The first step in our usual studies of the discovered attack has been to find the kernel of it and build from there. But what if the kernel doesn't yet exist? Then a bit more imagination is required: you need to experiment with moves—here, moves by bishops and the pieces they can mask—that create kernels; you learn to notice when a move or exchange you might make will bring the kernel of a discovery into existence. Here are some illustrations.
In the case shown here, White sees that he has a rook on the open e-file. He has a bishop that can reach the same file with Be7, masking the rook. Look at the position of the bishop and rook here; see how they thus are one move away from being in the classic formation for a discovery. The question is whether the move that creates the kernel also forces a target into place. Here the bishop’s move to e7 would threaten Black’s rook on f8. Black’s reply would be forced and thus easy to foresee: the rook would have to move to e8. This little sequence would create a working discovered attack: a masked piece, a masking piece, and a target all along the e-file. Now the bishop just needs a good place to go next. Bb4 threatens Black’s queen; and if Black replies QxB, White has RxR+ followed by RxQ when Black interposes his queen on f8. Notice how easy this position would be if it started with the kernel intact (i.e., with 1. Be7, Re8 already played). The challenge is just to think of White moving his bishop in front of his rook and to carefully consider the result.