Structure & the Fair-Play Contract
A detective story is two stories stacked: the crime that already happened, and the investigation that uncovers it. The spine is always the same — Crime → Investigation → Reveal — and the writer’s job is to hide the first inside the second so precisely that the solution reads as surprising yet inevitable. This module gives the agent the load-bearing structure and the one non-negotiable rule underneath it: do not cheat the reader.
§ 1.1 Map three acts onto the investigation arc
Adapt three-act structure to the mystery spine rather than importing it whole. The beats:
- Act 1 — Setup (~first 25%): establish the sleuth’s normal world, deliver the inciting crime (body or discovery, usually around the
10–15%mark), assemble the initial suspect pool, and fix the detective’s stakes. - Act 2 — Investigation (the long middle): run try/fail cycles — leads, interviews, clues, red herrings — where each new fact complicates rather than simplifies.
- Midpoint reversal (~50%): the false high / false low. Everything the sleuth believed is overturned, forcing a new lens; re-examine and re-interview under it.
- Act 3 — Reveal (final ~25%): narrow to one suspect, stage the climactic confrontation, deliver the solution, and resolve the subplots.
Mapped back to the spine: Act 1 exposes the Crime, Act 2 is the Investigation, the midpoint is its hinge, and Act 3 is the Reveal. Write the crime’s full backstory first as a straight timeline, then design the investigation as its mirror-image uncovering: Discovery → Early Investigation → Complications → Revelations → Climax → Resolution.
§ 1.2 Honour the fair-play contract
The genre runs on a contract: the reader must have access to every clue the detective has. Ronald Knox’s Decalogue (codified 1929) and S.S. Van Dine’s “Twenty Rules for Writing Detective Stories” (The American Magazine, 1928) spell out the specifics. Many individual rules are dated; the enduring core is the single line beneath them — thou shalt not cheat thy reader.
Treat the rules as a historical scaffold to understand and then break intelligently. Christie broke both “the detective must not be the criminal” and the narrator rule in The Murder of Roger Ackroyd — and still played fair, because the clues were on the page. P.D. James gives the clean modern summary: a central mysterious crime (usually murder), a closed circle of suspects with means, motive, and opportunity, a detective (amateur or professional), and by the end a solution.
The detective can know nothing which the reader isn’t also told. P.D. James, Talking About Detective Fiction, 2009
The break is licensed only when the fairness invariant survives it. Novelty in the mechanism, never in the disclosure.
§ 1.3 Plant clues for surprising-yet-inevitable
A clue must be visible and camouflaged at once. Three techniques carry the load:
- Bury it in the mundane — fold the decisive detail into ordinary description so it passes as texture.
- Misdirect the interpretation — give accurate information the reader will attach to a wrong assumption.
- Use the incongruous detail — a small wrongness the reader registers but does not weigh.
Two boundary rules make the contract auditable. First, the reveal has a defined opening: the first scene in which the detective acts on a settled solution. A clue first shown at or after that point is not planted — it is disclosure during the verdict. Second, evidence gathered off-page and first reported in dialogue (“a neighbor saw two cars”) counts as planted only when it breaks a false lead; every clue load-bearing for the culprit’s guilt must be gathered in scene, on the page, where the reader can weigh it before it is used.
Apply the Rule of Three: introduce an important detail three times in different contexts so it registers subconsciously without announcing itself. Calibrate against a hard target — a careful reader could solve it roughly 2–3 chapters before the reveal; most will not, but on finishing all should feel I should have noticed that. Avoid the blinking-neon “this is a clue”; equally avoid burying it so deep it cannot be retrieved on a reread.
§ 1.4 Deploy red herrings without cheating
The fairness test for a red herring is mechanical: on a reread, could a careful reader distinguish the false lead’s clues from the true solution’s? If not, it cheats. To generate suspicion honestly, give every significant character a secret so they all behave evasively — and decide why each one hides it. Deploy in waves: resolve one false lead as you introduce the next; plant each early enough to earn the reader’s investment and close it before the climax. “Resolved” is a two-part test, not a vibe: the reader has been given the innocent explanation (the secret itself, disclosed) and understands why it produced the guilty-looking behaviour. Dropping the murder suspicion while leaving the secret undisclosed past the reveal leaves the thread open. How many to run is a genuine dispute: ProWritingAid warns that more than one or two leaves readers feeling cheated, while puzzle-forward writers favour a fuller suspect field. Default when unspecified: two herrings for a character-forward story, a fuller field for a puzzle-forward one.
§ 1.5 Construct the crime backward
Know the who, the how, and the why before writing a line of investigation. Give the culprit a believable, strong motive — usually money, passion, or both. Build the suspect pool by the means/motive/opportunity lattice: hand several suspects two of the three, and reserve all three for the killer. The sleuth’s entire job is to find who holds the full set. A morally grey victim with messy relationships helps here: every grudge the victim earned in life is another plausible suspect.
§ 1.6 Subplots and the B-story
The B-story is typically a relationship — love interest, partner, rivalry — that both obstructs and aids the protagonist while carrying the emotional and thematic weight the A-story cannot state directly. It runs alongside the investigation, not underneath it. Resolve subplots before or at the reveal. Naming the killer settles your account with the reader; a subplot still dangling after that reads as a broken promise, not a tease.