Again we begin with no possible checks for White and no loose Black pieces (except the queen, which makes a bad target, and bishop, against which nothing is possible). Still, we do have some familiar geometry: Qd5 aims the queen at both the Black king and the rook at a8. Neither side of the attack yet works, but the basic pattern is something to keep in mind. The next step is to experiment with checks (White has none) and with captures. The obvious spot for an exchange is f5, where Black has a knight that White attacks. How many times is it attacked, and how many times protected? Twice and twice. So White imagines 1. NxN, BxN; 2. BxB, RxB. Here is the key moment in the exercise: imagine the board after those exchanges, and re-ask what checks and loose pieces are available. Now Qd5 looks quite different. It’s a check, and the rook at a8 would be loose and so would be lost to the double attack. This problem is a good illustration of two things: how much a couple of exchanges that don’t seem to do much by themselves can change the board, and the importance of asking simple questions about how the board will look after such exchanges are complete.
Take the opportunity of this position to think, too, about the coordination of Black’s pieces—and White’s disruption of it. In the diagram Black’s pieces protect each other nicely: his rooks guard each other, and his bishop guards both knights and prevents White from giving check on d5. With a couple of exchanges White ruins this coordination, suddenly leaving Black’s position much more exposed to brutal tactical shots. It shows how vulnerable a position can be when pieces are protected by other pieces (rather than pawns). A series of captures can force the pieces all over the board and rather abruptly leave one or two of them loose.