Observe the offensive tension on the board. Black would like to use his queen to take White’s rook on f5, but then he would lose his own rook to White’s queen. How to break the jam? By use of the queen-rook-queen alignment on the e1-a5 diagonal—the three-piece kernel of a discovery. White’s queen is ready to be taken if the rook can vacate c3 with sufficient violence. The rook has no way to give check, but study White’s king: it’s trapped on the back rank. If Black could land a rook there he might have a mate threat—except for the protection furnished by White’s rook on b1; if Black could take that rook with his own, he would indeed threaten mate. So the threat by Black of RxR is almost as good as a check, and Black can achieve this with 1. ...Rb3.

What next? White is confronted with multiple threats, so he might naturally look for a way to defuse them both—a square where his queen would be safe and would defend against RxR#. One possibility is 2. Qc1. This takes care of those two problems, but not a third one: Black now can play QxR with no worries, because White’s queen no longer is in position to retaliate by taking the rook on c3. (But before playing QxR Black has to exchange rooks on b1; otherwise White meets Black’s QxR with c2xR.) White is left with a queen against Black’s bishop and queen. A better reply idea for White would be 2. a2xRb3. Black then plays QxQ next move, so this time White has lost a queen for a rook; but at least he ends up with two rooks against Black’s queen and bishop.