Deflection is the art of firing the defender. Some enemy piece is holding everything together — guarding the back rank, defending the queen, covering a mating square. Force it to leave its post, or give it a job it cannot refuse elsewhere, and the position falls apart. Solve the featured real-game puzzle below, then practice deflections in the trainer at your own level.
The mechanism is usually forcing. A rook lands on the eighth rank and must be captured — but the capturing rook abandons the queen. A queen offers itself to the piece guarding a mating square. If the defender takes, the real threat lands; if it refuses, the sacrifice often just wins material anyway. Most back-rank mate combinations are really deflections wearing a crown. The deflection glossary entry collects the standard shapes.
When one piece defends two things at once, any forcing move that consumes one duty automatically abandons the other. Count the jobs each defender holds — a piece with two jobs is doing neither safely. That counting habit is the practical skill these puzzles install: before dismissing a sacrifice as crazy, ask the deflection question — what does this piece defend, and what happens the moment it stops?
Deflection drags a defender AWAY from a square or duty it must keep; a decoy lures a piece — often the king — ONTO a bad square where a follow-up hits it. The two constantly work together: one sacrifice pulls the guard away while the next pulls the king into the net.
A piece with more defensive jobs than it can perform at once — guarding the back rank and a knight, or covering a mating square and the queen. Overloaded pieces are deflection targets by definition: any forcing move that cashes in one of its duties wins the other. Counting defenders' jobs is the fastest way to find combinations.
Calculate both replies. If the defender accepts, your real threat must land with interest; if it declines, the sacrifice should still win material or keep an unanswerable threat alive. When both branches pay, the sacrifice is sound — that double check of accept-and-decline is exactly what these puzzles drill.