Verified against ChatGPT · 2026-08-14
Construct a fair-play mystery plot where an attentive reader could solve it, but almost nobody does
Builds a mystery plot with a clue ledger tracking exactly what's genuine evidence, what's a red herring, and what misdirection is doing the work of hiding the solution in plain sight.
The prompt
Ready to copy — highlighted parts are example details you can swap.
You are constructing a fair-play mystery plot — one where the solution is genuinely deducible from clues available to the reader, but where misdirection makes it unlikely most readers actually get there first. CRIME AND VICTIM A wealthy vineyard owner found dead after apparently falling from a cellar staircase; ruled accidental until the detective notices an inconsistency. DETECTIVE OR INVESTIGATING POV A visiting wine critic who happened to be staying at the vineyard, an amateur with sharp observational instincts but no formal authority. TRUE CULPRIT AND REAL MOTIVE The vineyard's longtime accountant, who had been quietly embezzling for years and killed the owner after being confronted, staging it to look like a fall. NUMBER OF RED HERRINGS 3 SETTING An isolated vineyard estate in rural France during harvest season, with a small, closed circle of live-in staff and family. STEP 1 — Work backward from the solution Starting from The vineyard's longtime accountant, who had been quietly embezzling for years and killed the owner after being confronted, staging it to look like a fall., identify every piece of physical evidence, testimony, or behavior that would genuinely exist in this scenario as a direct consequence of what actually happened — this is the pool of fair clues, and every one of them must be something the culprit could not have avoided leaving behind given how the crime was actually committed. STEP 2 — Build the clue ledger For every fair clue identified in Step 1, decide how it will be presented to the reader: in plain sight but misattributed to a different suspect, buried among unrelated detail so its significance isn't obvious on first encounter, or stated accurately but interpreted incorrectly by the POV character in a way the reader is likely to accept. For each of 3 red herrings, build one that points toward a different, innocent suspect who has a genuine, plausible-looking reason to have done it — a red herring only works if the wrong suspect actually looks guilty on their own merits, not just coincidentally present. STEP 3 — Fairness audit Check the full ledger against one rule: could a reader who correctly noticed and correctly interpreted every fair clue arrive at the true solution using only information given on the page, without requiring a fact revealed for the first time in the final reveal? Flag anything that fails this test, since a mystery that depends on withheld information isn't fair-play, it's a trick. OUTPUT FORMAT 1. The clue ledger: each fair clue, how it's presented/disguised, and where it appears in the story. 2. Each red herring and the innocent suspect it points to, with their plausible-looking reason. 3. The fairness audit result, including anything flagged and fixed.
Customize
Optional — swap in your own details for the highlighted parts above.
Why this works
Asked to build a mystery, GPT-5.1 will readily generate a plausible-sounding solution and a set of clues, but without an explicit backward-construction step it tends to generate clues and solution somewhat independently, which is exactly how unfair mysteries happen — a clue that supports the solution only in retrospect, or a reveal that depends on a fact never actually shown to the reader, which is the single most common complaint against amateur mystery writing. Working backward from the solution and requiring every fair clue to be something the culprit 'could not have avoided leaving behind' forces logical consistency between what actually happened and what evidence exists, rather than generating evidence that merely sounds suggestive without a real causal link to the crime as committed. The instruction to build red herrings around suspects who look genuinely guilty on their own independent merits, not just coincidentally present, targets a specific weak-mystery pattern where misdirection relies purely on narrative suspicion-casting (a suspect acts nervous, for no clear reason) rather than the suspect having their own real, freestanding reason to look guilty — the latter is what makes a red herring satisfying to have been fooled by rather than annoying in hindsight. The final fairness audit — checking whether a reader who noticed and correctly interpreted every fair clue could reach the solution using only on-page information — is the mechanism that actually enforces the fair-play constraint, since it's a specific, checkable test rather than a vague aspiration to 'make it fair,' and running it as an explicit final pass catches the common case where an otherwise well-built mystery quietly depends on one piece of information that's only ever revealed in the final chapter.
What you get back
Fair clue: the accountant's ledger entries for the past two harvests show unusually round numbers in a business that never otherwise produces round numbers — presented on the page as a background detail the critic notices only because she happens to be numerically minded, not flagged as significant by the narrative itself. Red herring: the owner's estranged son, who inherits the estate and had a public argument with his father the week before...
Verified against
ChatGPT GPT-5.1 · 2026-08-14
Changelog
- 2026-08-14 — Initial publish, verified against ChatGPT GPT-5.1.
Need this built into your business?
If a prompt isn't enough — what Scult builds, built and maintained for you — that's Scult's day job.
EXPLORE WHAT SCULT BUILDS
