Start your engines

I recently fulfilled a promise made a couple of months ago and sent off 34 euros to IM and chess programmer Vasik Rajlich for a copy of Rybka 2.3. I had hoped to be writing that this new release exceeds all expectations but this has not proven to be so, at least as far as my copy is concerned. The problem is that Rybka 2.3 runs much more slowly than other programs and this not only affects its playing strength but also causes strange lag effects in the board display. A posting on the Rybka Forum indicates a hashtable bug that is due to be corrected very shortly through a beta release. That's fine with me as long as it doesn't cost anything.

In order to compare Rybka 2.3 with my other playing programs I decided to assign a position from the game Szabo-Bisguier, Buenos Aires 1955, which is shown in the diagram. Szabo concluded matters by force in only three more moves: 1.Bxh7+ Kxh7 2.Qh3+ Kg8 3.Rg4! and Black resigned because he cannot prevent Rh4-h8 mate. His king is blocked in by his own pieces and any move of the f-pawn is met by g5-g6. This is a good test position for humans because the key move (3.Rg4) is slightly counterintuitive; it is usually the queen that is on the sharp end of h-file attacks.

Unlike human players, computers are tactically perfect within a short range and are thought to have few biases so one would expect them to solve this position fairly quickly. I have a number of UCI engines that run under the Fritz interface and I gave them all the same position in turn. Of course they all solved it but they took varying amounts of time to do so. Here are the results:

Rybka 1.0 beta: 2 seconds
Fritz 8: 17 seconds
Pro Deo 1.1: 60 seconds
Fruit 2.1: 62 seconds
Toga II 1.2.1: 68 seconds
Rybka 2.3: 72 seconds
Crafty 19.15: 189 seconds

Yes, a little joke at the end... I wonder what a match between Rybka 1.0 and Crafty 19.15 would look like? In any case I think you can sense my concern over spending 34 euros on Rybka: it was outperformed by the free programs Pro Deo, Fruit and Toga, and badly outperformed by Fritz 8. They were all running on an AMD Athlon 2600+ with 64 MB hashtables.

Any and all feedback is welcome on this topic.

Zwischenzug

Last time we looked at some interesting play from the 2006 B.C. Championship and today there is more on the menu. The game McLaren-Wu reached the first diagram after several moves of a Fianchetto Grunfeld. The position is absolutely symmetrical with the important proviso that it is White's move. It has been understood for many years that maintaining symmetry can be an effective equalising strategy for Black as long as he realises that a moment will come when the symmetry must be broken. In the present game this moment arrives after White's next move:
9.Ne5
If Black now plays 9...Ne4?! there follows 10.Nxc6 and he cannot continue with 10...Nxc3 because of the line 11.Nxd8 (also possible is 11.Nxe7+ followed by 12.bxc3 but capturing the queen is stronger) 11...Nxd1 12.Nxf7 Nxf2? (the lesser evil was 12...Rxf7 13.Rxd1 leaving White a solid pawn ahead) 13.Nh6+ Bxh6 14.Bxh6 and White wins the exchange after 14...Nh3+ 15.Bxh3 Bxh3 16.Rxf8+, incidentally breaking the symmetry in decisive fashion. Therefore Black must find a more effective way. As in the Exchange Slav a key question is whether capturing on c6 and transforming Black's b-pawn into a backward c-pawn constitutes a positional threat. On the one hand this pawn could be a handy target for White's pieces; on the other hand it might be difficult to attack or even blockade the pawn without allowing counterplay on the central dark squares. These ideas would take concrete form in the sample line 9...Bf5!? 10.Nxc6 bxc6 11.Bf4 Nd7!? threatening to obtain complete freedom with 12...e5. In the game Black apparently decided that 10.Nxc6 was a threat so he took the decision to capture on e5 himself:
9...Nxe5 10.dxe5 Ng4
Countering White's attack on the d-pawn with his own attack on the front e-pawn.
11.Nxd5
Also possible was 11.Qxd5 since exchanging queens loses a pawn for Black and 11...Nxe5 12.Qc5! gives White a small but definite advantage on account of his more active pieces, the disparity between the two light-squared bishops being especially striking.
11...e6?! (second diagram)
It appears that Black did not like the variation 11...Nxe5 12.Bg5! Nc6 13.Rc1 (threatening 14.Rxc6!) but in fact the resource 13...Be6! would solve most of his problems. Now 14.Rxc6 is met by 14...Bxd5! 15.Rc2 Bxg2 16.Kxg2 Qa5 17.Bxe7 Rfe8 18.Ba3 Rad8 and Black's active pieces give him excellent compensation for the pawn. In the actual game White now played 12.Nf4?! and after 12...Nxe5 13.Qb3 Qa5! Black had largely equalised the chances. What interests us here is what would happen after the desperado move
12.Nf6+!?
At first glance this looks pointless because multiple captures on f6 will leave an almost symmetrical position with little opportunity for either side to play for an advantage. This is seen clearly in the lines 12...Nxf6 13.exf6 Bxf6 or 12...Nxf6 13.Qxd8 Rxd8 14.exf6 Bxf6. But in the latter variation White has an effective zwischenzug:
12...Nxf6 13.Qxd8 Rxd8 14.Bg5!
Black's knight is pinned against his rook, so he must acquiesce to the loss of a pawn. The main variations are in White's favour:
a) 14...Bd7 15.exf6 h6 (15...Bh8 and 15...Bf8 are both met by 16.Rac1) 16.Bf4 Bxf6 17.Bxb7, winning the exchange;
b) 14...h6 15.Bxf6 Bxf6 16.exf6 e5!? (not 16...Rd2 17.Rfd1 Rc2 18.Rd8+ Kh7 19.Rf8 Rc7 20.Rad1 and wins) 17.Rfd1 Bg4 18.Kf1! Rab8 19.Rac1 b6 20.Ke1! Rxd1+ 21.Kxd1 Rd8+ 22.Ke1 Rc8 23.Rxc8+ Bxc8 24.Bd5! and Black is in great difficulties because of his weak f-pawn.

Let's peel back to the position after 12.Nf6+!? and take a look at the alternative:
12...Bxf6
This cannot be satisfactory because it gives up the natural defender of the dark squares around Black's king.
13.exf6 Qb6!?
Or 13...Qxf6 14.Qa4 Qf5 (and not 14...Ne5 15.Rd1 Rd8 16.Rxd8+ Qxd8 17.Bf4 Nc6 18.Rd1 Qe7 19.Bd6 Qe8 20.Qh4! and White has a winning attack) 15.Qh4 e5 16.Rd1 Qe6 17.Bh6 Re8 18.Rd2 with advantage to White.
14.Qa4 e5 15.b3 Nxf6 16.Be3 Qe6 17.Rfd1 Re8
Stronger than 17...Bd7 18.Qh4!
18.Bg5 Kg7 19.Rac1
White has an obvious advantage. Black's queenside remains undeveloped and under pressure from White's active pieces.

The actual game ended in a draw after some further adventures.

Fighting for the draw

There were some interesting moments in the game Wu-Pechisker, British Columbia Championship 2006. In the first diagrammed position Black is to play his 52nd move. He is a pawn up and the natural question is whether this gives him any real winning chances. In effect there is only one meaningful plan: create a passed pawn and try to queen it. Before executing this plan Black could have tried improving the position of his other pieces but in the end he decided not to wait:

52...d5 53.exd5+ Kxd5
White must now decide on a plan of defence. The most natural move is 54.Rg4 aiming for a blockade on the square e4. After 54...Ra8 55.Nd2 Ra3+ 56.Ke2 Be7 57.Nc4 Ra2+ 58.Kd3 Bc5 59.Re4 Black appears to have no way of breaking White's blockade; for example 59...Bd4 60.Rg4 Rh2 61.Ne3+ Kc5 (if 61...Bxe3 62.Kxe3 Rh3+ 63.Ke2 and Black cannot win) 62.Nf5 Rh3+ 63.Ke4 Rh1 64.Nxd4 exd4 65.Rg8 Re1+ 66.Kd3 and again Black can make no progress. White must have seen some problems with this plan because instead he chose the radical
54.Nxe5!??
sacrificing his knight to reach the theoretically drawn ending of rook vs. rook and bishop. I have given this move one exclamation mark for hubris and two question marks for impracticality. Unless White is completely certain of his defensive technique, taking on this endgame is a very risky enterprise.
54...Bxe5 55.Rh3 Ra8 56.Rh5 Ra3+ 57.Kc2 Rg3 58.Rf5 Ke4 59.Rh5 Rc3+ 60.Kd2 Rc6 61.Rh7 Bf4+ 62.Ke2 Rc2+ 63.Kd1 Rd2+ 64.Ke1 Rb2 65.Kd1 Be3
White's king has been driven to the edge of the board, and he must now play absolutely perfectly in order to survive.
66.Rd7!
The only move to draw; for example, 66.Re7+ Kd3 67.Rd7+ Bd4 68.Rc7 Ra2 69.Rc8 Rf2 70.Re8 Bf6 71.Re6 Bc3 72.Rd6+ Bd4 73.Re6 Rd2+ 74.Ke1 Ra2 75.Kf1 Rf2+ 76.Ke1 Rf5 77.Rd6 Rg5 and White must give up his rook to avoid immediate mate.
66...Rh2 67.Rd8 Bd4
(second diagram)
68.Kc1?
White has defended accurately until now, but this move is a mistake because the king will run into trouble in the restricted space on the queenside. White could maintain a defensible position with 68.Rf8!; for example, 68...Kd3 69.Ke1 Rg2 70.Rf3+! Be3 71.Rf8 and Black has not made any progress.
68...Kd3 69.Kb1 Rb2+ 70.Kc1 Ra2 71.Rb8 Rc2+?!
Black does not notice the winning manoeuvre 71...Rf2! 72.Kb1 Rf1+ 73.Ka2 Ra1+ 74.Kb3 Rb1+, picking up White's rook.
72.Kb1 Kd2?
A second error that allows White a clever drawing resource. It was not too late to get back on track with 72...Rf2 73.Rc8 ( 73.Rd8 Rb2+ 74.Kc1 Ra2 75.Rb8 Rf2 transposes to the previous note) 73...Rb2+ 74.Kc1 Ra2 75.Kd1 Rf2 76.Re8 Bf6 and wins as in the note to White's 66th move.

73.Rb4! Bc3 74.Rd4+!
With this rook sacrifice White forces stalemate, and so the players agreed to a draw here.
½-½

Lessons learned

Many chessplayers find losing a game to be a painful experience. All too often the scoresheet ends up falling behind a bureau where it is eventually swept up with some dustballs and thrown out, never to be reexamined or even seen again. This is an unfortunate outcome because a lost game is an opportunity to find out what sorts of errors we are making – provided we are brutally honest with ourselves. This information can then be used to design a plan of study aimed at reducing or eliminating those errors. This will definitely improve our play and may lead to dramatic rating gains. It might be said that there is no shame in losing a game of chess; if there is any shame it consists in not learning anything from the experience.

The first diagram shows a position from the famous game Bernstein-Capablanca, Moscow 1914. It is Black to play, and both sides have aimed for this position. For his part, Bernstein has won a pawn in apparent security. He sees that Capablanca cannot do any harm with 1...Qb1+ 2.Qf1 Rd1 because of the reply 3.Rc8+. Capablanca, on the other hand, has seen a winning shot based on White's weak back rank and undefended pieces: 1...Qb2!! This threatens both 2...Qxe2 and 2...Qxc3, so the only real defensive try is 2.Rc2. But on c2 the rook is now vulnerable to a double attack and Black wins with 2...Qb1+ 3.Qf1 Qxc2.

Every serious chessplayer is familiar with Capablanca's wonderful combination. His opponent Ossip Bernstein was probably unhappy to lose this game, but it must have been a great learning experience for him. I say this for two reasons. For one thing, he went on to become a very strong grandmaster and even drew a match with Alexander Alekhine while the latter was at the peak of his powers. But he obviously learned some more specific things as well.

Take a look at the second diagram, which shows a position from the game Kahn-Bernstein, Paris 1926. The lesson given by Capablanca on loose pieces, weak back ranks and deflection sacrifices has obviously had its effect, because Black found the very strong move 1...Rxf2! After the reply 2.Rxe6?! he finished the game with 2...Qe2!! There are two threats: 3...Qxe1+ and 3...Rf1+, and White cannot defend with 3.Bxf2 because of 3...Qxf2+ followed by mate. Instead of 2.Rxe6?! White can play 2.Qe4 but Black is still winning after 2...Bf5 3.Qd5 Qxd5 5.Rxd5 Rxg2+!

White in turn must have learned something from this loss, for he went on to co-author a classic book entitled The Art of the Checkmate.

Artillery

Positions in which all the minor pieces have been exchanged and the players are left with a queen and two rooks each were examined by Alex Angos in a fascinating 1982 monograph entitled Endgame Artillery. Such positions present an interesting blend of middlegame and endgame features. If there are no passed pawns or serious pawn weaknesses and the two kings are relatively safe the outcome is often an uneventful draw. But when there are significant imbalances the initiative becomes a key factor and the play can be extremely complicated. In this regard the classic encounters Rubinstein-Alekhine, Dresden 1926; Alekhine-Euwe, Nottingham 1936; and Korchnoi-Spassky, Belgrade (m/7) 1977/78 are recommended for further study.

In a recent team match I managed to win a pawn but at the same time had to give my lower-rated opponent dangerous kingside pressure. After some exertions I managed to exchange off the minor pieces and blunt the attack, arriving at the “endgame artillery” position shown in the first diagram. My opponent's last move had been 42...Reg8 and I could sense that he was now looking for a draw by repetition in the variation 43.Rg1 Re8 44.Re1 Reg8, etc. In unhappily typical fashion I had left myself five minutes for all the rest of my moves; meanwhile my opponent had nine minutes on his clock. I briefly considered acquiescing to the draw because the match score at that moment was 4-0 for our side with three games to go including this one. But I rejected this notion almost immediately – I was a pawn up and I wanted payback for the difficult defence I had conducted! On top of it all there was an interesting counterattacking line available, so without further ado I plunged ahead...

43.e4! fxe4
After 43...Rxg3 44.Rxf5+ Black is forced to play 44...R8g5 because 44...Kh7 loses immediately to 45.Rxh5+ while 44...R3g5 allows the simplifying operation 45.Rxg5 Rxg5 46.Rg1 Qe7 47.Qxg5+ Qxg5 48.Rxg5 Kxg5 49.Kg3 with an easily won pawn ending for White. After 44...R8g5 White must play 45.Re2 (definitely not 45.Rxg5? Qh3#!) and now there are two lines:
a) 45...Kg6 46.Qf4! R5g4 (the alternative 46...R3g4 47.Rxg5+ Rxg5 48.Rg2 leads to a queen ending that is very difficult for Black to defend) 47.Qb8! Rd3 48.Qd8! (taking control of key squares) 48...Rgg3 49.d5 Rh3+ 50.Kg2 Rhg3+ 51.Kf1 Rd1+ 52.Kf2 and Black is losing at least a rook;
b) 45...h4 46.Qf4 Kh5 47.Rxg5+ Rxg5 48.d5 Qd7 ( or 48...Qe5 49.Qxe5 Rxe5 50.d6 Re8 51.e5 and wins) 49.Rf2! and White will either mate, win material or promote a pawn (and then mate!)
44.Rfxe4
Stronger was 44.Rexe4; for example, 44...Qd7 45.Qe3! R8g6 46.Re8 Qg7 47.Rff8 Kh7 48.Re7 and wins.
44...Qf5 45.Rf4 Qd3?! (second diagram)
Perhaps Black was afraid for his king but in the circumstances he should not be the one forcing off the queens. More tenacious was 45...Qg6 although after 47.Re7! Rg7 48.Rxg7 Kxg7 49.Qe3! White should still win with proper technique.
46.Qxd3?!
The time factor... Black's last move had surprised me and I was forced to spend one of my remaining minutes seeing if I could win after exchanging queens. This left me with two minutes to play the rest of the game and no time to look at the alternative 46.Rf6+! which is in fact much stronger. The main line then is 46...R8g6 47.Rxg6+ Kxg6 48.Re6+ Kf7 49.Qxg5 Kxe6 50.Qe5+ Kd7 51.Qxb5+ Kc8 52.Qe8+ Kc7 53.Qe5+ Kd7 54.d5 Qf1 55.a4 with two passed pawns for White and no visible counterplay for Black.
46...cxd3 47.Rf6+ Kh7?!
This loses quickly. Black should have abandoned the attack on g3 and played 47...Kg7! After 48.Rf3 Rc8! 49.Rxd3 Rc2+ 50.Kh3 Rxb2 51.d5 Rf5 52.d6! (not 52.g4!? immediately because of 52...hxg4+ 53.Kxg4 Rf8 54.d6 Rg2+ 55.Kh3 Rg5 56.Re4 Rh8+ 57.Rh4 Rd8 58.Rg4 Rxg4 59.Kxg4 Kf6 60.Rd5 Ke6 61.Rxb5 Rxd6 and White cannot win) 52...Rff2 53.g4! Rh2+ 54.Kg3 Rbd2 and now White must find 55.Ree3! (stronger than 55.Red1 because it creates a safe passage for White's king) 55...Rdf2+ (or 55...hxg4+ 56.Ke3 Rxd3 57.Kxd3! Rg1 58.Kd2! Rg2+ 59.Kc3! and wins) 56.Ke4 Rxg4+ and now after 57.Kd5! it is clear that Black will be resigning soon. The problem with the text move 47...Kh7 is that White can immediately force off a pair of rooks. The resulting single rook ending can then be won with almost no thought, a pleasantly untroubling task when one has only two minutes left on the clock.
48.Re7+! R8g7
Much weaker is 48...Kh8? 49.Rh6#!
49.Rxg7+ Rxg7 50.Rf3 Rc7 51.Rxd3 Rc2+ 52.Kh3 Rxb2 53.d5 (third diagram) 53...Rc2 54.d6 Rc8 55.d7 Rd8 56.Kh4 Kg6 57.Rd5!
Winning the h-pawn and bringing White's g-pawn into play. The rest requires no comment.
57...Kf6 58.Kxh5 Ke6 59.Rd3 Rh8+ 60.Kg4 Rg8+ 61.Kf4 Rf8+ 62.Ke4 Rd8 63.g4 Rg8 64.d8Q Rxg4+ 65.Kf3

Black resigns 1-0