The Black king’s position looks well-fortified, but White’s pieces are arranged powerfully enough to force mate in three moves. His knight is on g4, a great square since it allows Nh6 or Nf6, both of which are brutal attacking moves. He has two attackers that can use the long dark diagonal: his queen and his dark-squared bishop. The g7 pawn is what holds Black’s position together; it blocks the dark diagonal and protects f6 and h6 against occupation by White’s knight. Put these factors together and they should suggest a classic bishop-and-knight mate, made possible by a queen sacrifice to open g7: 1. Qxg7+, KxQ; 2. Bf6+, Kg8; 3. Nh6#. Notice these additional virtues of a knight on g4: not only that it can jump to h6 and f6, but that it can protect a bishop as it arrives on one of those squares as well, while sealing off the other as a flight square.

And notice as well the crucial work done by the pawn on h5. Without it, Black’s second move is Kg6 and the immediate threat is over. The offensive power of a pawn advanced past the fourth rank should not be underestimated. Such pawns can provoke your opponent into captures that create valuable open lines; they can support attacks by pieces on the squares they protect; they can deny your opponent squares where he would like to place defenders; and they can seal off important flight squares for the enemy king.

Presumably you realize that if Black’s g7 pawn instead were on g6 at the start of this position, White would mate immediately with Nh6. But if you instead imagine the h7 pawn on h6 at the start, the sequence no longer works. Black’s second move is Kh7 and again his king escapes. The lesson is to be careful not to use mating patterns (or any patterns) mechanically. They are meant to suggest ideas, but you still have to test the idea as it will play out in your position. These mates that involve trapping the king on squares in the corner are a good example, as they depend on the king’s limited range of motion. If it has flight squares, all of them must be accounted for.