Dvoretsky's Analytical Manual

Part 1

Immersion in the Position

A bottomless pit opens,

Full of stars

Stars uncountable,

A bottomless abyss.

– Mikhail Lomonosov

In this section we shall look at interesting positions (and if we examine an entire game, then we shall still confine ourselves to its crucial moment). And still – behind the scenes of one little episode, there may sometimes be concealed more fruitful, deep and beautiful ideas than one could perhaps carry out in an entire tournament.

Chapter 1

Combinative Fireworks

Not long ago, I reexamined an exercise from my notebook, in which White won thanks to a beautiful attack. I ran it past Fritz, which told me the attack was refuted – several different ways. I could not believe that – Black’s king position looked much too dangerous. And in fact, soon the computer went from dour to smiley-face, showing equality in lines that it had previously considered won for Black. This was more believable; but I am not a fan of situations in which there are a number of equivalent ways to play (even though, in practice, such situations occur all the time). I continued searching; and finally, all the defensive tries went down, save one – now this result, I could live with. And in the end, we found a difficult win here, as well.

The course of analysis uncovered more and more subtleties, some of them utterly fantastic, non-human. People cannot play chess at this level yet – in fact, I hardly think they will ever be capable of learning how, especially taking into account the current tendency toward faster time-controls and the ever-growing predominance of sporting (or should I say, financial) aspects over the creative side of chess life.

I do not believe that I have ever seen a position under analysis, filled with such an enormous variety of combinative content. This one example contains a multitude of exercises geared primarily toward the training of combinative alertness, but also toward accuracy in the calculation of variations, and of knowing when to attack and when to defend.

For the top-class player, it might prove an even more effective form of training to play out the starting position against a strong opponent. Play several such games, alternating white and black, allowing yourself and your opponent enough thinking time (not less than an hour apiece), and only afterwards, begin your acquaintance with my analysis.

Lelchuk – Voronova

USSR 1983

White is a pawn down; what is more, she has four (!) pieces en prise. She could minimize her losses by playing 1 Rxd6? Qxd6 2 Bh6+ (or 2 Qxd6 Bxd6 3 Nd7 fg) 2…Kxh6 3 Nf7+ Rxf7 4 Qxf7 Ng7, but this is completely hopeless, of course. So finding a combination that will set the opponent problems is no luxury here, but a matter of sheer survival; there is no point in even thinking about checking its correctness.

The careless 2…Rxd1?? ends in mate after 3 Ng4+ Kh5 4 Qxh7+ Kxg4 5 Qh3, as well as 2…fe?? 3 Qxf8+ Kh5 4 g4+! Kxg4 5 Rg1+ (5 Qh6!) 5…Kf3 6 fe+ (or 6 Rxd6 first).

And 2…Rxf7?! 3 Nxf7+ would leave White the exchange up after either 3…Kg7 4 Nxd8 or 3…Kh5 4 Nxd8 Rxd1 5 Rxd1 Bxf4 6 Rb1 (6 Rd7 ). Of course, such an endgame could not appeal to Black. Thus, Tatiana Voronova decided to go for a position in which her king would stand out in front of its whole army.

When warding off a dangerous onslaught, sometimes we find ourselves facing a difficult dilemma: should we attempt to cast doubt on our opponent’s idea in the most principled variations (usually involving the win or retention of material), or should we avoid them, and select a roundabout path instead, one involving significant concessions? Both approaches are valid – the choice between one or the other is determined by our assessment of the situation on the board, and by concrete calculation of variations – and sometimes, by intuition.

Now the hyper-aggressive 4 Rde1? does not work: 4…Kxg4! 5 Re3 (the line 5 Rxe7 Kh5 6 Re3 Qd7! 7 Rh3+ Qxh3 8 gh Kh6 leaves Black with great winning chances) 5…Qd7! 6 h3+ (6 Rg3+ Kh5 7 Rh3+ Qxh3 8 gh Kh6–+) 6…Kh4! 7 Rxe7

Black has two tempting continuations at his disposal; the task is to select the more effective one.

7…Qf5!

7…Rd1 8 Kh2! Qxe7! 9 g3+ Kh5 10 Qxe7 Rxf1 11 Qe2+ Kh6 12 Qxf1 is weaker.

8 Rxe8 Rd7! (a necessary insertion, as 8…Rxe8? 9 Qxe8 offers Black no advantage) 9 Re7 Rxe7 10 Qxe7 Bxf4!, and White cannot withstand Black’s considerable advantage in material.

The knight must be preserved – it will prove useful in the attack.

4 Ng4-e3!

Defending such a position is not at all easy. If Black plays 4…Kh6?, for example, White will not repeat moves, but play 5 Rde1! instead, with the terrible threat of 6 Rf3.

And nevertheless, Black still has quite a few defensive resources – which may, in fact, only make it harder for her to come up with the right decision. There are few weighty positional considerations in favor of one or the other continuation, and it would hardly be feasible to calculate all the variations over-the-board.

Let us examine 4…f5?! first.

With this move, Black parries 5 Rde1?, as now she would have 5…Rf6 6 Rf3 Bxf4!.

And the consequences of 5 Rxd6 Bxd6 6 Nxf5! (6 Rf3 would meet the same reply) 6…Bxf4! or 5 h3!? e6! 6 g4+ Kh6 (6…Kh4? 7 Kh2!+–) 7 g5+ Kh5 8 Qb7! Rd5! are unclear.

However, she wins spectacularly with 5 Nxf5!! Rxd1 6 g4+ Kxg4 7 Ne3+ Kh5 8 Rxd1 Qc8 (8…Qc7? 9 Rd5+) 9 Qd5+ e5 10 Qg2! Kh6 11 Qg5+ Kg7 12 Qe7+ Kh6 13 Rd7 Qxd7 14 Qxd7 ef 15 Ng4+ Kg5 16 Nf2 – White has both the material advantage and the attack.

Black is also unable to survive the line 4…e5?! 5 fe Rxd1 6 Rxd1 Qc8 (6…Qc7 7 Qc4+–).

True, after 7 Rd7?! Qc6 (7…Bxe5? 8 Qe6!) 8 h4! (threatening 9 Qxh7+!) 8…Kh6 9 Ng4+ Kh5 10 Nf2(e3) Kh6, it looks like White has to settle for the repetition.

But 7 ef! (threatening 8 Rd5+ or 8 Qd5+) is stronger: 7…Nc7

Black’s king is not to be envied; the only question is how to mate it? 8 Qc4 is met by 8…Qe6! 9 Qe2+ Kg5. If 8 Rf1 (intending 9 Rf5+), then 8…Ne6! 9 Rf5+ Ng5, and the threat of 10…Qxf5 11 Nxf5 Nxf7 leaves White no time to press the attack.

The solution lies in the hidden quiet move 8 g3!!, which strengthens all White’s threats. For example, 8…Qe8 (8…Qe6 9 Rd5+!; 8…Qb7+ 9 Rd5+ Nxd5 10 Qb7; 8…Qg8 9 Qd7 Qc8 10 Qc6) 9 Qc4!, and the only way to prevent mate by Qh4 is 9…g5, when there follows 10 Qg4+ Kh6 11 Qh3+ Kg6 12 Qf5+, etc. And if 8…b5, then 9 Rf1! Ne6 10 Rf5+ Ng5 11 Rxg5+! Kxg5 12 h4+ Kh6 (12…Kh5 13 Qd5+) 13 g4!, and Black cannot parry the twin threats of 14 Qg7# and 14 Nf5+! gf 15 g5#.

However, there are other paths to the goal; I present but one, also very pretty: 8 Qe7! Qe8 (on 8…Qe6, simplest is 9 Qg7 Qg8 10 g4+ Kg5 11 h4+ Kf4 12 Qh6+).

Here, White delivers the brilliant 9 Rd8!! For example, 9…Qxd8 10 g4+ Kg5 11 h4+ Kxh4 (11…Kf4 12 Ng2+ mates quickly) 12 Ng2+ Kg5 13 Qe5+ Kxg4 14 Qf4+ Kh5 15 Qh4# – queen and knight, uniting against the opposing king, demonstrate once more the power of this particular tandem. Or 9…Qxe7 10 fe Re8 11 Nd5! Nxd5 (11…Kg5 12 Rxb8!) 12 Rxe8 Bd6 (12…Be5 13 c4) 13 Rd8, with an easy win.

Voronova traded rooks at once, which somewhat reduced her opponent’s attacking potential and allowed her to develop her inactive knight.

Another tempting way to attack was 6 g4+?! Kh6 (but not 6…Kh4? because of 7 Ng2+ Kxg4 8 Qe6+ Nf5 9 Qe2+, forcing mate)

The endgame after 7 g5+?! fg 8 Ng4+ (8 fg+ Kh5! 9 Qf3+ Kxg5) 8…Kh5 9 Qd5 Ne4! 10 Qxd8 Rxd8 11 Rxd8 Bxf4 12 h3 Kh4 13 Kg2 Nxc3 favors Black.

Stronger is 7 Nf5+! gf (a complex, but apparently equal position arises after 7…Nxf5 8 Rxd8 Rxd8 9 gf) 8 g5+! fg 9 fg+ Kxg5. Now after 10 Rg1+? Kf4! 11 Qe6 Ne4! 12 Rf1+ Ke3 13 Qxf5 Rf8!? 14 Re1+ Kd2 15 Qxe4 Bf4, Black’s king penetrates deep into the enemy position, while successfully eluding pursuit. White has to continue 10 Qg7+ (10 Qe6!?, preparing 11 Rg1+, also maintains the balance) 10…Kf4 11 Qg3+ Ke4 12 Qe1+ Kf4, with a perpetual check.

We have arrived at the deciding moment of this game.

Unfortunately, in the game Black, unable to bear the tension, played 6…f5?, and lost quickly: 7 g4+ Kh6 (on 7…Kh4 8 Qe6!+– is most exact) 8 Rxd6! ed 9 Nxf5+ gf 10 g5+ 1-0

Zoya Lelchuk conducted the attack in outstanding fashion. But what would have happened if Black had pushed a different pawn?

It is vital to cut off the king’s retreat. 7 Qe6? Kh6 8 fe Kg7 9 ed Re8 10 Qh3 Bxd6, or 7 g4+? Kh6 8 g5+ Kh5 would be much weaker: Black must win this.

Not 8 Qd7? at once, in view of 8…f5! 9 g4+ Kh6 10 fe Qe8!? 11 Rxd6 Bxd6 12 Qxd6 Qd8!?–+.

Threatening 10 Ng2+ Kh3 11 Rd3#. How should Black defend?

9…c4? loses quickly: 10 Ng2+ Kh3 11 g5+ f5 12 Rxe5.

9…f5?! is not so simply refuted. White continues 10 Rxe5 (10 fe? Ne4 11 gf gf would be considerably weaker).

Now if 10…Qf6?!, then 11 gf. Alexander Motylev continued the line as follows: 11…Nc4 12 Ng2+ Kh5 13 Qd3 Kh6 14 Re6 Qd8 15 Qh3+! Kg7 16 f6+, or 11…Nf7 12 Re7 Qd6 (12…Rd8 13 Qxd8!) 13 Qb7!, and wins.

10…Qd8 11 Qe6 Re8 (11…Ne4 12 Rxe4! Qd6 13 gf+–) 12 Qd5 Ne4

13 Rxe8 Nf2+! (the d1-square must be covered – 13…Qxe8? fails to 14 Ng2+ Kxg4 15 Qd1+, and mates) 14 Kg1(or Kg2) Qxe8 15 Kxf2 Qe4! (the only defense to the mating threats) 16 Qd8+ Kh3 17 Qxb8 fg 18 Qe5! Qf3+ 19 Ke1, when the extra knight ought to be enough to win for White. She has the consolidating 20 Nf1 at her disposal, while capturing the h-pawn loses quickly: 19…Kxh2? 20 f5+ g3 (20…Kg1 21 f6) 21 Nf1+ and 22 Nxg3.

Things are even more complicated with the other two defensive tries, 9…e4!? and 9…Kh3!? In both these cases, achieving success would require White to execute a series of difficult, only moves, while avoiding a number of temptations along the way. It is hard to say which variation is the main line, but I got the most enjoyment out of analyzing the first one.

1

When we attack after sacrificing material, we generally try to force the issue; in particular, we

give check at the first opportunity. Here too, 10 Ng2+?! Kh3 11 g5+ f5 (but of course, not 11…Nf5? 12 Rd3+! ed 13 Qxd3+ Kg4 14 h3+ Kh5 15 Qf3#) suggests itself:

White’s task here is to find a way to bring the queen closer to the enemy king.

A subtle move – now Black will have to think about 13 Rd3+.

12…Kg4 is riskier: 13 Qxb6 (the queen wants to approach the enemy king via b1) 13…Nb5!! (the only defense) 14 ab Bxf4 15 Nxf4 Kxf4 – White can either give perpetual check – 16 Qc7+ Kxg5 17 Qg3+ Kh5 18 Qh3+ – or carry on the fight with 16 ba!? Kxg5 .

13 Rd3+! Kg4! 14 Rg3+ Kh5 15 Qd5 Qd6 16 Rh3+ (16 Qxc4?? Qd1+) 16…Kg4 17 Rg3+ Kh5, and perpetual check.

White must play something else.

A quiet move, providing a decisive strengthening of the attack. White’s main threat is not at all obvious. She will execute it, for example, after 10…Kh3 (or 10…c4).

It is mate, in no more than ten moves, after a spectacular rook sacrifice.

11 Rh5+!! gh 12 g5+! f5 13 Qd5, followed by 14 Qd1(d2), and Black’s king is defenseless. I note that 11 Rd3!! also accomplishes this, although by a much more complicated route. We shall be examining this position further, when we analyze 9…Kh3!?

White also has a pretty mate after 10…Qf7?

Here it is mate in six!

11 Ng2+ Kh3 12 Rd3+!! ed 13 Qe3+ Kxg4 14 Qg3+ Kf5 15 Qh3+ Ke4 16 Qxd3#.

The only acceptable defense remaining does not, however, save Black from the storm of combinations.

Black counts on 11 gf? Qf7! 12 Rxd6 Qxe6 13 Rxe6 Bxf4 14 Rxe4 g5, with about even chances. 11 Rd1 is not dangerous either, the simplest reply being 11…b5, preparing 12…Nc4.

The rook is immune: 11…ed? 12 Ng2+ Kxg4 13 Qe1, and Black’s king will be mated in two or three moves. On 11…Qf7?, White decides with 12 Ng2+ Kxg4 13 Rg3+ Kh5 14 Ne3! Kh6 (14…Kh4 15 Kg2) 15 Qe5.

In order to break up the mating net, Black has to capture the f4-pawn with her bishop as quickly as possible. For this, she must retreat the knight – but where? It must be to a square that is twice attacked!

It is important to attack the enemy knight, or else White will have time for the decisive strengthening of her attack by 12 gf! Here are some sample variations:

11…Nc8 12 gf! Qe7 (12…ed 13 f6+–) 13 Ng2+ Kh5 14 Rh3+ Kg4 15 Rg3+ Kh5 16 Rg5+ Qxg5 (16…Kh6 17 Rxg6+) 17 fg; 11…Nb7 (11…Nb5 is hardly different) 12 gf! ed 13 fg! (13 f6? Qc8) 13…Qxf4 (13…Qc8 14 Qf6+ Kh3 15 Qg5 Qc6+ 16 Ng2+–; 13…Kh5 14 Qg4+ Kh6 15 g7+–) 14 Ng2+ Kg5 15 Nxf4 dc (15…Bxf4 16 g7+–) 16 Nh3+! Kh5 (16…Kh4 17 Qe3!? c1Q+ 18 Qxc1 Kxh3 19 g7+–; 16…Kh6 17 gh+ Kg7 18 Qe7+ Kg6 19 Qe4+ and 20 Qxc2+–) 17 Qf5+ Kh6 (17…Kh4 18 Qg5+ Kxh3 19 Qh5#) 18 Qxc2+–.

The knight is immune: neither 12 Qxc4? ed nor 12 Nxc4? Bxf4 works.

This is precisely the right moment to remove the knight – while the rook at h3 is hanging (since its capture results inescapably in mate: 15…Kxh3 16 Qe2!, threatening 17 Qe3+). After the superficial 15 Rh4+? Kf3 16 Qxc4 Kf2!!, the game takes a 180-degree turn.

Here, I would have made do with one of the smaller diagrams; but instead, I am presenting the largest one in the entire book! I have never seen such a fantastic situation!

White has not only almost restored the balance of material, but driven the enemy king deep into her own camp. Black’s other pieces all spin uselessly at the furthest possible distance from their king – back on the eighth rank; meanwhile, White’s pieces have the king surrounded. And with White to move, as well, it would seem that mate is just around the corner.

However, nothing of the kind is happening. As a matter of fact, it is Black who is winning here! Long ago, the great Steinitz wrote that the king is a strong piece, and can take care of itself. Just so here: the king not only withstands an entire enemy army by itself, but also intends to participate in the decisive counterattack!

Here is the main, and most indicative line: 17 Rh3 Qg8! 18 Qxa6 Qa2! 19 Re3

19…Qb1+ 20 Re1 Qxe1+! 21 Nxe1 Rd8!!, and Black wins.

Moving the white queen to the first rank does not help: 17 Qb3 Qd6 18 Qb1 Qd2 19 Rh3 Bxf4 20 Qg1+ (on 20 Nxf4, simplest is 20…Qe1+! 21 Qxe1 Kxe1, and the black pawns decide) 20…Ke2 21 Nxf4+ Qxf4 22 Rg3.

22…Kd2 is not completely clear: 23 Rg2+ Kxc3 24 Qa1+. She should continue: 22…e3! 23 Qg2+ (Black’s task would be simpler after 23 Rg2+?! Qf2!) 23…Qf2 (23…Ke1!? 24 Rf3! Qxf3! 25 Qxf3 Re8! would also be very strong) 24 Qd5! Re8!? (24…Qf1+? does not work: 25 Rg1 Qf3+ 26 Rg2+; however, 26…Rd8!? 27 Qxd8 Qf1+ 28 Rg1 Qf3+ 29 Rg2+ Ke1 30 Qd3 c4! 31 Qxc4 e2 32 Qd4! Qe4 is worth considering: despite the extra rook, White has a difficult defense) 25 Qd3+ Ke1 26 Rg1 Qxg1+ 27 Kxg1 e2 28 c4 Re4!, and Black appears to be winning.

There is nothing else – Black must shed a piece. Once it is taken, the king returns home, and Black’s strong central pawns secure her good saving chances. So White must be accurate to the very end.

White can win the bishop several different ways, but not all ways are equal. 16 Rh4+ Kg5 17 Rxf4 Qd6 or 16 Qe2+ Kg5 17 Qf2 Kf6 18 Nxf4 Qd6 19 c4 Qd1+ 20 Kg2 Rd8 21 Nd5+ (21 Qh4+ g5!) 21…Kg7 are unconvincing.

Black can neither take the rook (16…Kxh3? 17 Qxf4) nor save the bishop: 16…Be5 17 Nf4!! Kg5 (forced) 18 Ne6+.

More accurate than 17 Qxf4+ Kf6 18 Ne3 Qb8.

The knight obtains the excellent central post d5, and the queen obtains the a1-h8 diagonal. Black cannot take on f4, because of 19 Rg3+. White has a winning position.

2

The king runs away from the knight check on g2 – while, of course, rendering itself vulnerable to other checks, such as g4-g5 or Rd3. I began to study this move after completing the analysis of the 9…e4!? variation, and for a while, I thought that this cheeky king sortie deep into the enemy position would let Black save the game.

After 10 Ng2?! e4, the game transposes into the drawing variation 9…e4 10 Ng2+?! Kh3 we examined earlier. Black also has another way to maintain the balance: 10…f5!? 11 gf Qf7! (but not 11…Qe8? 12 Rd3+ Kg4 13 Rg3+ Kh5 14 Qg7+–) 12 Rd3+ (12 Rxd6? Qxd7 13 Rxd7 ef would be bad for White) 12…Kg4 13 Rg3+ Kh5 14 Rh3+ Kg4, with perpetual check.

10 g5+?! f5 leads to entrancing complications. Taking the pawn by 11 Rxe5? brings White to the brink of disaster: 11…Qd8! 12 Qe6 (12 Qg7 Nf7!–+) 12…Re8 13 Qd5 Ne4 14 Qc4 (14 Qb7 Bc7 15 Qxa6 Nd2!, intending 16…Qa8+).

The obvious 14…Qd2 15 Qf1+ Kh4 would only be enough to draw. Black is right to expect more.

14…Ng3+!! 15 hg Bxe5 16 Qf1+ Kxg3 17 Qg2+ Kxf4 18 Nd5+ Qxd5 or 16 Nd5 Kxg3 17 Qd3+ (17 fe Qxg5!–+) 17…Kh4! 18 Qd2 Qxd5+! 19 Qxd5 Bxf4 .

So 11 Rd3 e4 must be played.

Now what? 12 Nxf5+? fails to 12…ed 13 Nh6+ (13 Nxd6+ Kh4+–) 13…Nf5 14 Qxd3+ Kh4 15 Qf3.

With such a large advantage in material, Black is quite justified in giving up the queen.

15…Qxh6!! 16 gh Re8, with a winning position. Retreating the white rook by 12 Rd1? also meets with a pretty refutation.

The careless 12…Qf7?? would allow White to conclude the game with a spectacular mate: 13 Rg1!! Qxd7 14 Rg3+ Kh4 15 Kg2, with the unstoppable threat of 16 Rh3#. On 12…Qd8?, White plays the same queen sacrifice, except that now it is only good enough for a perpetual check: 13 Rg1!! Nc4! 14 Rg3+ Kh4 15 Qg7 Qf8 16 Ng2+ Kh5=. And another drawing line is 12…Nf7? 13 Qd2 Bxf4 14 Qf2 Qb8!! 15 Rg1!.

The winning line is 12…Nc4!! 13 Nxc4 (13 Ng2 Kg4) 13…Bxf4.

12 Qc6!!

The only way for White to avoid losing.

12…Nc4!?

Black is not worried about losing the knight here: 13 Nxc4+? Kh4, followed by 14…Bxf4.

13 Nxf5+ (or 13 Ng2+ Kg4 14 Rg3+ Kh5, with our familiar perpetual check) 13…Kg4 14 Qxe4 Qxf5.

White can check in many different ways, but cannot mate Black’s king.

White has one more attacking line at her disposal: 10 Rd3. I had thought that 10…e4 11 g5+ f5 transposes into our last line. But grandmaster Motylev, to whom I showed my analyses, disagreed with my conclusion, and found a difficult path to victory. I present his variations (with some retouching) herewith.

10…c4 11 Qc6! e4 encounters a refutation that is quite difficult to see.

White only gets a draw from 12 Nxc4+? Kxg4 13 Nxd6 Bxd6 14 Qxe4 (14 Rxd6 Qa8) 14…Qa8 (14…Bxf4 is not bad, either) 15 Rd5 Re8 16 Qg2+ Kxf4.

12 Qxb6!!

We saw this idea in an earlier variation. The queen takes a roundabout path to get to the enemy king – via the first rank. Taking the rook with either pawn allows 13 Qb1!, forcing mate. If Black blocks the queen’s path by 12…Nb5, 13 Nd5+ Kxg4 (13…ed 14 Qg1!) 14 Nxf6+ Kxf4 (14…Kf5 15 Rd5+) 15 Qf2+ Ke5 16 Nd7+ decides.

12…Nf5!? 13 Nxf5+ Kxg4 (13…cd 14 Qg1!) 14 Rd5! gf 15 Qg1+ Kh5 16 Rxf5+ Kh6 17 Qg4 Qf7

A simple tactical stroke leads to the win of material.

18 Rg5! Qg6 (the only way to prevent an immediate mate) 19 Rxg6+ hg 20 Qh3+ Kg7 21 Qd7+ Kh6 22 Qd4 Re8 23 Qxf6 (with the terrible threat of 24 f5), and White wins.

Just so – without the insertion of 11 g5+?! f5. As we have seen more than once in this game, a quiet move brings far greater dividends than a tempting check.

Equally valid is 11 Qc6!

After 11…c4, play transposes into our familiar line 10 Rd3 e4 11 Qc6 c4 (in which, let me remind you, White wins by 12 Qxb6!!).

On 11…g5, White replies, just as she would after 11 Qe6 – with 12 Qd5.

Black could still try 11…Qe8. Once again, checks lead nowhere: 12 Ng2+? Kxg4 13 Rg3+ Kh5 14 Rh3+ Kg4 15 Rh4+ Kf3 16 Qd5 (16 Rh3+ Kg4=) 16…Ke2! (the only defense) 17 Ne3! (after White’s slightly risky refusal of the repetition, she finds she has but one means of escaping defeat) 17…Kxe3 18 Qd1, forcing perpetual check.

12 Qd5! c4 13 Rd1

On 13…Nf5 14 gf Bxf4, the quickest way to win is 15 Rg1! Bxe3 16 Rg3+ Kh4 17 Qd1.

Now, we really should take a closer look at the line 13…g5 14 Qd2 gf 15 Qf2, since this position could also arise after 11 Qe6.

On 15…Nc8, White wins by 16 Rg1! (threatening 17 Nf5) 16…Ne7 17 g5!. And mate soon follows after 15…h5 16 Rg1, as well.

But after 15…Qe5! this recipe no longer works: 16 Rg1? Rg8 17 Rg3+! Kh4! and Black wins. But in that case, White has 16 Rd5! fe (16…Qxd5 17 Qxf4 Nf7 18 Qf1+ Kh4 19 Nxd5) 17 Qg2+ (but not 17 Qxe3+? Kxg4 18 Rxe5 fe–+) 17…Kh4 18 Rxe5 fe 19 Qg3+ Kg5 20 Qxe5+.

White’s queen has time to take the rook, and still get back home in time to stop the passed e-pawn. 20…Kxg4 (20…Kg6 21 Qxh8 e2 22 Qg8+ Kf6 23 Qd8+ Ke6 24 Qh4+–) 21 Qxh8 e2 22 Qg7+ Kf4 (22…Kf5 23 Qxh7+ Kg5 24 Qg7+ Kf5 25 Qf8+ Ke5 26 Qf2+–) 23 Qg3+ Kf5 24 Qf2+ Ke5 25 Qxe2, and White’s material advantage must tell.

And now, let us see what happens if we move the queen to e6.

11…ed? 12 Qd5! loses at once for Black. And 11…c4 leads to unstoppable mate after 12 g5+ f5 13 Nxc4+ Kh4 14 Ne3 Kh5 15 Ng2 Qf7 16 Rh3+ Kg4 17 Rg3+ Kh5 18 Ne3! Kh4 19 Kg2.

The only defense.

12 Nd1+? Kh4 13 Nf2 gf 14 Rh3+ Kg5 15 Nxe4+ Kg6 16 Nxf6 (White does not have enough for the exchange after 16 Rd3 Qf7 17 Rxd6 Bxd6 18 Qf5+ Kg7 19 Nxd6 Qe7).

Black unexpectedly goes for the counterattack by 16…Qf7! 17 Qe5 Qa2!!, intending 18…Qb1+ and 19…Qxc2+.

12…Qg8 13 Nf5+ Kxg4 14 Nh6+ Kxf4 15 Nxg8 ed 16 cd is hopeless for Black.

12…Rg8 13 Nf5+ Kxg4 14 Rg3+ Kh5 15 Qe6 g4 allows White a tactical resource to finish off the attack.

16 Ne7! – a decisive double attack! The rook is en prise (16…f5 17 Nxg8+–), and retreating it leads to a mate in two: 16…Rg7 17 Rh3+! gh 18 Qxh3#.

13 Nxc4+? would let the win slip after 13…Kh4 (but not 13…Kxg4? 14 Rg3+ Kh5 15 Rxg5+!) 14 Kg2 Nxc4 15 Qxe4 Qe8 16 Rh3+ Kxg4 17 Rg3+ Kh4 – White has no time to take the knight, in view of the threatened Bxf4.

Stuck in the opposing camp, Black’s king needs the help of its pieces – which they cannot readily provide. 14 Qd2 is threatened, and Black cannot capture on f4 for the moment, because the queen mates at h5.

If 13…Kh4 (to answer 14 Qd2 with 14…gf), 14 Rg1!, threatening 15 Rg3 and 16 Ng2#, is decisive.

On 13…h5, the most convincing win is 14 Rf1 gf (14…hg 15 Qd1 f5 16 Qe1+–) 15 Rxf4, followed by 16 Qd1 (15 Ng2 is also good, with its threat of 16 Nxf4+ Kh4 17 Ng6+).

On 13…Qf7, there follows 14 Qd2 gf 15 Qf2, and Black is helpless. If Black plays 13…Qe8 (instead of 13…Qf7), then in the resulting position, she could defend herself by 15…Qe5 – but then she would still be doomed, since this would be the same situation we analyzed when we looked at 11 Qc6.

A last attempt, but this also fails. Motylev demonstrates two ways to win.

Another good way is 14 Rf1 gf (14…Rg8 15 Qd2 gf 16 Rxf4 Kh4 17 Ng2+ Kg5 18 Rxf5+ Kxg4 19 Qf4+ Kh3 20 Qh4#) 15 Rxf4 Qh6 16 Ng2 fg 17 Qd2

The quickest and most elegant way to win after 17…g3 is 18 Rxe4!, and after 17…Qg5 – 18 Rf5!

If 15…Ne8, then 16 Nxf5 Kxg4 17 Rd5! Kg5 18 Nd6+ Kg6 19 Qg2+ Kf6 20 Qg5+ Ke6 21 Qe5+ Kd7 22 Nxe8+.

One last effort:

Far be it from me to consider my analysis errorfree – I have often seen the old maxim of Bent Larsen confirmed: Long analysis, wrong analysis. Experience tells us that any serious new immersion in a position is capable of introducing fresh ideas, corrections and additions to those previously found. In fact, the note to Black’s third move (concerning the attempt 4 Rde1?) came about after the position had been played over by two strong grandmasters – to say nothing of Motylev’s contribution to the analysis of the important variation 9…Kh3!?

Let us summarize. We have found an exemplary game – long enough (seventeen or eighteen moves in either of its two main variations), spectacular enough, and also unusual. No doubt, this analysis will get a further working-over from a number of readers. They may confirm and fill out the picture I have drawn. They might also arrive at the opposite result, after which our pretty story falls completely apart – but then, we would still have many glittering fragments.

Postscript

After this analytical article had been published at ChessCafe.com (in August 2006), a reader sent in a letter, suggesting two new defensive tries for Black. I analyzed the new situation thoroughly, and came to the conclusion that the reader was right: Black is saved.

First case: instead of 4…Rxd1!, Black can play

4…e6!

After 5 Rxd6 Bxd6 6 Rf3 Bxf4!, White must either accept being a pawn down, with extremely nebulous compensation after 7 Qxe6 Kh6 8 Rxf4 Kg7, or else try to save the inferior endgame that comes about by force after 7 Rxf4 Qc7 8 Rxf6 Qf7 9 Rxf7 Nd6 10 Re7.

5 Rde1 is stronger (5 Rb1 is possible too; for example, 5…Rd7 6 Rf3 Bxf4 7 Qxe6 Kh6 8 Rxf4 Kg7, with unclear play) 5…Rd2 6 Rf3 Bxf4!

a) 7 Rxf4 Qd7 (or 7…Qc7) 8 Rxf6 Qxf7 9 Rxf7 Nd6 10 Re7 – with all four rooks on the board, White probably stands no worse.

b) 7 Qxe6!? – White keeps sufficient compensation for the pawn (but probably no more than that) in the lines 7…Kh6 8 Rxf4 Kg7 or 7…Qd7 8 Qe4 Kh6 (8…g5?! is weaker: 9 Rh3+ Qxh3 10 gh Rxh2+ 11 Kg1 Rxh3 12 Nf5 ) 9 Qxf4+ Kg7.

Instead of 6 Rf3, 6 g4+!? Kh4 7 Qxe6 is also worth consideration, but then White can really hope for no more than a draw.

There could follow 7…Nd6 (7…Nc7 8 Qe4 Nd5 9 Rf3=; 7…Ng7 8 Qf7 Qf8 9 Qb7!= or 8…Rd7 9 Qc4! Rd2!=) 8 Ng2+ Rxg2 9 Kxg2 Re8 10 Qd5 f5! (Black threatens 11…Ne4) 11 Rf3 Ne4 12 Rh3+ Kxg4 Rg3+ with perpetual check, as the capture of the knight leads to mate in two (shown by Karsten Müller).

And secondly, after 4…Rxd1! 5 Rxd1, the natural 5…Nd6 now appears to be a mistake: 5…Qc8! is stronger, retaining the important defensive resource Bxf4!.

The inferior endgame for White that we are already familiar with arises after 6 Rf1 e6 7 Rf3 Bxf4 8 Rxf4 Qc7 9 Rxf6 Qxf7 10 Rxf7 Nd6 11 Re7. So she plays 6 Rd5+ Kh6 7 h3 f5 (in the line 7…Nc7 8 Ng4+ Qxg4 9 hg Nxd5 10 g5+ fg 11 fg+ Kxg5 12 Qxd5 Kh6 13 Qc4!?, threatening 14 Qh4+ and 14 Qxa6, White retains the advantage) 8 Rxf5 (or 8 Qxe7 Bxf4 9 Qh4+ Kg7 10 Qe7+ Kh6, with perpetual check) 8…gf 9 Nxf5+ Qxf5 10 Qxf5 Kg7, with about even chances.

Chapter 2

Chess Botany – The Trunk

From our search for brilliant combinations, we turn to the more prosaic, though no less important issue: calculating combinations accurately.

In his book Think Like a Grandmaster, Alexander Kotov introduced the concept of a tree of variations, that being the collection of variations which needs to be analyzed. He also enumerated three kinds of trees – the bare trunk, the shrub, and variational debris. In this and the chapters that follow, we will be looking at examples of each kind of tree.

The bare trunk is a long, forcing variation practically devoid of alternatives. (In point of fact, there are almost always alternatives; but if they are of little significance, we can honestly assign our calculations this kind of label.)

For less highly-skilled players, the main impediment is the need to accurately foresee each of the many positions that come up over the course of the calculation. Over a lengthy calculation, they are prone to lose the thread, and find themselves unable to continue the variation.

But for trained chessplayers, too, there are difficulties – above all, psychological ones. The deeper one goes into the variation, the stronger grow the doubts: should I extend this line? Did I calculate everything correctly? Did I overlook something important? You can increase your confidence in your calculating by moving down the line without haste, stopping at each step to check carefully whether or not there might be a strong alternative, either for yourself or for your opponent.

Spraggett – Browne

New York 1987

Your task is to evaluate 32 Bxa5.

Even with the question phrased this way, focusing on analyzing one concrete continuation, it is still necessary to think, if only for a little while, about the starting position: who is better, and what will happen if White plays a quiet move.

For if White stands better, then we will not be satisfied if, for example, the sharp variation that we must calculate ends in a draw. We would also be justified in cutting our analysis short if we see that it would lead to a situation that would be difficult to evaluate, and involve considerable risk.

On the other hand, these circumstances do not require us to just give up on the main continuation, if we assess the starting position in our opponent’s favor.

In the actual game, Spraggett chose 32 Rd3. Black defended the a5-pawn with 32…Ra8 – although 32…Qa8!?, intending 33…Qa6! was also worth considering. An exchange of queens would render the b3-pawn more vulnerable, with Black apparently taking over the initiative.

White’s next move was a serious error, which his opponent failed to exploit.

33 h4? (33 Kf2 was correct)

The simple, but elegant tactical blow 33…Bxh4! would have won a pawn, since 34 gh?! would be bad in view of 34…Qxd3! 35 Qxd3 Nxf4+.

33…h6? 34 Kf2 g5?

This is suicide! After 34…Kh8, the position remains unclear.

35 Re2 Qb7 36 hg hg 37 fg Kg7 38 Bd2

Of course, this does not give us a clear-cut answer to the assessment of our starting position, but some definite ideas have probably surfaced already. It seems to me that Black’s chances are preferable – his position is easier to play.

And now, let us start calculating the main line.

Of course not 35…Rb2+? 36 Rd2+–.

White attacks the rook, and plans to create threats against the king by Bd6. After 37…Re8? 38 Bd6, he would have an obvious advantage. The counterattack 37…Rb1? fails in view of 38 Qc8+! (we shall examine the alternative 38 Bd6 after completing our analysis of the main line) 38…Kg7 39 Qe8+–.

This would all be very nice for White, if his opponent did not have the following powerful move, forcing the exchange of queens:

The endgame looks good for Black, in view of the power of his connected passed pawns.

Here is exactly the place to recall those impressions we got while considering the starting position. If you liked it for White, then perhaps it would make sense to stop here and reject the capture of the a5-pawn in favor of a quieter continuation. But if that position was not to your liking, then you should continue with the analysis: perhaps, in the endgame, White’s counterplay – advancing his a-pawn – will be enough to save him. If so, then by taking the pawn, you avoid defending an inferior position, force a draw (that is, if your calculations were correct), as well as pose a problem for your opponent somewhere down the road (although it is true that the likelihood he will miss the move 37…Qb3!! is not very great).

Now if 39…c4, 40 Ra2 c3 41 a6 Rb2+ 42 Rxb2 cb 43 a7 b1Q 44 a8Q+ Kg7 leads to equality.

The a-pawn has been stopped, but now White has the chance to work up threats against the enemy king with e5-e6, Rb2 and Be5.

White likely has no reason to investigate the sharper line 43 Rb7 Ra2+ 44 Kh3 (44 Kf1?! Rxh2), although here too, the most likely outcome is a draw. For example, 44…Re2!? 45 Bg7+ Ke8 46 Rb8+ Ke7 47 Bf8+ Kxe6 48 Bxc5 d2 49 Rd8 c3 50 Bb4 Re1 51 Bxc3 d1Q 52 Re8+ and 53 Rxe1.

Forced.

The passed pawns cancel each other out; the position is even.

And now, I am going to offer an additional task for you to solve, also on the bare trunk (or more precisely – almost bare): calculate the variation 37…Rb1? 38 Bd6?! Bxd6 39 ed (as Karsten Müller points out, after 39 Qxd6+ Kg7! 40 Qf6+ Kh6 41 Qf8+ Kh5 42 Qg7 Qh1+ 43 Kh3 Rh7 44 Qf6 Kh6 45 Qf8+ Rg7 46 Qxc5 d3, Black would be quite justified in expecting a draw).

The first moves are forced: 39…Qg1+ 40 Kh3 Qxf2 41 d7 Rb8 42 Qd6+ Kg7

43 Qe5+!

43 Qxb8? Qf1+ 44 Kh4 Qf3 would lead to an immediate draw.

43…Kf7!

43…Kh6? would lose: 44 Qxb8 Kh5 45 Qb5 (or 45 Qb2!?).

44 Qxb8 Qe2!

44…Qf1+? 45 Kh4 Qf3 no longer works: 46 Qe8+ Kg7 47 Qe7+ Kh6 48 Qf8#.

45 d8N+!

Queening the pawn would allow Black to give perpetual check.

45…Kg7 46 Qb7+!

Retreating the king to f6 or to g8 leads to mate in one or two moves. 46…Kh6? is a mistake too: after 47 Nf7+ Kh5 48 Ne5+–, the king takes a vital checking square away from the queen.

46…Kh8! 47 Nf7+ Kg8 48 Nh6+ Kh8 49 g4! (the only way to continue to play for a win) 49…Qe3+ 50 Kh4 Qxf4 51 Qa8+ Kg7 52 Qg8+ Kf6 (52 Kxh6?? 53 Qf8#) 53 Qh8+! Ke6 54 Qxh7

This is probably a good place to stop, since the forcing play has more or less ended. White retains the advantage, but it is difficult to say whether it will be enough to win.

But now let’s return to the position after Black’s 44th move. Instead of the tempting promotion of the pawn to a knight, Pablo Pena suggested another, even stronger move, leading to a forced win: 45.Qe5!! Q:e5 46.fe Ke7 47.Kg2! c4 48.Kf2 c3 49.Ke1! (but not 49.Ke2? g5! 50.a5 f4 51.a6 f3+ 52.K:f3 c2 53.a7 c1Q 54.d8Q+ K:d8 55.a8Q+ Ke7=) 49…K:d7 (with the king on e1, any counterplay with g6-g5 is too late) 50.a5+–

For further training, I offer a couple of exercises from my notebook.

Kholmov – Jakobsen

Kislovodsk 1972

Calculate 37 Bxg6+.

White has an overwhelming advantage; there is no need for him to combine here. The game continued 37 Bf3 Rd3 38 Qf6 Qb7 39 h5 gh 40 Qxe5! 1-0

The bishop sacrifice would have risked throwing away the win.

If you can accurately calculate the main variation to the end, there is no need to distract yourself by calculating alternate lines. Nevertheless, I will demonstrate that immediately sacrificing two pieces will not give Black a draw: 41…Be2+? 42 Kxe2 Qb5+ 43 Kxd1 Qd3+ 44 Kc1 Qxa3+ 45 Kd2 Qb2+ 46 Ke3+– or 42…Rd2+ 43 Kxd2 Qxf2+ 44 Kc1 Qe3+ 45 Kb2 Qd4+ 46 Rc3 Qd2+ 47 Rc2 Qd4+ 48 Ka2+–.

43 Kh3 Bf1+.

But not 44…Bg2+? 45 Kh2!+–.

Black gives perpetual check.

We were convinced that the attempt to mate Black’s king would fail. In order to give a final evaluation of that bishop sacrifice, we must also make sure that White cannot strengthen the long variation we have found anywhere along the way. Since, after 39 Rc7, Black’s counterplay develops along forcing lines, with checks, we must look for an alternative on move 39.

A move worth looking into is 39 Qf7, defending his f-pawn, White threatens 40 Rc7.

Black responds with 39…Qd6!. Now 40 Rc7 Rd7 is pointless, and meanwhile Black wants to drive out the enemy queen by 40…Rf8.

But the attack is still not over: there is 40 h5! gh (40…Be8 41 Qf6 Kh7 is a mistake in view of 42 Rc5!, followed by 43 Rxe5) 41 g4! Now 41…hg? 42 Qf6+ Kh7 43 Kg2 would lose quickly for Black. And 41…Qd7 42 Qxh5+ Kg7 43 Qxe5+ is dangerous too, although White would still have to demonstrate an advantage.

The best defense is 41…Be8! 42 g5+! Kxg5 43 Qg7+ Bg6 (but not 43…Kf4? 44 Re1!+–). I do not see a win here; for example, 44 Kh2 Qd2 45 Rg1+ Kf4 46 Qh6+ Kxe4 47 Qxg6+ Kd4!?. The king is ready, if necessary, to go to c3 and beyond. The game we examined in the preceding chapter showed that mating such an active king can sometimes be extremely difficult, if possible at all.

I also looked at the zwischenzug 39 a4!!, but I thought that Black could defend successfully after 39…Be2! – he wants to trade rooks on d1; after 40 Rc7 Rd1+, he has the same draw as in the main line. However, Chiedu Maduekwe, who read my article analyzing this example, has found a way to win.

40 Qf7!

41 Rc7 is threatened, and now 41…Qd6 is useless, since the bishop no longer controls d7.

40…Rd1+ 41 Rxd1 Bxd1 42 Qg8! Qb7 (42…Qc7 meets the same reply, and on 42…Kh5 43 Kg2 decides) 43 b5! ab 44 ab.

The black queen is overloaded, having both to prevent mate and stop the passed pawn. It cannot carry out both tasks, and the bishop is not able to help, because if 44…Ba4, then 45 g4!.

Arnason – Miles

Reykjavik 1978

Black has a healthy extra pawn, and with a normal continuation, he will most likely win. So White should carefully examine the combination beginning with a sacrifice on g5, because there just might be a draw there.

In the game, Arnason decided against calculations, and deservedly lost: 36 Rg3+? Kg6 37 Bd5 Qc8 38 Qg2 Rh8 39 Rf3 Rh4 40 Rf5 Rf4 41 Bf7+ Kxf7 42 Qxg5 Rxf5 43 Qxf5 Qxf5 44 ef Kg7 45 Kf2 Kh6 46 Kf3 Kg5 47 Ke4 a5 48 b3 Kg4 49 Kd5 Kf4 50 Kxc5 e4 0-1

There is no other way to escape the checks. Now the game becomes a pawn ending.

Now Black will strengthen his position as much as possible, advancing his pawns and king (in this stage, the individual moves do not have much significance). In order to ensure that one of his pawns will queen, at some point he will have to leave the square of the protected passed d-pawn, which will advance at once. Here, once again, concrete play resumes, requiring accurate calculation.

It is useful to bear in mind here that the pawns could be set up a bit differently, starting with 45 a4!?. But for now, we should not distract ourselves with this – we can return to it if our calculation shows us that the queenside pawn structure will factor significantly into the coming queen endgame.

Two necessary elements of Black’s play in such endings are choosing the optimal pawn placement on the front line, and choosing the right moment to run the king in. Here, Black should advance the f-pawn, since after 50…e3+?! 51 Kf3 Kf5 52 Ke2 Ke4? does not work: 53 d6 f3+ 54 Kf1 e2+ 55 Kf2 Kd3 56 d7+–, and White’s pawn queens with check.

Inserting the moves 52…a4 53 ba here does not change the assessment of the position.

54 Kg1? Kg3 55 d7 e2 loses for White. And on 54 Ke1?! f2+ 55 Ke2 Kg3 56 d7 Kg2 57 d8Q f1Q+ 58 Kxe3 Qe1+, the endgame looks dangerous for him.

In Chess Informant 25, Miles assessed the concluding position of this variation as . True, he arrived at it by a somewhat different route: 54 Kf1 f2 55 Ke2, etc. But 54…f2?? here is a gross blunder, which actually loses for Black after 55 Kg2!.

White manages to obtain a more pleasant version of the queen endgame here.

This position is drawn.

For those who would like to solve similar exercises, I recommend looking in my book Dvoretsky’s Endgame Manual (3rd ed.) for the endings Pillsbury – Gunsberg, Hastings 1895 (p. 306) and Stangl – Schneider, Berlin 1992 (p. 74).

Chapter 3

Chess Botany – The Shrub

The Shrub is what Kotov called the situation in which one must calculate not a single long variation, but a number of comparatively short ones. Here, perhaps, the main problem is to identify all the variations worth examining. All the calculations we perform could prove fruitless, all our time wasted, if we neglect to study a strong candidate move at the very outset!

The problem position that follows arose right out of the opening; I shall therefore give those opening moves with brief notes.

Smyslov – Gufeld

USSR Spartakiad, Moscow 1967

1 c4 Nf6 2 Nf3 g6 3 b4 Bg7 4 Bb2 0-0 5 e3 b6 6 d4 c5! 7 dc bc 8 b5 a6 9 a4 Ne4!?

9…Bb7= was simpler.

10 Bxg7 Kxg7 11 Qd5!?

With this and his next move, Smyslov accepts his opponent’s challenge to enter irrational tactical complications. On the other hand, after quiet development with 11 Bd3 Qa5+ 12 Nbd2 Nxd2 13 Qxd2 ab! 14 cb d5, with Nb8-d7-b6 to follow, White could not expect an advantage out of the opening – Eduard Gufeld’s opinion.

11…Qa5+ 12 Ke2?!

This was also necessary as after 12 Nbd2 Nxd2 13 Nxd2 ab!, Black is already better. Still, in my opinion, 13 Qxd2 (or 12 Nfd2 Nxd2 13 Qxd2) would have been safer.

Black has two pieces en prise. Let us calculate the variations, so we can understand who is better in this position and how we should play it.

The first thing that attracts our attention would be 12…Nc3+? 13 Nxc3 Qxc3

14 Qxa8? Qxa1 15 Qxb8 would be a mistake, which could be strongly met not only by 15…d6!? (Black can expect at least a perpetual here – Gufeld), but also by 15…Qb2+ 16 Nd2 d5!, intending 17…Bg4+ or 17…dc (Dvoretsky).

Smyslov had prepared 14 Rd1! Ra7 15 Qxc5, with an overwhelming advantage (once the rook retreats, White can trade queens by Qd4+).

In the game, Gufeld executed a pretty idea: 12…Bb7 13 Qxb7 Nc6 (meeting 14 bc with 14…Rab8). As will shortly become clear, the combination leads to an approximately even position.

The grandmaster examined no other possibilities for Black. If in fact there were none, we could stop our calculations here. But, remembering the candidate-move rules, we shall not jump to conclusions.

The amusing 12…Qc3? pops into our head, but does not inspire confidence. Even the endgame, with 13 Nxc3 Nxc3+ 14 Kd2 Nxd5 15 cd favors White and 13 Qxa8 instead leaves Black badly off wherever you look: 13…Qxa1 14 Qxe4+–; 13…Qb2+ 14 Nfd2+–; 13…Qxc4+ 14 Kd1! Nxf2+ 15 Ke1 Qc2 16 Na3+–.

However, there is one more combinative resource, and this one is much stronger: 12…Nd6!