The First DraftFinds the Story.Revision Makes It Work.

Sorkin answers a direct question about second drafts by describing what completion reveals. A writer may begin on one course and arrive somewhere adjacent. At the end, the script can be about something different from what the writer expected when the opening pages were written.

His compact description—“I don't write scripts, I rewrite scripts”—is not merely a preference for polishing. He describes going back to the beginning, removing what no longer serves the discovered story, bringing necessary material into clearer relief, sharpening weak jokes and dialogue, and reducing excess. The first draft has supplied information that the opening could not yet contain.

That does not mean every writer must retype every page, or that every draft hides one correct subject waiting to be uncovered. It means Sorkin treats completion as a diagnostic event. The end provides a new position from which to judge the route that produced it.

Sorkin gives the most specific formulation in this cluster: a problem that appears in the third act may exist because the first act did not establish it properly. The repair therefore moves backward. Fixing the visible symptom at the end would leave its missing cause untouched.

BFI's description of Frank's lecture supplies a related but separate emphasis. It identifies the importance of opening scenes, the challenges of screenwriting, the “magic of rewrites,” and the importance of words, with examples from Get Shorty, Minority Report, The Lookout, and Out of Sight. The description does not provide a detailed Frank method, and it does not justify assigning Sorkin's first-act/third-act statement to him.

The connection is PasteLint's inference: openings are not protected merely because they were written first. Later pages can reveal that an opening prepared the wrong expectation, withheld a needed condition, or emphasized material the finished work no longer needs.

Later in the WIRED interview, Sorkin urges writers to move beyond saying that a scene was good or bad. After the immediate experience, the writer should ask what produced the effect. Was it how a joke was set up? Camera movement? The absence or presence of score? The question converts reaction into something specific enough to test.

Some of those causes extend beyond the written page. Performance, camera, score, pacing, dialogue, setup, and structural expectation can all influence the result. That makes the diagnostic habit broader than line editing, while also making it less mystical. Judgment is not simply having better taste. It is learning to locate the choice that created the response.

Revision becomes more deliberate when “this does not work” turns into “this moment asks for information the opening never supplied” or “this line repeats an emphasis the scene has already earned.” Diagnosis does not guarantee the right repair. It gives the writer a clearer object to revise.

Across the source cluster, rewriting can involve removal, setup repair, attention to openings, sharper language, rebalanced emphasis, and response to constraint. Those operations occur at different scales. A clumsy line may need a better line. A late structural weakness may require an earlier scene. Material can be well written and still belong to the story the writer thought they were making rather than the one they completed.

This is why “make it shorter” is an inadequate summary. Sorkin discusses excess, but reduction is one action inside a larger diagnostic process. Frank's publisher description keeps words and openings in view together. Solomon's account places change inside the longer journey from pre-production through post-production.

PBS describes Solomon tracing the genesis and evolution of Men in Black from pre-production through post-production. The program description includes being fired and rehired, working with cast and crew, and watching the story change across that process.

That evidence supports a bounded point: a screenplay can continue evolving when it meets collaborators, performance, production conditions, and professional disruption. Those encounters can expose assumptions that pages alone did not test. They can also impose compromises. Nothing in the source establishes that disruption automatically improves a script, or that collaboration always produces the best version.

Solomon's role in this cluster is therefore different from Sorkin's and Frank's. The pressure is not only the writer looking backward from the ending. It is the work meeting a world of people and constraints that did not exist inside the initial act of drafting.

A revision tool should help writers align a draft with what writing the draft taught them. It should not substitute its own interpretation for that discovery.

Second Draft operates inside a much narrower boundary than screenplay diagnosis. It can reveal exact eligible repetition, remove defined filler, preserve uncertainty and modality, preserve recommendation strength and who is acting, protect tested timing and value patterns, and make transformations reviewable. Those behaviors can reduce mechanical friction without pretending to understand the writer's story.

The distinction matters. A tool may show that one sentence appears twice. It cannot decide whether a repeated image is the work's governing motif. It may remove a bounded filler opening. It cannot know whether the first scene establishes the promise the ending requires. Cleaner prose is useful only while the writer remains responsible for what the work means and what the next draft must become.

The next layer of evidence should include comparative screenplay drafts, production drafts, script editors, directors discussing specific rewrite decisions, and documented before-and-after scenes. Research on expert revision could test whether retrospective alignment appears outside film writing and where the analogy breaks.

Counterexamples matter too: scripts whose first drafts changed very little, production constraints that weakened a story, or revisions that diagnosed the wrong cause. Those cases could narrow the synthesis and keep an appealing sequence from hardening into a rule it has not earned.

The most useful future cases would connect a stated diagnosis to an observable change: an opening rewritten after the ending clarified the central conflict, a scene removed because production exposed its redundancy, or a line retained because performance gave it a function the page did not reveal. That evidence would make the relationship between later knowledge and earlier repair more inspectable.