skills/ai-film-pipeline-master/references/phase-01-story-generation/symptoms-information.md

# Information complaints

Part of the symptom router. Open this file when the complaint is about information. Each complaint below is followed by the cards that repair it, best first; `LOOKUP.md` says which file holds each card.

---

### "I don't know how much the audience should know"

This is the first question of every scene, and most drafts never ask it. There are only three positions and they produce different stories from identical events.

1. Audience ahead of the characters — the suspense position
   `info.dramatic_irony` · `info.suspense_shared_knowledge` · `tens.bomb_under_table`
2. Audience level with the protagonist — the mystery position
   `info.mystery_audience_behind`
3. Audience behind, briefly — the surprise position, and the most expensive
   `info.surprise`
4. Audit it properly: draw the grid of who knows what, scene by scene
   `info.knowledge_ledger` · `info.character_asymmetry`

### "The reveal didn't hit"

1. A reveal must land on a character, not on the audience alone — someone in the scene has to be changed by it
   `info.reveal_lands_on_character`
2. Stage it in steps rather than delivering it whole
   `info.staged_reveal` · `info.double_reveal`
3. The strongest reveal re-reads what came before rather than adding something new
   `info.recontextualising_reveal`
4. If it arrives at the midpoint it should change the question, not just raise the stakes
   `info.midpoint_reveal` · `struct.stc_midpoint`

### "The audience guessed it"

1. You telegraphed rather than foreshadowed — you gave the conclusion instead of the materials
   `info.telegraphing` · `info.foreshadowing`
2. Bury the plant inside a moment that is interesting for another reason
   `info.buried_plant`
3. Misdirect by attention, not by lying
   `info.attention_misdirection` · `info.double_meaning_clue`
4. Give them a smaller guilt to find so they stop looking for the larger one
   `info.lesser_guilt` · `info.red_herring`
5. Then stop patching scene by scene and run the procedure that catches it before the next draft does it again. It is the inverse of the knowledge ledger — an account not of what the audience has been *told* but of what they can now *derive*, checked at every structural seam to confirm the derivation you are saving is still out of reach
   `info.deducibility_audit`
6. Two limits on that pass. Do not run it on a first draft; solvable is fine that early, and the audit belongs after the structure is fixed. And never run it on a piece whose pleasure is the audience being ahead — on a tragedy it will systematically remove the thing that makes it one
   `info.dramatic_irony`

### "The twist feels like a cheat"

1. Run the three tests on it
   `info.the_cheat`
2. Characters may lie; the narration may not, unless unreliability was established first
   `info.fair_play_contract` · `info.unreliable_narrator`
3. If the narrator is unreliable, establish it early and by a rule
   `info.naive_narrator` · `info.restricted_narration`

### "The prose explains too fast"

1. Give the reader the sensation before the name for it
   `info.delayed_decoding`
2. Withhold the viewpoint itself where the identity is the information
   `info.withheld_pov`

### "The serial episode doesn't pull to the next one"

1. End on a reveal rather than on an answer
   `info.cliffhanger_reveal` · `tens.suspended_ending`
2. Run a ladder of small reveals so every episode pays something
   `info.micro_reveal_ladder`
3. Open the episode on a reveal, not on a recap
   `info.cold_open_reveal` · `open.episode_opening`

### "The audience knows something I want them to sit with"

1. Hand them a secret the characters do not have and let the ordinary scenes carry it
   `info.audience_only_secret` · `dial.half_truth`

### "I've lost track of what the audience knows"

1. Draw the information map for the whole work before fixing any single scene
   `info.information_map` · `info.knowledge_ledger`
2. Decide the disclosure budget — how much you are prepared to spend and when
   `info.disclosure_budget`
3. Decide deliberately what the audience will never learn, and what stays open
   `info.never_disclosed` · `info.deliberate_unanswered_question`
4. The ledger has a negative half that most drafts never write down — not what they have been told but what they can now work out from it. Run it as a pass rather than a rule, and only after the structure is fixed
   `info.deducibility_audit`

### "The secret isn't doing any work"

1. Price the telling. A secret with no cost to disclose is not a secret
   `info.cost_of_telling`
2. Decide which kind it is: one that binds people together or one that isolates
   `info.binding_secret` · `info.isolating_secret`
3. Give the keeper someone to tell
   `info.the_confidant`
4. Or make it an open secret nobody will name, which is a different and often stronger scene
   `info.open_secret` · `sub.social_prohibition`

### "The audience doesn't believe the claim"

1. Build the credibility ladder deliberately rather than asserting
   `info.credibility_ladder`
2. Choose the proof form: a document, a witness, or a physical object
   `info.the_document` · `info.the_witness` · `info.physical_proof`
3. Let two sources contradict each other and show the work of deciding
   `info.contradictory_sources` · `info.hearsay_evidence`
4. The fastest single move, if you have enough standing for it to cost you something: admit one real weakness, early and flatly. A source that will say one true thing against itself is assumed to be telling the truth about the rest, and it converts an audience from grading claims to weighing an argument
   `info.damaging_admission`
5. Concede on a second-order axis and only on one. A concession on the audience's actual decision criterion does not buy belief about anything else, it ends the conversation — and a concession from a speaker nobody knows yet is just a bad review
6. Across a series or a campaign, stop treating belief as something each piece earns from zero. It is one account: decide in advance which pieces are deposits, which are withdrawals, and where the claim that needs the whole balance is going to sit
   `info.credibility_spend`
7. That ledger does not exist in a format the audience arrives at randomly and leaves — a retail loop, a pre-roll. There every viewer is at balance zero and the belief has to be earned inside the piece

### "The documentary reveal falls flat"

1. Documentary reveals are structured differently — access is the spine
   `info.doc_reveal_structure` · `info.access_as_structure`
2. The admission in an interview is the form's climax; build toward it
   `info.interview_admission`
3. Archive can be a reveal rather than illustration
   `info.archive_as_reveal`
4. And the hardest question: what you withhold from your own subject
   `info.withholding_from_the_subject`
5. Before rewriting anything, check whether the moment is a writing problem at all. Locating the moment that already exists — in rushes, a transcript, an interview, a comment thread — is a separate discipline with its own search strategy and its own judgement, and it is not composing
   `info.finding_the_moment`
6. The call that saves a month: if the material genuinely does not contain it, that is usually knowable in week one and usually taken in week nine. Searching forty hours for a moment that was never shot is how a documentary edit loses its schedule
7. A found moment is by definition one the speaker did not intend to give, so where the contributor has a right of review there is a consent question in it that no card settles for you

### "The audience already knows how it ends"

This is an asset. Historical, heritage, true-story and adaptation work all run on it.

1. Structure for known outcome — the dread is in the arrival
   `info.known_ending_structure` · `tens.known_outcome_dread`
2. An adaptation has two audiences and must pay both
   `info.adaptation_double_audience`
3. A true-story frame makes a promise about accuracy; know which one you are making
   `info.true_story_frame`
4. In heritage work, part of the audience will correct you, and that is a craft constraint
   `info.heritage_audience_correction` · `expo.heritage_specificity`

### "A fact changed, or was always disputed"

1. Retconning has rules, and breaking them costs the audience's trust in everything
   `info.the_retcon`
2. A fact the story disproves later must have been honestly presented first
   `info.disproved_fact`
3. Where the history itself is contested, say so in the structure rather than picking silently
   `info.disputed_history` · `info.communal_version`
4. Information that travels through people degrades — that is usable
   `info.rumour_chain`
5. If what you owe is a published repair of a published error, that is its own form with its own disclosure problem, because a correction re-releases the wrong fact in order to kill it
   `info.the_correction`
6. And do not correct a disagreement. Conceding a contested interpretation to satisfy a complainant teaches their side that pressure works and teaches your audience that your positions are negotiable — the hybrid that concedes tone while insisting on substance reads as neither, and is the commonest form of the failure

### "The serial loses people between episodes"

1. Design what the audience must carry, and what you will re-supply
   `info.between_episode_memory`
2. Reorient through action instead of a recap
   `info.recap_without_recap` · `open.episode_opening`
3. In interactive work, a reveal may never be reached — plan for that
   `info.missable_reveal` · `info.multiple_paths_one_fact`

### "The reveal landed and nobody cared"

1. Nobody was waiting for it — the named failure
   `info.reveal_nobody_wanted` · `info.answer_to_unasked_question`
2. Or the mystery had no stakes attached to its answer
   `info.mystery_without_stakes`
3. Or you have said it three times already
   `info.thrice_told_fact` · `dial.known_information`

### "There's a fact I'm not allowed to state and the hole shows"

1. Then make the bar itself visible rather than letting the hole pass as silence. An NDA, a subject's approval, a court order, a source-protection promise, a legal review, an embargo — the piece can carry the *shape* of what it is barred from saying, and that is information
   `info.embargoed_fact`
2. The limit that decides whether you may: if declaring the bar identifies the thing it protects, do not declare it. An audience who can name the withheld fact from the shape of your refusal has been given the fact, and a source-protection promise announced too precisely is a source burned by arithmetic
3. And do not do it for an ordinary constraint. Every film has things it cannot say about everyone; announcing that is self-importance rather than transparency
4. If the withheld thing is a plot fact rather than a legal one, it is a different question entirely
   `info.never_disclosed`

### "I can't prove it and I'm not leaving it out"

1. Publish the claim with the *state of the checking* attached, so the audience receives the claim and its exact weight in the same breath
   `info.unverified_allegation`
2. Test your reason for keeping it first. "It is too important not to report" is the sentence that precedes most of the form's disasters and it is not a standard — a claim you cannot weigh cannot be handed over with the weighing attached, and publishing it anyway transfers your problem to a named person who did not agree to receive it
3. Never composite. Three unverified claims presented together read as corroboration and are not
4. The attaching itself is a design job rather than a compliance one: decide per claim what the audience is told about where it came from, in which channel, at what moment, and in how much detail
   `info.source_on_screen`
5. Two things that go wrong there. Attributing your own reporting to yourself on screen reads as insecurity — your reporting is asserted, and the film's standing is the source. And a citation supered over testimony converts a person into evidence, which is almost always the wrong choice at that moment
6. Where the audience simply does not believe you, that is the other complaint
   `info.credibility_spend` · `info.damaging_admission`

### "Half the film is reconstruction and the only place that's admitted is the credits"

1. The version of this with a name is `scene.reconstruction_unmarked` — a reconstruction shot in the visual grammar of archive, with grain, gate weave, desaturation, hands and backs of heads, and cut in without a label, so no viewer can tell what is evidence and what is invention. It works once and then destroys the piece: the moment one viewer identifies an unmarked reconstruction, every other image in the film becomes a question, including the real ones, and evidentiary trust cannot be re-earned inside the same film
2. Run it as a pass. Ask of every image whether a viewer could tell, from this shot alone, that it happened; where the answer is no, either label it or change how it is shot. The same applies to a still treated to imply motion it never had, which is the identical offence in a subtler form
3. Then move the admission to the moment it applies and write it inside the film's own grammar — spoken, supered, or carried in the narration — declaring that this shot or this sequence was re-staged, composited or arranged
   `expo.disclosed_staging_note`
4. Signal it in the picture and not only in a corner caption half the audience misses, because a signalled reconstruction is dramatically *better* than a disguised one: an audience that has been told what it is watching attends to what the scene shows instead of spending it wondering whether the thing is real. Clean modern frame, no grain, faces deliberately unresolved, the caption inside the first five seconds
5. Do not label what is obviously a device. A silhouetted actor, a hand pouring tea, a map animating — announcing those trains the audience to think you would otherwise have cheated
6. Nor does an essay film need to announce that it is not news; the register has already said so
7. The same fabrication committed in the cut rather than in the image is a separate named failure, and it is held in Scene shape
   `scene.false_adjacency`
8. If what the film is asserting is a person's *change* rather than a shot, that is the other declaration, and it has a harder standard
   `info.true_story_frame` · `arc.imposed_arc`

---