Figure 4.5.11.3[White to move]

Look for useful alignments between Black’s pieces. See that his queen and rook are lined up on the fifth rank, and that his queen and knight both are on the d-file. White has nothing he can do with the horizontal alignment, but he has a rook available to send pressure down the d-file with Rd1. White’s rook is safe there because his queen protects it, but does the move accomplish anything? The target of the skewer would be the knight behind Black’s queen, but the queen can protect the knight when it moves by going to b7. Attacking a protected knight with your rook generally is a bad idea. Generally—but not here, because from the outset Black’s knight already is under attack by White’s queen as well. Once Black moves his queen the knight will be attacked more times than it is defended and must be lost. White’s queen does dual service as guardian of the skewering piece and additional attacker of the targeted piece in the rear. The simple lesson of the position: when you have the geometry of a skewer and are deciding whether it can be made profitable, be careful to consider not just whether the enemy pieces involved are loose or have protection, but also whether they already are under attack from other directions.

But now let’s go beyond the simple lesson and think about how the knight will be lost after White plays Rd1. If Black replies Qb7, White has to choose between QxN and RxN. Which is better? QxN, because of how the board looks afterwards: White then has a battery of queen and rook on the d-file, with the queen ready to invade the Black king’s position. Black is obliged to prevent this by trading queens—playing QxQ, and letting White recapture RxQ. That’s fine with White, as he is a piece ahead; his advantage gets larger as other pieces are exchanged away. If White instead takes the knight with his rook, he poses no such threat. Black can play Qb1+ and harass White’s king.

But Black doesn’t have to reply to White’s initial Rd1 by playing Qb7. Another option is Ra5, meeting White’s threat not with defense but with a counterthreat. Black is hoping that White’s queen will move to a square where it no longer attacks the knight on d7. The threat doesn't achieve that goal, but it still turns out better for Black than Qb7; instead play goes 2. RxQ, RxQ; 3. RxN, Rxa2 and Black has traded a knight for a pawn instead of giving up his knight for nothing. (Or it can run 2. QxN, QxQ; 3. RxQ, Rxa2. The result is the same.)

Finally, consider one other way for Black to respond to 1. Rd1: he can play Nb6. Notice the appeal of the move. Again it counterattacks White’s queen; and it also removes the knight from the d-file, preventing White from following RxQ with RxN the way he did in the previous paragraph. The fantasy is that play will go 2. RxQ, NxQ; 3. RxR, NxR—and Black is even. Unfortunately this all fails because White replies to Nb6 with Qe8+. (When Black moved his knight, he inadvertently opened a line to his back rank for White’s queen.) Now Black’s king and queen are both being attacked; after Black moves his king, he will lose his queen in return for a rook. So Nb6 is a disaster, but it’s still important to see. If his king were on h7 here rather than g8, Nb6 would save the position for Black. Play would go as described in the fantasy sketched a few sentences ago. The point: a relative skewer can be broken just as a relative pin can be broken if one of the pieces being skewered can go make a serious enough counterthreat of its own. And this brings us to our next topic.