How far I've progressed you can judge from the games at the end of this piece. I do know that I feel a great difference in my willingness to go in for complications, and greater confidence in attempting to whip up an attack.
My aim here is to lay out what I see as the principal elements of my last two years' study which have led to these changes. In short, I offer a pattern (not a complete course!) of home study which, in my case, has resulted in a significantly sharper (more interesting? stronger?) style.
It is not my aim to consider how to calculate variations. Not that I don't agree this is an essential ability, but I do believe it is over-estimated in its own right. Practice is actually very easy: set up complicated positions, either from Winning Combinative Play or from games collections such as those of Tal, and analyse -- against the clock! The one rule you must adhere to is not to waste time. Go over lines 10 times if you want, but only if you are asking new questions at each repetition. Depth, accuracy and thoroughness will come with practice, as you compare what you worked out in your head with what is there on paper.
The point to understand is that no-one can calculate perfectly (not even the computers -- chess is too big for that) and so it is irrelevant whether you can calculate 10 moves deep or only 2: at some point you have to stop and make an assessment of the position.
Instead of asking: "does that really work? What am I overlooking?", you have to develop the intuition that lies behind a sacrifice.
So, after this lengthy introduction, I'm concerned with a related pair of questions:
- How can we raise the courage to play sacrificial moves?
- And, how can we improve the likelihood of getting the chance?
Botvinnik claimed to have learnt about Queen+Knight co-operation from Capablanca. To see what he learnt, consider the following finish of Silwa-Botvinnik, Moscow 1956 (for full analysis see Botvinnik on the Endgame (1985)):
White has more than enough material compensation for the knight,
and the black king is insecure. Therefore, attempts to win the
Q-side pawns, which lift the mate threat, would allow white to
escape with perpetual. The winning plan is extra-ordinary. By
sacrificing his last two pawns, black opens up the white king and
forces 'mate, due to the co-operation of his queen and knight.
41... Qc6 42.Qg1 Qxc3 43.a4 g4!
(43... Qxd4 44.a5 Nxc2 45.a6 would be harder.)
44.hxg4+ Kg5! 45.a5 h3!
There goes the last pawn, opening up the long diagonal. Because both 46.gxh3 Qc6+ 47.Kh2 Qf3! and 46.g3 Qc6+ 47.Kh2 Qxc2+ 48.Kh1 Nxg4 quickly lead to 'mate -- White resigned.
And now see the finish of V.Anand-V.Topalov (Wijk Aan Zee, 1998):
(Acceptance is forced. Although all white's pieces are on the back rank, the weak dark-squares are decisive. If 26...Nxd5 27.Nf5 Qf8 28.Nh6+ Kg7 29.Qd4+ Kg6 30.h4 etc.)
27.Qd4+ Kf8 (27...Kg8 28.Bh6 or 27...Kg6 28.h4) 28.Bh6+ Ke8 29.Re1 1-0
(For we mortals: 29...Kd7 30.Rxe7+! When it's B+Q vs K+others. e.g. 30...Kxe7 31.Qe4+ Be6 32.Qxh7+ Bf7 33.Bg5+ Kf8 34.Qh8+ Bg8 35.Qh6+ Ke8 36.Qg6+ Bf7 37.Qe4+ Kf8 38.Bh6+ Kg8 39.Qd4 forces the win of the queen, with the advance of the g and h-pawns winning the game.)
Amazing stuff, but note that after the demolition sacrifice Anand finished playing Q+B vs K+scattered defence; as the black K could never escape to d7 (because of Qf5+ and Bg5 forcing the win of the queen) Anand could judge that a 'mating position was forceable (of course, he may have calculated it too!). The choice is simple. Either you can calculate perfectly (me neither), or you must understand piece co-operation enough to make an accurate assessment of such positions.
Studying simple positions will help us gain the courage to make the initial sacrifice -- trusting our assessment will remove the worry of "overlooking something"; if your assessment is right, you can find the good moves when you get there.
The thematic move here is 1.Bxh7+, but before
playing it, there are FOUR variations that need to be considered:
1) 1... Kh8 Worth remembering that sacrifices do not have to be accepted. In this position 2.Bd3 is possible, winning a pawn.
and after 1... Kxh7 2.Ng5+
2) 2... Kg8 3.Qh5 Re8 4.Qxf7+ Kh8 5.Qh5+ Kg8 6.Qh7+ Kf8 7.Qh8+ Ke7 8.Qxg7 'mate. A typical procedure, driving the king to its fate.
3) 2... Kh6 3.Nxf7+ xd8 wins the queen.
4) 2... Kg6 3.h4 (sometimes 3.Qg4 is strongest) Rh8 4.h5+! Rxh5 5.Qd3+ f5 6.exf6+ Kxf6 7.Qf3+ Ke7 8.Qf7+ Kd6 9.Qxh5 and wins (10.Nf7+ is threatened too).
[5) 2... Kh8 is never really an option 3.Qh5+ Kg8 4.Qh7 'mate.]
Once mastered, this is a powerful weapon, ready to be used whenever possible. A light-hearted example:
"Almost any decent player would make this sacrifice instinctively, without bothering to calculate the consequences." CHERNEV
9... Kxh7 10.Ng5+ Kg6 11.h4 Nxd4 12.Qg4 f5 13.h5+ Kh6 14.Nxe6+ g5 15.hxg6 en passant, 'mate!
This is, of course, an example of "castling into it". What is to be done when your opponent tries his/her level best to avoid falling for such basic attacking motifs? Just how can we force an opponent into accepting the pre-conditions for a combination?
Consider an example of Capablanca's, where a combination is used to set up the pre-conditions for a Classic Bishop Sacrifice.
Black's last move prepares to develop his last piece, but allows
JRC an attractive combination. 11...h6 would have avoided
immediate danger. Note that white's pieces are right for the
Classic Bishop Sacrifice, but black's knights are irritatingly
covering f6. Capablanca uses one decoy sacrifice and one exchange
sacrifice to clear his way.
12.d5! Nxd5 13.Bxe7 Nxe7
step 1: white moves the N from f6, making the sacrifice possible:
14.Bxh7+! Kxh7 15.Ng5+ Kg8 16.Rxd7!
step 2: white removes the other N before it goes to f6, e.g. 16.Qh5? Nf6 and black defends.
16...Qxd7 17.Qh5 Rd8 18.Qxf7+ Kh8 19.h4
(not 19.Nh5 Qd1+ 20.Rxd1 Rxd1 'mate!)
19...Nf5 20.Nh5 Qe8 21.Nf6! 1-0
(because 21...Nd6 22.Nxe8 Nxf7 23.Nxf7+ Kg8 24.Nxd8 Bd7 25.Nxe6 Rxe8 26.Ng5 leaves white with a won endgame.)
Studying games such as these (more from Capablanca in a later piece), and other masters like Tal, helps us understand the pre- conditions behind a finishing combination, and how to set them up.
I want to illustrate all this with some examples of my own play. One from last year:
This is a standard position from the French Defence, Tarrasch
Variation. Some books may tell you that the French stresses the
"defensive" aspect of Defence, whereas I have always thought of the
opening as a counter-attacking one, and particularly attractive to
those aspiring to be "combinative players".
In particular, this position is wonderfully unbalanced. Note Black's attack on d4, the half-open f-file, and the free piece play. The battle is usually about control of e5, because if black manages to play ...e5 his game generally becomes very free -- taking over the central squares, freeing his Bc8, and also creating a passed d-pawn (not always a trivial factor!). The danger for black is in allowing white to establish a bind on e5 and the Q-side, which can make white's wins look effortless.
12.Nc3 O-O 13.a3 Bd7 14.b4 Kh8
White is making progress towards a Q-side bind, but leaving the centre undefended -- so black strikes there.
15.Na4 Qc7 16.Bb2 e5! 17.dxe5 Nxe5 18.Nxe5 Bxe5 19.Bxe5 Qxe5 20.Nc5
This is where the fact that I play the French all the time played
its role -- I knew that White had handed the advantage
over to Black. In fact, I am convinced that after ...e5 Black
manages to take control of these kinds of positions. Therefore, I
spent a long time looking for a decisive continuation. In short, I
wanted a combinative attack.
45 minutes is a long time to spend on one move, especially when it's as obvious as the next:
20...Ng4 21.g3 Nxf2!
That wasn't so obvious, but if I hadn't played it then 20...Ng4 would have been a bit pointless.
22.Rxf2 Rxf2 23.Kxf2 Bh3!!
I found this quiet move by a combination of remembered lessons and frustration at not finding anything by direct calculation. A quiet move like this is hard to find, especially with ...Qd4+ ...Qb2+ or ...Rf8+ as viable alternatives. However, I couldn't find a decent follow-up to any of these moves, and instead remembered: firstly, "the threat is stronger than the execution", and secondly, Vukovic has an extended discussion on weak square complexes, and using them to force checkmate. White's king already looks draughty on the dark squares, but how about the light ones? Therefore 23...Bh3, and I found this during my long think on move 20!
Stronger is 24.Ra2, though after 24...Rf8+ 25.Kg1 Qe3+ 26.Kh1 b6 Black forces a winning ending after, e.g., 27.Nb3 Rf2 28.Rxf2 Qxf2 29.Be4! Bg4! 30.Qg1 Bf3+ 31.Bxf3 Qxf3+ 32.Qg2 Qxb3.
24...Rf8+ 25.Kg1 Qe3+ 26.Kh1 Qf2 27.Qxd5 Qxe2 28.Rg1 Bg4 29.0-1
This kind of combination is standard for the French player. Another example of the same theme:
By transposition we've reached a theoretical position of the French Two Knights, where white gives up his central pawns for piece control of e5. In earlier games I struggled for a plan in these positions, but these days I'm armed with 11...Qe8, Watson's recommendation.
Best for white is 12.Nb5 when Qe7 gives black a decent game. Black's move is also a mistake: the theoretical 12...Nh5! forces the win of material, e.g. 13.Nxc6 Nxf4 14.Ne5 Qh5 15.Nf3 Nxg2 16.Kxg2 Qxf3+ etc. Other choices for white cost even more.
13.Bxe5 Qh5 14.Rae1 Ng4 15.Bg3
Again, that f2 square -- Black has to strike now or h3 follows.
16.Rxf2 Bxf2+ 17.Bxf2 Qxh2+ 18.Kf1 Bd7 19.Qg5 Qh1+ 20.Ke2 Qxg2 21.Qh4
I had overlooked this move. The threat to h7 allows Rg1, forcing an endgame with black's 3Ps vs the N. I had assumed such positions would offer both sides winning chances when offering the sacrifice on move 15 -- the kind of assessment I would not have trusted myself to make a couple of years ago -- and my opponent proceeds to misplay the endgame badly.
21...h6 (Playing the pawn to a dark square -- observe the Bd3) 22.Rg1 Qxf2+ 23.Qxf2 Nxf2 24.Kxf2 Rf8+ 25.Ke2? g5 26.Rf1? Rxf1 27.Kxf1 Kg7 28.Kg2 Kf6 29.Kg3 h5 (White's passive play [25.Ke3-d4!] and exchange of the rooks, have made the pawns a winning team.) 30.a4 h4+ 31.Kf3 Bc6 32.Kf2 g4 33.a5 g3+ 34.Kg2 d4+ 35.Ne4+ Ke5 36.Kf3 g2 37.Kxg2 Bxe4+ 38.Kh3 Bxd3 39.cxd3 Kf4 40.Kxh4 Ke3 41.0-1
An interesting system against the King's Indian: White simply captures the e5-pawn and sets Black some problems. I had assumed this was quite rare, but looking down the table I.George vs. D.Grossett had the same position -- 9...Nc6? 10.Nd5! won the exchange quickly. 9...Nbd7? is no good, 9...Na6!? is better, and 9...Re8! is probably best.
9...c6!? 10.Nxe5 Re8 11.O-O-O! Na6
If 11...Rxe5 12.Rd8+ Re8 13.Bxf6; 12...Ne8 13.Rxc8 etc. The whole point of the system is that White (almost) gets away with a blatant pawn grab!
12.f4 h6 13.Bh4 g5 14.fxg5 hxg5 15.Bg3 Nc5
Black tries to recover the e4 pawn. Although 15...Be6 doesn't
regain the material, it would have avoided the following
16.Nxf7! (the KB2 square bears the brunt of many sacrifices) Kxf7 17.e5 Ng4 18.Rd4 Nxe5 19.Bh5+ Kg8
A wise decision -- 19...Ng6 20.Rf1+ Bf6 21.Rd6 Re6 22.Rxe6 Bxe6 23.Be5 Nd7 24.Bxf6 Nxf6 25.Ne4 and, 19...Kf8 20.Bxe8 Kxe8 21.Re1 Nd7 22.Rxd7 Bxd7 23.Bxe5 leave White with his extra material and a simplified ending.
20.Bxe8 Ned3+ 21.Rxd3 Nxd3+ 22.Kd2 Bf5 23.Rf1 Bh7 24.Bf7+ Kh8
The net result of the last 9 moves is that White has kept his extra
pawn! At first I was a little worried about black's two bishops,
and the possibility of Rd8, but I decided my pieces were still
better placed than Black's, and thought I would just "push them
25.Ke3 Nxb2 26.Ne4 Rd8 27.Nd6 Rd7 28.Be6 Bd3
Or 28...Re7 29.Nf7+ Kg8 30.Nxg5+ Kh8 31.Rf7 Re8 32.Be5 Rg8 33.Bxb2; or 31...Nxc4+ 32.Ke2 Re8 33.Rxb7 are winning.
29.Nf7+ Kh7 30.Nxg5+ Kg6?
Not the best, but neither 30...Kh6 31.Rf7 Rxf7 32.Nxf7+ nor 30...Kh8 31.Rf7 give black time to regain his pawn.
31.Bxd7 Bxf1 32.Ne6! Bxg2?
Time trouble : but 32...Nxc4+ 33.Kf2 Nb6 (...Bd3 34.Nf4+) 34.Kxf1 Nxd7 35.Nd8 is still winning for white.
33.Nxg7 Nxc4+ 34.Kd4 Nb6 35.Ne6 Kf6 36.Bh4+ Kf7 37.Nf4 Bf1 38.Be6+ Kg7 39.Bb3 Nd7 40.Ne6+ Kg6 [TIME] 1-0