Structure

There is a particular kind of stuck that only happens on page sixty. The setup worked. The characters were interesting for a while. And now the script is a series of scenes in which people who used to be interesting have conversations about what might happen next. If that is where you are, the problem is almost never that you failed to hit a beat on the right page. It is that the machine underneath the beats was never running.

FADE UP:

INT. THE PROBLEM WITH TEMPLATES - DAY

Syd Field did not invent three-act structure. He watched a great many successful films, noticed that they tended to turn at roughly the same proportions, and drew a diagram of it. Blake Snyder did the same thing thirty years later with more granularity and better jokes. Christopher Vogler did it by running Joseph Campbell over a studio slate. All three books are useful. All three are descriptions, and every one of them has been read by a hundred thousand people as a prescription.

The difference matters enormously. A description tells you what a healthy script tends to look like from the outside. A prescription implies that producing those external features will produce a healthy script. It will not, for the same reason that painting spots on a horse does not get you a leopard. You can hit the Catalyst on page ten, the Midpoint on page fifty-five and All Is Lost on page seventy-five, and still have written nothing, because the beats are the symptoms of a story engine running properly, not the engine.

(This is not an argument against reading them. Read all three. Just read them the way a doctor reads a list of symptoms, not the way a cook reads a recipe.)

So before any of the shapes, the engine. It is much simpler than the diagrams, and much harder to do.

CUT TO:

INT. THE ENGINE - CONTINUOUS

David Mamet, teaching directors, reduced the whole apparatus to three questions he insisted be answerable about every scene and every film:

That is the engine. A person wants something specific, something is in the way, and the cost of failure is real and immediate. Everything else in this article is a consequence of those three questions being answered well or badly.

Notice what the questions demand. Specific. “Wants to be happy” is not a want; “wants to get the money out of the hotel before six” is. Opposed. If nobody is in the way, you have a to-do list. Urgent. “Why now” is the question that kills more dead second acts than any other — if the story could just as easily have happened a year later, you have not started it yet.

Lajos Egri, writing in 1946 and still the sharpest book on the subject, put the same thing differently: a play begins where the conflict begins, and conflict comes from a character who wants something so badly they will not stop. Egri called the controlling idea the premise, and insisted it be a sentence with cause and effect in it — “great love defies even death” — rather than a topic. A premise you can state is a script you can cut, because you know what is off the spine.

If you cannot say who wants what, and what it costs them to fail, you do not yet have a structural problem. You have a story problem wearing a structural problem’s coat.

CUT TO:

INT. THE UNIT OF CONSTRUCTION - LATER

Structure is not made of acts. It is made of scenes, and most writers are carrying around a definition of “scene” that is doing them damage: the idea that a scene is a location plus some dialogue.

Robert McKee’s definition is the useful one. A scene is an action through conflict that turns the value-charge of a character’s situation. Something is different at the end of it — safe becomes unsafe, together becomes apart, hopeful becomes desperate. If nothing turns, McKee’s term for what you have written is a non-event, and his advice is to cut it and redistribute anything it contained.

The test, applied honestly

Take any scene in your draft and answer two questions. What is the charge at the top — is the character up or down, winning or losing, close or far? What is it at the bottom? If the answer is the same, that scene is delivering information, and information is not drama. You can usually fix it three ways: give somebody in the room a want they do not currently have, give them an opponent, or cut the scene and let the audience infer it.

Dwight Swain, writing about fiction in 1965 but describing something universal, split the same material into scene (goal, conflict, disaster) and sequel (reaction, dilemma, decision). The insight worth stealing is that the quiet bits are not filler — they are where a character metabolises a defeat and chooses the next goal, which is what makes the next scene mean anything.

CUT TO:

INT. THE SCALE NOBODY TEACHES YOU - DAY

Here is the most useful structural idea in this article, and it is the one least likely to have reached you, because it was taught rather than published for most of its life.

Frank Daniel — who ran the film programmes at Columbia and USC and co-founded the Sundance Institute’s labs — taught that a feature is best understood as roughly eight sequences of ten to fifteen pages. The idea comes out of the silent era, when a projector reel physically held about that much film and each reel had to work as a unit. Each sequence has its own protagonist want, its own escalating obstacle, and its own turn at the end that reframes the next one. Paul Gulino wrote it down properly in 2004.

Why this beats three acts as a working tool:

• It is the scale you actually write at. “Write act two” is an instruction to write sixty pages, which is not an instruction. “Write a twelve-page sequence in which she tries to get the money and fails worse” is a morning’s work.
• It localises the sag. A broken second act is almost always two or three sequences that have no want of their own, sitting between two that do. At sequence scale you can see exactly which ones.
• It gives the middle its own spine. Each sequence asks a dramatic question and answers it — usually “no, and worse” — which is what keeps a long middle moving.

(Try it on a film you know cold. Name its eight sequences in one line each. You will find them, and you will never look at your own outline the same way.)

CUT TO:

INT. WHAT THE ACT BREAKS ACTUALLY ARE - CONTINUOUS

Acts are not containers you pour pages into. An act break is a point of irreversible change in the protagonist’s goal. That is the whole definition, and it explains why the page numbers are approximate: they fall where they fall because that is how long it takes to earn them.

• The first turn (Field’s Plot Point I, around a quarter of the way in) is where the protagonist commits to a course of action they cannot walk back. Before it they could still go home. After it they cannot. The debate that precedes it is not padding — refusing the call is what proves the cost is real.
• The midpoint is the most underused structural tool in circulation. It is where the protagonist stops reacting and starts acting — often via a false victory or a false defeat that reveals the real stakes. If your middle sags, this is nearly always the missing piece: the protagonist is still being pushed around at page sixty.
• The second turn (around three-quarters) strips away the last resource — the plan, the ally, the illusion — so the climax has to be won with whatever the protagonist has actually become.

CUT TO:

EXT. THE CAUSALITY TEST - DAY

This is the single most valuable structural diagnostic in existence, it takes twenty minutes, and it came from the creators of South Park.

Write your story out as a list of beats. Now go between every pair and try to insert the words “therefore” or “but.” If the only word that fits is “and then,” you have found a structural fault.

“And then” means the next event does not depend on the previous one. Events that merely follow each other are chronology; events that cause each other are plot. Aristotle said essentially this in the Poetics — that the incidents must follow by necessity or probability, not merely one after another — and it has not been improved upon in twenty-three centuries.

Therefore / but is plot. And then is a list. Run the whole outline and mark every “and then” in red — that map is your rewrite.

CUT TO:

INT. WRITE TOWARD THE ENDING - NIGHT

Michael Arndt — who wrote Little Miss Sunshine and Toy Story 3 — argues that the ending is not the last thing you figure out but the first, because the ending is what the film is about. Everything before it exists to make one final choice cost something. In his framing, a great ending fires three things at once: the external plot resolves, the internal need resolves, and the philosophical question the film has been asking gets answered by a character’s action.

Practically: know your final image before you write page one. Not the plot mechanics — the image. It will change, and it will still have saved you twenty drafts, because every scene now has somewhere to be going.

FADE OUT.