Roth wrote a monologue for The Insider. A page and a half. He thought it was pretty good. The morning of the shoot, Al Pacino called him: I can do this with one look.1 Roth was surprised. He called the director Michael Mann. Mann said: let's shoot it both ways. In the final cut, it's the one look.2
That's dialogue economy at its purest. A page and a half of words become a single moment of held gaze, and the gaze carries the entire emotional content the words were carrying. Pacino is doing the actor's version of subtext-via-dream-rule. The information arrives without language. The audience receives more because the actor was trusted with the moment.
Roth's reflection on this is generous: sometimes I'm trying to find ways to describe things that maybe I go overboard or something, you know, or maybe an actor can then take it make it their own.3 He's naming a specific writerly failure mode — over-writing scenes that should be played in silence — and a specific collaborative grace — recognizing when the actor knows better.
The inverse of Pacino's economy is shoe leather. Roth explains the term: Why is it called that? Because it's just wasting time walking around, you know.4 Shoe leather is dialogue that fills time without doing work. Characters chat to chat. Scenes extend because they exist, not because they're earning their length. Cut shoe leather, and the script tightens. Don't, and the audience starts checking watches.
Dialogue economy is the discipline of saying less than you initially wrote. The page-and-a-half draft is the writer's first pass — necessary work, because the writer needs to know what the scene is doing before they can know how short it can be. The cut version is the second pass — the work of recognizing that most of the words were the writer's exploration, not the scene's requirement.
Pacino's one-look is the asymptote: the moment when the right actor with the right director needs zero words. Most scenes don't get there. Most scenes can get closer. The discipline is to keep trimming until the scene is as short as it can be without losing the content — and then check whether one more cut still preserves the content.
Shoe leather is the negative space: dialogue that exists for transition, for filler, for we need to get to the next scene somehow. It usually announces itself as small talk, scene-bridging logistics, or characters narrating their own movements. Audiences tolerate small amounts; large amounts dissolve attention.
Dialogue economy becomes urgent in three situations:
When a scene's emotional content can be carried by an actor's face. The Pacino case. Any moment of decision, realization, regret, or held emotion — these are actor-territory, not writer-territory. Words explain; faces embody.
When a scene's information is already known to one character. The Water Commissioner trap variant. If one character speaks information the other already has, you've written shoe leather even if the dialogue is technically necessary for the audience.
When scenes are running long without escalating. Three pages of conversation that establish what one paragraph could establish. The writer is enjoying the dialogue. The audience won't.
Three mechanisms operate together:
The trust of the actor. Roth's craft includes leaving space for the actor's contribution. A line that could be cut for a look is a line the writer should consider cutting before the page is sent. Trusting the actor is also trusting the director and the editor — recognizing that the script is the first draft of a collaborative artifact, not the final form.
The audience's pattern-completion engine. Audiences fill in dialogue that wasn't spoken when the structural setup primes them. Two characters arrive at a moment everyone has been building toward; the unspoken weight of the moment is itself communication. The writer who insists on speaking the moment defuses it.
The scene's earning-its-length test. Every line should answer: does this line do something the previous line didn't do? If the answer is no, the line is shoe leather. Repeat: every line. The discipline is brutal. The result is dialogue that breathes because every breath is doing work.
Dialogue economy makes the dream-rule subtext discipline workable. Subtext requires that the unspoken content carries; economy ensures the unspoken content has room to carry by removing the spoken filler that would smother it. The Pacino look only works because Roth cut the page and a half. If the page and a half had stayed, the look would have been redundant.
Economy also makes the theme-as-rudder principle operable at scene level. Theme survives in scenes that have been compressed to their load-bearing lines. Theme suffocates in scenes that have been padded with shoe leather. Compress the scene; theme survives.
The Forrest Gump bench scenes are a study in dialogue economy operating inside apparent looseness. Forrest sits on a bus stop bench. Various people sit next to him. He talks. They listen. The form looks chatty — long monologues, episodic structure, Forrest's voice meandering.
Read it carefully: every line is earning its length. The shrimp speech (shrimp and potatoes, shrimp burger, shrimp sandwich5) does multiple jobs. It establishes Bubba's voice and character. It establishes Forrest's memory and his attention to people who matter to him. It earns the audience's emotional investment in Bubba's later death. It's also funny, which buys Roth space for the next emotional beat. Five paragraphs of shrimp; not a wasted line.
The Jenny moments are even tighter. I'm not a smart man, but I know what love is.6 One line. Carries the entire Forrest-Jenny relationship. Anything more would weaken it. Roth's discipline at the line level shows: the line was crafted to be small enough to land hard.
The shoe-leather counter-example: the original life is like a box of chocolates line in the source novel was being an idiot is no box of chocolates. Roth changed it. The replacement is tighter, more memorable, more line-worthy. The change is small. The economy is real.
You've written a scene. It's three pages long. The information is delivered. The characters are in the right emotional places. You're proud of some of the lines.
Make a copy of the scene. In the copy, cut every line that doesn't do something the previous line didn't do. Be ruthless. Cut a line you love if the line above it is doing similar work. Cut the line of small talk that bridges into the real content. Cut the closing line that explains what the audience just saw. The copy is now probably a page and a half. Maybe a page.
Now read the cut version aloud. If it lands harder than the three-page version, the cuts were right. Most of the time they will be. Save the cut version. Discard the original.
If you've cut a line you really loved, save it in a separate document of good lines that didn't fit. Eventually some of them will find a scene where they're load-bearing. Most won't. That's fine. Writing is overwhelmingly the work of producing more than you need so you can recognize what you actually need.
Pacino's one-look discipline is the next level: you cut not just lines but moments. The scene that needed dialogue at the page-draft stage may need silence at the production stage. Trust the director. Trust the actor. Trust the editor. Your draft is the first move in a longer game.
You'll know economy has failed when audiences describe scenes as talky. Talky is the lay term for dialogue-without-economy. The scene is doing its work but doing it more times than necessary. Audiences feel the inefficiency without naming it.
You'll know shoe leather is operating when you can describe scenes by saying they walk to the car and talk about their day and then they get in. If the description doesn't include emotional or informational stakes, the scene is shoe leather. The fix is usually to cut the scene entirely and let the next scene start from the destination.
The deepest failure is the fear of cutting — the writer who has internalized that their dialogue is the work and cannot bring themselves to remove lines. This is the harder discipline. Cutting lines that are technically good is not a betrayal of craft; it's the expression of craft. The Pacino call is the writer's master class: when the actor can do it in a look, the page-and-a-half was the writer's exploration, not the script's content.
Dialogue economy can over-shoot. Some films have so compressed their dialogue that the audience loses orientation. The discipline isn't cut everything; it's cut everything that doesn't earn its length. Some scenes do earn longer dialogue — sustained argument scenes, monologues that build cumulative force, comedic riffs that depend on rhythm. The economy discipline still applies but the threshold is different.
A second tension: actor-trust is genre-specific. Roth works with elite directors and actors who can carry the Pacino-look moments. Writers working with less experienced collaborators may need to write the page-and-a-half because the actor cannot deliver the look. Economy is a collaborative-trust calculation, not an absolute principle.
A third tension is unresolved: how do you teach economy to writers who haven't internalized it? Roth doesn't address this directly. The implicit answer is: have them write the page-and-a-half, then have them cut it themselves until it's the look. The writing of the over-long version is part of how the cutting becomes possible. Skipping straight to the cut version produces thin work; cutting from the long version produces dense work.
Set Roth's economy against Karlsson's body-state writing7. Karlsson's prose is often dense and accumulative — sentences that pile clauses on clauses, paragraph rhythms that build by addition rather than compression. Both writers are doing work the other isn't. Karlsson is transmitting somatic state through prose-rhythm, which requires duration. Roth is transmitting emotional moment through compression, which requires absence.
The convergence: both writers refuse the information-transfer-as-prose failure mode. They diverge on whether the transmission medium is duration or compression. Reading both reveals that economy is medium-specific. Screenwriting's medium privileges compression because the actor and the visual carry duration. Essay prose's medium privileges accumulation because the words must carry everything alone. The dialogue economy discipline is a screenwriter's specialty; it transfers to other forms with care.
Whyte's poems sit between them: compressed at the line level but accumulative across the poem. Whyte's lines have Roth-level economy. Whyte's structures have Karlsson-level duration. The polymath read: economy is scale-dependent, and writers in any form should think about which scale their craft is operating at.
The compound insight: dialogue economy is operationally validated silence-deployment (behavioral mechanics) + philosophically aligned with how meaning arises (Eastern frame) + craft-specific compression discipline (screenwriting). The writer who knows only the craft layer treats it as technique. The polymath understands that compression is the writer's contribution to how the audience's own meaning-making apparatus works. Trusting that apparatus is the craft.
The Sharpest Implication
If you've been told your dialogue is good but your scripts feel long, the diagnosis is dialogue economy. The fix is not better dialogue. The fix is less dialogue. Specifically: less of the dialogue you wrote first, more of the silence between the lines you decided to keep. The Pacino one-look is unreachable for most scripts, but every script can get closer than its current draft.
Generative Questions