The Draft Is a Hypothesis

Dan Erickson budgets for failure before breakfast. The Severance creator has described the shape of his production days: the cast rehearses a scene in the morning, he hears the actors speak the lines for the first time, and the rewriting starts on the spot. An actor asks to try a different approach. A line that read clean on the page lands wrong in a human voice. Erickson expects all of it. He built his process around the assumption that the script arrives at set unproven.

Hold that posture up against how most writers treat their pages. We polish for months, defend every comma, and when actors strain against a clunky rhythm we explain that it plays better than it reads. Erickson runs the opposite protocol. The draft is a hypothesis. Rehearsal is the experiment. A line that fails the experiment was wrong, whatever it looked like in the document, and the failure is data rather than insult.

The reason this works sits in a gap most training never names: written conversation and spoken conversation obey different laws. On the page, a line of dialogue serves its function and moves on. In an actor's mouth, surrounded by breath and body and the weight of the take, the same line can turn to concrete. No amount of solitary polishing closes that gap, because the writer alone cannot run the experiment. The gap closes in the room, out loud, with other people. Erickson has also described the larger version of the problem: whole scenes that worked on paper and then, on the day, refused to become themselves. He treats that discovery as part of the schedule rather than a breakdown of it.

What the method demands is a specific separation: the writer's worth from the words' permanence. A writer who confuses the two cannot hear the room. Every suggestion registers as an attack, every stumble as sabotage, and the information the actors carry, which is real information available nowhere else, goes to waste. Erickson can rewrite on a shooting morning because the script functions as a working document, a thing under test, and his standing does not ride on any given sentence surviving.

Now speed the whole machine up, because that is where the craft is heading. Vertical drama shoots at a pace that makes network television look contemplative: small crews, compressed schedules, seasons of sixty to a hundred micro-episodes moving through production in weeks. There is no table-read week. There is no cushion between draft and camera. A writer in this format who needs distance from their own pages before they can revise them will drown. The Erickson posture stops being a luxury of prestige television and becomes the baseline requirement: arrive expecting the pages to bend, listen fast, rewrite standing up.

The good news is that most of the test can move upstream, to the desk, where it costs nothing. Read every line aloud before anyone else does; the lines that die in a mouth announce themselves in your own. Mark any line you cannot deliver in one breath. Hand the pages to a cold reader and watch where they stumble instead of asking what they think. Write dialogue that survives a phone speaker, because in the vertical format that is the instrument it plays through. None of this replaces the rehearsal discovery, but it raises the batting average of what reaches the day.

There is a structural version of the same humility, and serialized formats reward it. A season built as a chain of cliffhangers is itself a hypothesis about what an audience will chase. Writers who hold the whole season rigid, every beat locked before episode one shoots, lose the chance to learn from the audience the format hands them episode by episode. The ones who treat the season outline the way Erickson treats a scene, as a strong guess awaiting evidence, course-correct while the season still breathes.

Severance reads on screen as a show of obsessive control: the corridors, the cadence, the precision. The lesson of its process is that the control came through surrender, morning after morning, to what the rehearsal revealed. Precision was the output, never the input.

Hold your next draft the same way. It is a guess. Make it a good guess, then go find out.

Screenburn keeps version history on every story and every scene for this exact reason: drafts are experiments, and experiments deserve lab notes.