arrow's Point
1932
The name of the author sure gives associations... But it's a common name and the initial + year tells us that it couldn't be this Larsen nor that Larsen...; ) This one carries an indian theme. White has a nice and sharp arrow, formed by the pawns c2,d3,e4,f5,h7. On the other hand he's facing defeat since the black f2 pawn is about to promote. Looks like there is no decent way to stop this runner. But the arrow carries some poison and with an intriguing manouvre white is aiming for peace. And succeeds... Test: White moves for the draw. Fire off your arrow and show us your peace evals! The smoke signals are further down. ;)
Hm, my sources are giving me some trouble here... First we had a beautiful study by Schroeder, who proved not to be *that* Rebel-Schroeder. :) And now we are facing a very clean finish by Meyer. Which Meyer? Kahlen? Believe it or not...This ending is very simple, clean and with a distinct finish. A bishop ending (same colors), 4-3 pawns, but the position is blocked and it's difficult to make any progress. Something must be there because black's bishop looks just like another pawn... Test: White to move and win. Good evals for white, while just moving around the pieces, are not good enough. Can your program actually play out this position for a win? The distinct kill further down.
Now something to chew on for you fast drawers! No compromises any more. This is the real stuff! And some of the possible endings KQ vs KRP have been proved forced mate. Test: White to move and win. Nothing else counts. The ultimate solution further down.
Test your silicon
