Everybody likes to attack in chess – it's fun and exciting. From my experience, that can often be used against an opponent. Many players will continue to attack even when they would be better off thinking about more peaceful methods.

In particular, a rook lift to the kingside is a dangerous attacking plan that also burns some bridges. After all, if the kingside attack does not succeed, the rook is simply out of play. There are many examples of the defending side launching a strong counterattack or transposing into a good ending based on the poor placement of an "attacking" rook out of play for other operations.

The following game made a great impression on me when I first encountered it.

 E. Geller vs. M. Euwe
 1953 Zurich Candidates Tournament
 Nimzo-Indian Defense, Delayed Saemisch Variation

1.d4 Nf6 2.c4 e6 3.Nc3 Bb4 4.e3 c5 5.a3 Bxc3+ 6.bxc3 b6 7.Bd3 Bb7 8.f3 Nc6 9.Ne2 0-0 10.0-0 Na5 11.e4 Ne8

This is a key move in black's defensive scheme. If white is allowed to play Bg5, the pin would be extremely uncomfortable. The knight retreat also allows black to play ...f6 if necessary to block the advance of the f-pawn and give the black king an escape square.

12.Ng3 cxd4 13.cxd4 Rc8 14.f4 Nxc4

It is common in this line for white to sacrifice his weak c4-pawn for kingside attacking chances. It is clear that white will now aim at the kingside, and black will bank upon his extra pawn and play on the queenside.

15.f5 f6 16.Rf4 b5!

Black starts a counterattack involving his queenside forces. Other than the presence of white's d-pawn, the position could conceivably arise from a Sicilian Defense.

17.Rh4 Qb6 18.e5 Nxe5 19.fxe6 Nxd3 20.Qxd3 Qxe6 21.Qxh7+ Kf7 22.Bh6 Rh8!!

This is the culmination of black's play, sacrificing a rook to decoy the  queen from control of the c2-square.

23.Qxh8 Rc2 24.Rc1 Rxg2+ 25.Kf1 Qb3 26.Ke1 Qf3, 0-1.

This game deservedly won a brilliancy prize in a tournament filled with worthy games.

The following game, while not nearly as spectacular, is based on a similar idea: at the key moment, black lures white forward so that he may counterattack.

David Kerkove (2130) vs. Randy Bauer (2275)
1996 South Dakota Governor's Cup
Sicilian Defense

1.e4 c5 2.Nc3 Nc6 3.f4 g6 4.Nf3 Bg7 5.Bc4 e6

Black's set-up has taken a lot of the fun out of this variation for white. The problem is that 5.Bb5 Nd4! cuts across the white plan to play against doubled c-pawns, while now black is readying ...d5.

6.0-0

The most thematic try, 6.f5, runs into 6...Nge7! 7.fxe6 when black has good chances after 7...dxc6 (solid) or 7...fxe6 (enterprising). I've preferred 7...dxc6 8.d3 0-0 9.0-0 Na5 10.Qe1 Nec6! 11.Bg5 f6 12.Be3 b6 with good play for black (Mandeville-Bauer, Iowa 1991 and Kosanovic-Damljanovic, Belgrade 1993).

6...Nge7 7.d3 d5 8.Bb3 0-0 9.f5!?

This is a typical method for white in this line (see above). From my database, however, the only move in master play from this position is 9.Qe1. Then black has a variety of playable moves, with 9...Nd4 the most popular. I would probably prefer 9...Na5.

9...dxe4 10.f6

The only consistent response. 10.Nxe4 Nxf5 11.Nxc5?? Bd4+ wins a piece, and 10.dxe4 Qxd1 11.Rxd1 exf5 12.exf5 Bxf5 is clearly good for black.

10...Bxf6 11.Ne4 Bg7 12.Kh1

Again, 12.Nxc5?? loses to 12...Bd4+.

12...Na5

This is a practical choice from black, but 12...b6 is more principled. I decided that I didn't want to let white keep his light squared bishop (as with 13.a3).

Removing the white light-squared bishop is a key part of black's planned method of defense. A good way to deal with white's threats in this type of position is with the pawn formation ...f6 and ...e5. This keeps white's knights out of the dangerous g5 and f6 squares and allows black's light-squared bishop freedom on either the h1-a8 or h3-c8 diagonals. Of course, this would be too dangerous if white's light squared bishop was bearing down on the a2-g8 diagonal.

13.Qe1

Kerkove and I have played a similar position before, and I was pretty sure that he'd continue his pawn sacrifice. Still, 13.Nxc5 is probably best. After 13...Nxb3 14.axb3 b6 15.Ne4 Bb7 black has the two bishops but white has some attacking chances. I'd still rate it slightly better for black.

13...Nxb3 14.axb3 b6 15.Bd2 Nf5

I didn't seriously consider 15...Bxb2, since 16.Qh4 Nf5 17.Qh3 allows white's queen to get to the kingside. Even here black probably has a defensible position, but I saw no need, being a pawn up, to let things get potentially out of hand.

16.Bc3 Bxc3

16...Nd4 17.Nxd4 cxd4 18.Bb4 Re8 19.Nd6 is not what black has in mind. Black doesn't really mind the exchange of bishops. From above we've seen that the black defensive formation with pawns on f6 and e5 is effective in defending against a kingside attack. This is the primary reason that black goes to great lengths to eliminate the light squared bishop – the f6/e5 pawn structure would be too dangerous if white had his light squared bishop.

17.bxc3 Bb7 18.Qf2 Nd6 19.Nxd6

19.Neg5 Bxf3 20.Nxf3 f6 followed by ...e5 gives a similar black position without black keeping the strong bishop. The downside is that 19.Neg5 h6 20.Nh3 e5! leaves the knight misplaced on h3.

19...Qxd6 20.Ng5 f6 21.Nf3 e5 22.Qe3 a5!

If we assess the position in terms of imbalances, white can attempt to attack on the kingside. Black has an extra pawn and the chance to create a passed pawn with ...a5...b5...a4. Black starts that plan, and he goads white into moving his pieces to the kingside. Why? Because it will make it harder for white to stop the passed a-pawn.

23.Ra4

This is white's last chance to easily get this piece to the kingside, since otherwise black could play ...b5. Whether white should actually do this is another question, but I was certain that white would make this move.

23...b5! 24.Rh4 a4! 25.bxa4 bxa4

Of course not 25...Rxa4? 26.Rh3 when there is no passed a-pawn.

26.Qh6 Qe7 27.Ra1 Bc6 28.Nd2 Rfb8 29.Ne4 Rb7

Black makes a patient move. He first protects his kingside, then he can push his a-pawn.

30.c4

If white is going to hold, he needs to get his heavy pieces on the h-file back into play. Unfortunately for him, it may already be too late. After 30.Qe3 f5! 31.Ng5?? f4 wins. The best try would be 31.Qxc5, but 31...Bxe4! 32.Qxe7 Rxe7 33.dxe4 f4! 34.Rh3 Rc7! 35.Rd3 Rc4 is a won ending.

30...f5 31.Nc3 f4!

Black has hit upon a winning plan: white's heavy pieces on the h-file are basically out of play for the rest of the game. Compare Euwe-Geller's method for isolating white's pieces.

32.Rg4 Qe6 33.h3 Rb2 34.Nd5 Ra7

Defends from afar, although 34...Bxd5 is also good enough.

35.Re1

Threatening 36.Rxf4! or 36.Nxf4!

35...Bxd5 36.cxd5 Qf6

36...Qxd5 is also good enough, but it requires more calculating because of white's ideas of sacrificing a rook on g6.

37.h4 a3 38.h5 Rg7 39.Ra1 a2 40.Kh2 Rxc2 41.hxg6 Rxg6 42.Rxg6+ hxg6 43.Qh3 f3! 44.Qc8+ Kg7, 0-1.

45.Qc7+ Kh6 and there are no more checks; 45.Qd7+ Qf7 46.Qxf7+ Kxf7, and after an eventual ...fxg2, one of the black pawns on the seventh will queen.