Black’s bishop and d4 pawn are on the same diagonal, signaling a potential discovery. (Notice the recurring visual pattern: a pawn in the center and a bishop aimed through it and toward the enemy king.) The pawn can step out of the way and attack White’s queen at the same time. The problem is becoming familiar in structure: the pawn on f2 would block the bishop, preventing it from giving check. So remove the pawn by taking something it protects—the g3 pawn, which Black can capture with his rook. But there is a flaw: consider whether the recapture you want would be compulsory or could be made by a different piece; be aware of how many times any piece you might take is protected. Here Rxg3 is met by White not with f2xR but with NxR. So what to do? Just push the plan back a step and start by taking out the knight. Thus Black plays BxN; and now if White recaptures with BxB, Black is free to play Rxg3+. White then would have to reply f2xg3 (if he instead tries Kh2, White looks for his next check and finds Qxh4, which mates after a useless interposition by White's bishop from f5). The board is set up for the discovered check d4-d3+, winning the queen.

The probable result of all this is that Black simply takes White’s knight for free at the outset; this is an example of the power of working backwards from an idea, examining the obstacles to its success and considering how each might be removed. The threat can lead to gains elsewhere on the board, since the discovery hanging in the background leaves other enemy pieces more vulnerable than they seem.