Art of Attack in Chess

by Vladimir Vukovic

  Naturally, an attack on the fianchettoed king position is generally directed against these weak spots, and consequently, the assault by the h-pawn and actions aimed at the exchange of the fianchettoed bishop are the principal weapons in the attack against such a position.

  First, we shall give simple examples of two typical operations: namely, the elimination of the fianchettoed bishop and the breaking up of the pawn structure in front of it.

  The first position is from the game Richter-Abramavicius, Hamburg 1930. It is White’s move, and he sacrifices both his rooks for the fianchettoed bishop in order to be able to conclude the game by an attack on the weakened dark-square network.

  1 Rh8+ Bxh8 2 Rxh8+ Kxh8 3 Qh1+ Nh7 4 Nf6 Kg7

  Or 4 ... Qxf6 5 gxf6 Bxd3 6 Qh6 Rxc2+ 7 Kd1 Rg8 8 Ng5 and mates.

  5 Qh6+ 1-0

  The following position occurred in the game Szilagyi-Szabo, Budapest 1946.

  Black, to move, broke up the pawn chain h2-g3 by a sacrifice and proceeded to decide the game on a network of squares of the opposite colour to that of the fianchettoed bishop.

  1 ... Rxh2! 2 Kxh2 Ng4+ 3 Kg1 Bxg3 4 Re2

  White loses, whatever the continuation. because of his weakness on the corresponding focal-points h2 and f2. Mate can be postponed only by a severe ‘letting of blood’. After 4 Nf3 exf3 5 Bxf3 Qh4 6 Re2 Bf2+ 7 Kf1 Nh2+ 8 Kg2 Qg3+ 9 Kh1 Qxf3+ Black still mates with 10 Kxh2 Qg3+ and 11 ... Qh3#.

  4 ... Qh4 5 Nf1 Bf2+ 6 Rxf2 Qxf2+ 7 Kh1 Qb2 0-1

  For White loses either his rook or his queen.

  Where there is a fianchettoed king position there is naturally no opportunity to sacrifice the bishop on h7; moreover, many other forms of sacrifice, typical enough in the case of normal castling, are out of the question against the specific structure of the fianchetto. A form of sacrifice which can be designated as characteristic in the case of a fianchettoed king position is that of the knight on f5. An example of such a sacrifice is provided by the next diagram.

  This position arose in the game

  Alexander-Pachman, Hilversum 1947 . Against Black’s fianchettoed king position White has a knight aggressively placed at g5, a ‘little avalanche’ of pawns on g4 and h4, a rook on g1 and his queen covering some useful squares. Nevertheless, to ‘unlock’ Black’s solid position a sacrifice is necessary.

  1 Nf5!

  Given the favourable supporting conditions, this typical ‘anti-fianchetto’ sacrifice is absolutely correct. From f5 the knight controls the squares g7 and h6 and has the intention, above all else, of eliminating the fianchettoed bishop. If Black does not accept the sacrifice but plays, for example, 1 ... Rae8, White continues 2 Nxg7 and then 3 f4 with good prospects, while if 1 ... Bf6, then 2 Nh6. The acceptance of the sacrifice leads at once to a critical situation for Black as the reply gxf5 leads to a very dangerous attack, while the f5-square is well covered, and consequently there is no danger of Black making a counter-sacrifice on it.

  1 ... gxf5 2 gxf5

  In positions of a similar nature one is often faced with the question whether it is better to take on f5 with the g-pawn or the e-pawn. The answer depends on the various characteristics of the position; one takes with the g-pawn, if the opening of the g-file or the diagonal d1-h5 brings with it a decisive advantage, while in playing exf5 one takes into account other possible advantages, namely, freeing the square e4 (on which a knight, for example, can take up a strong position) or opening the b1-h7 diagonal for the bishop, or perhaps to keep in reserve the continuation g5. In this particular case the g-pawn is obviously indicated, since there is an immediate threat of Qh5.

  2 ... f6

  If 2 ... h6, then 3 Qh5, threatening 4 Nxf7+ as well as 4 f6. If 2 ... Ng8, Black does not get as far as consolidating by ... Nf6 because of 3 Nxh7 Kxh7 4 Rxg7+ Kxg7 5 Rg1+.

  3 Nxh7 Be8

  The only defence as the queen must not be allowed to reach h5.

  4 Rxg7!

  The fianchettoed bishop’s head has to roll!

  4 ... Kxg7 5 Nxf8 Kxf8 6 Bh6+

  When the fianchettoed bishop vanishes from the board, its opposite number usually becomes extremely powerful.

  6 ... Kf7 7 Qh5+ Ng6

  If 7 ... Kg8, then 8 Rg1+ Ng6 9 Rxg6+ Bxg6 10 Qxg6+ Kh8 11 Qxf6+ Kh7 12 Qg6+ Kh8 and either 13 f6 or 13 Bg5.

  8 fxg6+ Kg8 9 Qf5 and White won in a few moves.

  The importance of removing the fianchettoed bishop at the correct moment in this kind of attack is shown by the next position, one which I reached in an analysis of the Dragon Variation of the Sicilian in my book, Contemporary Opening Theory.

  In my book I examined 1 Qh4, following it up by 1 ... bxc3 2 Qh8+ Bxh8 3 Rxh8+ Kg7 4 R1h7#.

  However, fianchettoed king positions of this kind (with the defensive bishop, the knight on f6, and a pawn on e7) offer a great deal of resistance. Consequently, in spite of the thunder on the h-file, Black can answer 1 Qh4 with 1 ... Nh5! 2 Bxg7 bxa3! 3 Qg5 (if 3 gxh5, then 3 ... Kxg7, while if 3 Bd4, then 3 ... axb2+ 4 Kb1 Qa5) 3 ... axb2+ 4 Kb1 Kxg7, and now if 5 Rxh5, then 5 ... Ba2+ 6 Kxa2 b1Q+, while if 5 gxh5, then 5 ... Qxc3 is in Black’s favour.

  JN: In this line 4 ... Qxc3 is immediately decisive, since 5 Bxc3 Ba2+ leads to mate.

  Because of this I have had to amend my original analysis of the diagrammed position and apply more drastic measures to eliminate the fianchettoed bishop.

  1 Qh6!!

  Clearly, 1 ... Nh5 will not do now because of 2 Qxg7+, while if Black tries to parry the threat of 2 Qh8+ by 1 ... Kf8 2 Qh8+ Ng8, White mates by 3 Qxg8+ Kxg8 and 4 Rh8+.

  1 ... Bxh6+ 2 Rxh6 g5 3 Bd3!!

  After 3 Rh8+ Kg7 4 R1h7+ Kg6 White only has perpetual check, while if 3 Rxf6, Black can reply 3 ... Bc4 4 Rh8+ Kg7! 5 Rfh6+ e5.

  3 ... Bxg4

  Black is also mated after other continuations, e.g.:

  1) 3 ... Bc4 4 Rh8+ Kg7 5 R1h7+ Kg6 6 e5+ Bxd3 7 Rh6+ Kg7 8 exf6+ exf6 9 Bxf6#.

  2) 3 ... d5 4 Rh8+ Kg7 5 R1h7+ Kg6 6 exd5+ Ne4 7 Rh6#.

  3) 3 ... Nh7 4 Rxh7 f6 5 Rh8+ Kf7 6 R1h7+ Kg6 7 e5+ f5 8 gxf5+ Bxf5 9 Rh6+ Kf7 10 e6+ Bxe6 11 R6h7#.

  4 Rh8+ Kg7 5 R1h7+ Kg6 6 e5+ Ne4 7 Rh6+! Kf5 8 Bxe4+ Kf4 9 Ne2#

  The attack on the queenside castled position

  Before discussing the attack on the queenside castled position, one should first of all examine the differences between this type of castling and kingside castling. Some writers in the past have supposed that the difference is considerable and that castling long ‘is much weaker than castling short’; this is a great exaggeration. The castled position after long or queenside castling differs from the short castled position only in this respect, that the positions of the king and rook are both a square nearer the centre of the board. This can be useful for the rook, but it usually presents difficulties to the king, which is safer on g1 than c1. Reckoning in terms of tempi, the following picture sums up the differences: the position on the queenside which would correspond symmetrically to the one given by short castling (king on g1 and rook on f1) is king on b1 and rook on c1, and to reach such a position after castling long would require two tempi. But let us now apply a different method of calculation: the player has castled on the kingside and then spends one tempo on the move Re1. The corresponding position to this on the queenside would be king on b1 and rook on d1, and for this only one move is needed after castling long. The conclusion to be drawn from this is simple whenever the rook is well placed on the d-file, castling long is fundamentally as good as castling short; if the rook is in fact needed on the c-file it is weaker by two tempi, while if the rook is needed on a third square it is weaker by only one tempo.

  In practice castling queenside usually occurs when the rook is well placed on the d-file, and indeed the need for it to be placed there is often the prime reason for adopting long castling. When reckoned in these terms, the practical difference between long and short castling disappears or becomes insignificant. There is yet another difference between the two forms of castling, but it is secondary and concerns the activity of the queen. However,
here too the advantages and disadvantages tend to cancel each other out. Thus one can say that castling on the queenside is almost equal in status to castling on the kingside in general terms of tempo and security. If there are still players who doubt the security of castling on the queenside, it might be suggested that their reasons for this are probably of a subjective, rather than an objective, nature. That is to say, many players remember some bad experiences which they have had with castling long and are not aware of a very common mistake, one which is repeated over and over again: they accept the tempo gained on the d-file by castling long as a gift, without wanting to pay the price for it; they neglect the consolidating move Kb1, and then, when their opponent suddenly sweeps down, it is already too late and there is no time left. The results, of course, are blamed on queenside castling.

  In master chess castling on the kingside predominates, but this is not the result of some fundamental weakness of castling long; the reasons are of another, structural, kind. Castling on the queenside requires the development of three pieces, while in the case of castling short only two need be moved to clear the back rank. Another important point is that in order to castle on the queenside the queen has to be moved first, and the best moment for doing this in many openings occurs rather later, as moving the queen at an early stage of the game can turn out to be a loss of a tempo. This is an important reason for the comparative rarity of castling on the queenside, and the trend is given further impetus by the structure of modern openings, the majority of which are directed towards a struggle over the central squares. In this the c-pawn plays an important part both in covering the d-pawn and in effecting a lateral attack; this delays the queen’s development on the one hand, and also slightly weakens the queenside on the other, consequently reducing the circumstances which favour castling long.

  There are quite a few situations in the openings when castling on the queenside can be usefully applied, and in some cases one can say that it in fact represents the best solution to the basic problem of building up the position. We shall now list the main factors which would usually endorse the choice of long castling in a game:

  1) The pawn position on the kingside has already been weakened, while that on the queenside is sound; this may incline the player towards castling long. (If castling on the kingside is quite impossible or is obstructed, then naturally there is no choice at all.)

  2) The opponent’s forces (pieces or pawns, or both) are favourably placed for an attack on the short castled position, but unfavourably for one on the long castled position.

  3) The kingside is not sufficiently developed, while on the queenside the back rank is already clear. If this is the case when the situation demands castling as soon as possible, then castling takes place on the queenside.

  4) One’s opponent has castled on the kingside, and the weaknesses in his castled position can only be properly exploited by means of an assault by one or more pawns. If in such a case castling on the queenside is readily available and there are no other factors militating against it, one castles long rather than short. An assault by the h-pawn with the aim of opening up the h-file for the rook is often the answer to a fianchettoed king position, and in this case too castling on the queenside is indicated.

  5) A rapid deployment of the rook on the d-file gains an important tempo or some other advantage, and in such a case castling long can afford an excellent solution.

  The factors listed here are only intended as general hints for the player, and the particular characteristics of the actual position should be surveyed each time one makes a choice between castling moves. Black, particularly, should be careful if he intends to castle on the queenside, since experience has shown that the necessary conditions for long castling more often arise for White than for Black. When Black’s long castling coincides with short castling by White one gets a case of opposite-side castled positions; this leads to an intensification of the game, which naturally favours the better developed side, and this in the majority of cases is White.

  However, caution over deciding to castle on the queenside should not be overdone, since many cases arise where castling long is indicated but where the player does not do so because a prejudice about its risk deters him. Here is an example of unprejudiced castling on the queenside from the game Keres-Reshevsky. World Championship Tournament, The Hague/Moscow 1948:

  Black is to move; his position is apparently inferior, since he is threatened not only by Nxd5 but also Nxb5. Besides this, he has not yet castled, so that he is faced with a double task: to discover a reply to White’s threats and also make the time to castle. Here is how Reshevsky solved the problem:

  1 ... g5 2 Bg3 Qb7!

  Black has already shown by 1 ... g5 that he does not intend to castle on the kingside, and now he gives up his pawn on d5 in order to use castling long to obtain an important tempo and gain control over the central squares d4 and d5 (2 ... Ne7 would be weak because of 3 Nxb5).

  3 Nxd5 0-0-0!

  Black’s castling on the queenside in the Open Variation of the Ruy Lopez (which was how this game began) qualifies as a curiosity, and many masters shook their heads when they saw what Reshevsky had done. In this case it was correct to overcome the prejudice about the unsuitability of castling long in the Ruy Lopez, for here this very move provides a happy solution to the problem: it gives Black the time he needs and consolidates his king position better than short castling would. It should also be noticed that the weaknesses at a6 and b5 are not particularly marked, since White does not possess the move a4 which would properly probe them. Another point is that castling long has lent power to Black’s g-pawn; its advance to g4 is important in the struggle for the square d4.

  4 Nf6

  The knight here is ‘attacking empty squares’, but 4 Nc3 would have been no better. The most important point is that White cannot play 4 c4 to support the knight on d5, since it would be answered by 4 ... g4 5 Nh4 (or 5 Ne1 Nd4 6 Qd3 Bf5) 5 ... Nd4 6 Qd3, after which either 6 ... c6 or 6 ... Bxd5 is in Black’s favour.

  4 ... g4 5 Ne1 Nd4 6 Qf1

  If 6 Qe4, then 6 ... Qa7!, threatening 7 ... Bf5.

  6 ... h5

  The rest of the game is not of interest to us here; Black’s active piece play serves as sufficient compensation for the pawn, there is no question of his castled position being in danger, and it is really White who is in difficulties. In the end White did in fact lose the game after making some rather weak moves.

  Our further consideration of the attack on the long castled position will be confined to those features which are specific to queenside castling. This means, in the main, that we need only consider positions where the king is still on c1, since if it has already retired to b1 a position has arisen which corresponds symmetrically to that after castling kingside.

  This position is taken from the game Rabinovich-Romanovsky, Moscow 1925. Black, to move, has no way of preventing exf6 and in desperation plays for a counterattack on the enemy king based on the pattern of Boden’s Mate.

  1 ... Ba3! 2 exf6?

  White, who has in fact the better of the game, here takes the wrong path and spends a tempo on ‘burying the corpse on f6’ instead of using it to consolidate his threatened king position, e.g. 2 Rxd3! Rxd3 3 Kb1! and 3 ... Qc5, then 4 Na4 and finally 5 exf6, etc.

  2 ... Qc5 3 Bd2?

  After this second mistake Black has a winning position. White should have played 3 Rxd3 Rxd3 4 Kb1! and if 4 ... Bxb2, then 5 Na4, or if 4 ... Rxc3, then 5 bxc3!, when Black can probably draw but cannot achieve more.

  3 ... Bg6!

  The precious bishop retires and thus stops White playing Ne1 followed by Nc2 with gain of tempo.

  4 Qa4

  While this does not give his opponent much trouble, White’s other moves are also of little use. The situation is now typical of attacks on the queenside castling position and a similar arrangement of pieces often arises in practice. So let us look at the whole repertoire of combinations at Black’s disposal which lead to mate.

 
First, it is clear that White could not play 4 bxa3, because of 4 ... Qxa3#, with the typical pattern of Boden’s Mate. Next, it can be seen that if 4 Qf4, Black has 4 ... Rac8, threatening 5 ... Qxc3+ and mate, thanks to the formidable effect of Black’s bishops. It is true that White could meet Rac8 with 5 Qe5, but then 5 ... Qb4 would be decisive. There still remain some other defences, which are worth setting out:

  1) 4 Rde1 Rxd2! (if 4 ... Rac8, White would reply 5 Kd1) 5 Kxd2 (5 Nxd2 is also inadequate, e.g. 5 ... Qxc3+ 6 Kd1 Qc2+ 7 Ke2 Qd3+ 8 Kd1 Rc8 9 Re4 Bb4 10 Rxb4 Rc1+ 11 Kxc1 Qc2#) 5 ... Rd8+ 6 Ke2 Bd3+ 7 Kd1 Be4+! 8 Ke2 (or 8 Nd2 Rxd2+ 9 Kxd2 Qd4+ and mates) 8 ... Qc4+ 9 Ke3 Rd3+ and Black mates.

  JN: Black can mate in two by 6 ... Bc2+, while 4 ... Rac8 also wins as after 5 Kd1 Bxb2 White is defenceless.

  2) 4 Rhe1 Rac8 5 Re3 Qb5! (yet another threat arising from the position of the rook at c8!) 6 bxa3 Qb1#.

  3) 4 Ne1 Rd4 5 Qg5 Qb4 and mates.

  4) 4 Be1! (relatively the best method of defence) 4 ... Rxd1+ (if 4 ... Qf5 5 Rxd8+ Rxd8, White has the reply 6 Qe4!) 5 Kxd1 Rd8+ 6 Bd2 (if 6 Kc1, 6 ... Rd3 is decisive) 6 ... Bxb2 7 fxg7 Rc8 and White must either give back the piece or consent to 8 Na4 Qc2+ 9 Ke2 Qd3+ 10 Ke1 Bc1! 11 Qb4 a5 12 Qf4 Rc4, when Black wins.

  4 ... b5 5 Qxa3

  5 Qxb5 is defeated by 5 ... Rab8! 6 Qxc5 Bxb2#. Still another of the numerous mating patterns in this kind of position!

  JN: Vuković does not mention the far more resilient defence 5 Qb3, when Black has to play very precisely to retain an advantage: 5 ... b4 6 Be3 Qc7 7 Rxd8+ Rxd8 8 Nd4 Rb8! 9 bxa3 bxc3 10 Nb5 Qc6 11 a4 a6 12 Qxc3 Qxg2 13 Rd1 axb5 and White’s king remains extremely exposed.