Docs/Skills and code/skill video-script-production

.claude/skills/video-script-production/references/implicit-pull.md

Markdown56 lines6.1 KBSkill: video-script-production

What the file says about itself

Implicit pull: keeping a viewer waiting between the questions: Added after the Moser Baer v4 review: "the open loop that keeps the viewer watching through might not be satisfied for the full script. It doesn't have to be explicit;

Mentioned on: page 12

Implicit pull: keeping a viewer waiting between the questions

Added after the Moser Baer v4 review: "the open loop that keeps the viewer watching through might not be satisfied for the full script. It doesn't have to be explicit; it can be subtle, implicit and implied."

The problem

Explicit loops (a question asked, then answered) go stale. v4 declared a spine loop and a beat loop at the end of each beat, and the linter passed, but a beat loop opened in a beat's last two sentences is only alive until the next beat answers it. Inside a beat, in the middle of a paragraph of dates, nothing was pending. The viewer had no reason to keep listening right now, which is when people leave.

What a pull is

A pull is something the viewer is waiting for without anyone having asked a question. It is made of information the viewer already holds plus a tension the script keeps in view. It stays live from its signal to its resolution, and it only works if the script touches it (a word, a date, an image, a callback) often enough that the viewer still has it in mind. The linter treats a pull as warm for 60 seconds after each touch.

TypeThe viewer is waiting forExample from Moser Baer
clocka deadline that is comingThe bond falls due on 21 June 2012; every later beat can say how far off it is
ironythe moment the people in the story find out what the viewer already knowsWe know the date the bond falls due; the company, in 2007, is selling it as ordinary financing
jeopardywhat happens to a person or group the viewer has metWorkers who built the plant: introduced early, and the viewer wants to know where they end up
predictionthe verdict on something the script said should be true"If that's right, sales will barely move and profit will vanish. Check March 2008."
motifwhat a recurring image or claim meansThe shut gate; the founder's habit of leaving whatever was running out
gapa fact the script says it does not have"I have not found where the bond's money went"
patternwhether a repeated shape repeats or breaksPrice halves in discs, drops to a chocolate's price in films, falls by two-thirds in solar: the viewer starts to expect the next price to fall

Why it should work, and how sure we are

SupportGradeWhat it says
Loewenstein's information-gap theory (above, Lesson B)BThe viewer must know enough to notice the gap. A pull is a gap built from facts already given, so it needs no new question.
Brewer and Lichtenstein (1982), structural-affect theoryBSuspense comes from an initiating event with significant consequences for someone the audience cares about, plus uncertainty about the outcome. That is jeopardy, and with a known date it is clock.
Gerrig and Bernardo (1994), readers as problem-solversA (experimental, fiction only)In seven experiments suspense rose when readers believed the number of ways out had been restricted, even when their estimate of escape did not change. A deadline and a shrinking set of options are the same lever.
Hitchcock to Truffaut on the bomb under the table (dramatic irony)CCraft doctrine: tell the audience what the characters do not know. Experiments on whether knowing the outcome reduces suspense disagree (a 2018 Frontiers in Psychology study found outcome knowledge changed emotional response separately from suspense). Use irony for the known date or known result, not as proof.
E. M. Forster ("the audience wants to know what happens next")CThe simplest pull is a person the audience wants to follow.
Zeigarnik effectDReplicates inconsistently; do not cite it for any of this.
Ours: loop-timing raters agree poorly (0.08 to 0.39)OursSo pulls are declared with quotes and checked mechanically; whether they work is for a fresh reader.

No lesson here has been tested on audience retention. The rule is a design constraint: no stretch of the script should depend on the viewer's goodwill alone.

How to place them (Step 3 of SKILL.md)

  1. Pull inventory. After the chain is drafted, walk the beats and for every minute of runtime write one line: what is the viewer waiting for right now? If the answer is "the next beat", the minute is uncovered.
  2. Choose at least four pulls of at least three kinds. One of them should be a person (jeopardy) so there is a human in the waiting, not only a mechanism. One should be a motif that comes back as the closing image.
  3. Plant early, touch often, resolve late. Signal the pull in the first third where you can. Touch it at least twice more, no more than about a minute apart, in a way that sounds like narration: a repeated word, a date, a callback to the scene. Resolve it in the beat where the facts resolve it, or admit the gap aloud and mark it left_open.
  4. Touches are not announcements. "Remember this date" is script-talk (F2). "Five years before it falls due" is a touch.
  5. Keep them true. A clock needs a real date, jeopardy needs real people who are in the sources, irony must rest on a fact the viewer is told. Do not build suspense on hindsight smugness ("how did they not see it?"), and do not invent a person.
  6. Fewer than you think are explicit. The spine and at most two beat loops are asked aloud; everything else rides on pulls. If every pull is spoken as a question, the voice becomes the staged run-up that F2 counts.

What the E rules check

RuleChecks
E1every pull has a valid type, a signal quote in its beat, at least two touches in the text, in order, and either a resolution or a line that admits the gap
E2after the hook, no stretch longer than 45 s (warn) or 75 s (fail) without an explicit loop or a pull touched in the last 60 s
E3at least three kinds of pull used
E4every middle beat carries a pull signal, a touch or a beat loop

The checks confirm the declared pulls are in the text where the chain says. They cannot confirm that a touch feels like tension. A fresh reader can: ask "at which paragraph did you stop caring what happens?"

Show the raw text
# Implicit pull: keeping a viewer waiting between the questions

Added after the Moser Baer v4 review: "the open loop that keeps the viewer watching through might not be satisfied for the full script. It doesn't have to be explicit; it can be subtle, implicit and implied."

## The problem

Explicit loops (a question asked, then answered) go stale. v4 declared a spine loop and a beat loop at the end of each beat, and the linter passed, but a beat loop opened in a beat's last two sentences is only alive until the next beat answers it. Inside a beat, in the middle of a paragraph of dates, nothing was pending. The viewer had no reason to keep listening *right now*, which is when people leave.

## What a pull is

A **pull** is something the viewer is waiting for without anyone having asked a question. It is made of information the viewer already holds plus a tension the script keeps in view. It stays live from its signal to its resolution, and it only works if the script **touches** it (a word, a date, an image, a callback) often enough that the viewer still has it in mind. The linter treats a pull as warm for 60 seconds after each touch.

| Type | The viewer is waiting for | Example from Moser Baer |
|---|---|---|
| `clock` | a deadline that is coming | The bond falls due on 21 June 2012; every later beat can say how far off it is |
| `irony` | the moment the people in the story find out what the viewer already knows | We know the date the bond falls due; the company, in 2007, is selling it as ordinary financing |
| `jeopardy` | what happens to a person or group the viewer has met | Workers who built the plant: introduced early, and the viewer wants to know where they end up |
| `prediction` | the verdict on something the script said should be true | "If that's right, sales will barely move and profit will vanish. Check March 2008." |
| `motif` | what a recurring image or claim means | The shut gate; the founder's habit of leaving whatever was running out |
| `gap` | a fact the script says it does not have | "I have not found where the bond's money went" |
| `pattern` | whether a repeated shape repeats or breaks | Price halves in discs, drops to a chocolate's price in films, falls by two-thirds in solar: the viewer starts to expect the next price to fall |

## Why it should work, and how sure we are

| Support | Grade | What it says |
|---|---|---|
| Loewenstein's information-gap theory (above, Lesson B) | B | The viewer must know enough to notice the gap. A pull is a gap built from facts already given, so it needs no new question. |
| Brewer and Lichtenstein (1982), structural-affect theory | B | Suspense comes from an initiating event with significant consequences for someone the audience cares about, plus uncertainty about the outcome. That is `jeopardy`, and with a known date it is `clock`. |
| Gerrig and Bernardo (1994), readers as problem-solvers | A (experimental, fiction only) | In seven experiments suspense rose when readers believed the number of ways out had been restricted, even when their estimate of escape did not change. A deadline and a shrinking set of options are the same lever. |
| Hitchcock to Truffaut on the bomb under the table (dramatic irony) | C | Craft doctrine: tell the audience what the characters do not know. Experiments on whether knowing the outcome reduces suspense disagree (a 2018 *Frontiers in Psychology* study found outcome knowledge changed emotional response separately from suspense). Use `irony` for the known date or known result, not as proof. |
| E. M. Forster ("the audience wants to know what happens next") | C | The simplest pull is a person the audience wants to follow. |
| Zeigarnik effect | **D** | Replicates inconsistently; do not cite it for any of this. |
| Ours: loop-timing raters agree poorly (0.08 to 0.39) | Ours | So pulls are declared with quotes and checked mechanically; whether they work is for a fresh reader. |

No lesson here has been tested on audience retention. The rule is a design constraint: no stretch of the script should depend on the viewer's goodwill alone.

## How to place them (Step 3 of SKILL.md)

1. **Pull inventory.** After the chain is drafted, walk the beats and for every minute of runtime write one line: *what is the viewer waiting for right now?* If the answer is "the next beat", the minute is uncovered.
2. **Choose at least four pulls of at least three kinds.** One of them should be a person (`jeopardy`) so there is a human in the waiting, not only a mechanism. One should be a `motif` that comes back as the closing image.
3. **Plant early, touch often, resolve late.** Signal the pull in the first third where you can. Touch it at least twice more, no more than about a minute apart, in a way that sounds like narration: a repeated word, a date, a callback to the scene. Resolve it in the beat where the facts resolve it, or admit the gap aloud and mark it `left_open`.
4. **Touches are not announcements.** "Remember this date" is script-talk (F2). "Five years before it falls due" is a touch.
5. **Keep them true.** A clock needs a real date, jeopardy needs real people who are in the sources, irony must rest on a fact the viewer is told. Do not build suspense on hindsight smugness ("how did they not see it?"), and do not invent a person.
6. **Fewer than you think are explicit.** The spine and at most two beat loops are asked aloud; everything else rides on pulls. If every pull is spoken as a question, the voice becomes the staged run-up that F2 counts.

## What the E rules check

| Rule | Checks |
|---|---|
| E1 | every pull has a valid type, a signal quote in its beat, at least two touches in the text, in order, and either a resolution or a line that admits the gap |
| E2 | after the hook, no stretch longer than 45 s (warn) or 75 s (fail) without an explicit loop or a pull touched in the last 60 s |
| E3 | at least three kinds of pull used |
| E4 | every middle beat carries a pull signal, a touch or a beat loop |

The checks confirm the declared pulls are in the text where the chain says. They cannot confirm that a touch *feels* like tension. A fresh reader can: ask "at which paragraph did you stop caring what happens?"

Shown as it is in the repository at build time (2026-11-03); the home directory appears as <home>. If this page and the file differ, the file is right. Back to the catalogue.