The script library lives in a converted broom closet at the back of the Parish Hall. It smells of old paper, dust, and the particular sweetness of binding glue that has outlasted three mayors. The shelves are plywood, cut by someone who measured twice and still got the left corner wrong. On them sit 147 plays in various states of annotation—dog-eared, sticky-noted, coffee-stained, and, in one case, held together by a single rubber band that everyone knows not to touch.
Marta Okonkwo has been the keeper of this library for fifteen years. She is not a director. She has never auditioned. She does not take bows at the end of the spring showcase. What she does is arrive forty minutes early to every rehearsal with a plastic bin of scripts, a three-hole punch, and a mental catalog of which actor needs which version of which scene because the director cut the monologue in Act Two last Thursday and nobody told the new understudy.
When the filmmaker first emailed the theater group offering to “document your process,” Marta was the one who read the message aloud at the monthly business meeting. She read it flat, without inflection, the way she reads stage directions when someone is sick. The filmmaker—let’s call him Daniel—had credentials. He’d done a short about a mural collective in Detroit. He had a grant from a regional arts foundation. He wanted to capture “the raw, authentic community-building that happens in spaces like this.”
The room got quiet in the way rooms get quiet when someone has said something that sounds generous but lands wrong.
The Warm-Ups Are Public. The Crying Is Not.
Community theater, the kind that rehearses in church basements and borrows folding chairs from the senior center, runs on a different clock than professional houses. Scenes are built slowly, over weeks of Tuesday nights and Saturday afternoons. People show up after double shifts. They forget lines and remember them again. They bring casseroles when someone’s mother dies. The work is not just the play. The work is the showing up.
Daniel’s proposal was specific: he would film rehearsals for six weeks, interview cast and crew, and produce a twenty-minute documentary the group could use for grant applications and social media. He offered to bring his own equipment. He offered to train someone on the board to use an AI transcription tool so they could generate searchable records of every rehearsal conversation. He mentioned, almost as an aside, that he’d been experimenting with an AI script generator that could help structure narrative arcs from raw footage transcripts, which he thought might be useful for shaping the final piece.
Marta wrote all of this down in the meeting minutes. Next to the AI script generator mention, she put a small question mark in the margin. She has been putting question marks in margins since 2009.
The debate that followed took three meetings to resolve, and it was never really about Daniel. It was about what the group owed itself, what it owed the people who walked through the door on Thursday nights carrying personal disasters they did not name, and whether the tools used to capture the work would change the work itself.
What the Script Library Knows
To understand the debate, you have to understand what Marta’s library actually holds. It is not just scripts. It is the sedimentary record of fifteen years of community decision-making.
There is the 2011 production of The Glass Menagerie where the actor playing Tom had a panic attack during dress rehearsal and the director rewrote the blocking so he never had to face the audience directly. That blocking is penciled into the margins. There is the 2016 original work about the factory closing, written collectively over six months of potluck dinners, where the third act was never finished because the playwright’s husband got sick and the group decided, without a vote, to perform it as a two-act with a discussion afterward. The unfinished third act is still in the folder, held together by that rubber band.
There is the 2019 Our Town where the woman playing Mrs. Gibbs had just lost her own son and asked, during the second week of rehearsal, if she could sit out the cemetery scene. The director said yes. The understudy stepped in. No one filmed it. No one tweeted about it. It is recorded nowhere except in Marta’s note on the cast list: “E. requested omission of Act III. Approved. Understudy J. covered. E. attended all remaining rehearsals and brought lemon bars for strike.”
This is the archive that Daniel’s camera wanted to enter. This is the archive that his AI transcription tool would have rendered searchable, taggable, grant-ready. And this is the archive that several cast members, when the discussion opened, said they did not want to see turned into content.
The First Meeting: Who Controls the Story
The first meeting about the documentary was held on a Wednesday in June. The air conditioning in the Parish Hall had broken again, so they propped the fire door open with a brick and let the moths in. Seventeen people sat in a circle of mismatched chairs. Daniel was not present; the group had decided, after some back-and-forth, to discuss the offer internally before inviting him to respond.
Luis, who has directed four of the last six spring shows, spoke first. He said he was tired of writing grant reports that no one read. He said a documentary might finally show funders what the group actually did. “They think we’re just putting on plays,” he said. “They don’t see the rest of it.”
Cheryl, who runs the youth summer program, disagreed. She said the rest of it was precisely what shouldn’t be filmed. “The kids tell us things in here they don’t tell their parents,” she said. “You put a camera in the corner, they stop talking. Or they start performing. Either way, it’s not the same room.”
Marta, as always, took notes. She wrote: “Luis: documentary as proof of existence. Cheryl: documentary as interruption of existence.” She did not write her own opinion. She has a policy about that.
The conversation circled for forty minutes. Someone raised the question of who would own the footage. Daniel had said the group would have final cut approval, but no one in the room had ever negotiated a licensing agreement. Someone else asked whether the AI transcription tool would store data on a third-party server. “I don’t want our conversations about casting disagreements ending up in some cloud we can’t delete,” said Pat, the treasurer, who works in IT and has opinions about data sovereignty.
By the end of the first meeting, they had not decided anything. They had only generated a list of questions, which Marta typed up and emailed to Daniel with the subject line: “Clarifications requested before further discussion.”
The Second Meeting: What Gets Recorded
Daniel’s response arrived within a week. It was professional, accommodating, and slightly missing the point. He offered to let the group review all raw footage before editing. He offered to use a local, offline transcription tool instead of a cloud-based one. He offered to blur faces of anyone who didn’t want to appear on camera. He did not offer to leave the room during the scenes where people cry.
The second meeting was smaller—eleven people, because it was August and several regulars were on vacation. This is a thing that happens in community groups: the most consequential decisions sometimes get made when the room is half-empty, and everyone has to live with that.
This time, the conversation turned to the specific tools Daniel had mentioned. The AI transcription, the script generator, the multi-camera setup. Pat had done some research. She explained that most AI transcription services are built on large language models trained on vast datasets of unknown provenance. She referenced, without naming it directly, the kind of guidance that professional writers’ organizations have begun issuing—the principle that AI outputs are generic mashups of pre-existing works, contrasting with the unique human voice and thinking that make a writer who they are. “If we let a machine transcribe our rehearsals and then feed that into a script generator,” Pat said, “whose voice is the output? Ours? Or the average of every play that thing has ever ingested?”
This landed. Several people in the room had spent years developing their craft—not professionally, but with the kind of devotion that comes from needing a place to put the things you cannot say at work. The idea that their halting, imperfect, deeply specific rehearsal dialogue might be processed into something generically “dramatic” felt like a violation of a different order than simply being filmed.
Marta, who rarely speaks during debates, raised her hand. “The scripts in the library,” she said, “are valuable because they’re marked up by specific hands. You can see where someone crossed out a line and wrote ‘no’ in the margin. You can see where the director changed ‘angrily’ to ‘quietly’ and dated it. That’s the record. If we replace that with a searchable transcript that nobody touched, we lose the thing we’re trying to preserve.”
The room nodded. Not unanimously—Luis still wanted the documentary—but with the kind of gravity that signals a shift.
What Screenplay Format Has to Do With It
There is a deeper tension here, one that the group only half-articulated. It has to do with what happens when community work gets formatted for outside consumption.
Professional screenwriting, as any guide will tell you, follows strict conventions. Screenplay format rules—font, margins, scene headings—impose a structure that could conflict with the fluid, emotional nature of rehearsal work. A script is a blueprint for production. It assumes a finished product. It assumes that the mess has been edited out.
But community theater, the kind Marta’s library documents, is mostly mess. It is the unfinished third act. It is the blocking change penciled in because someone had a panic attack. It is the lemon bars at strike. To format that into a documentary—to shape it into a narrative arc with rising action and resolution—is to impose a structure that the work itself never had. It is to pretend that the group’s story is a story, rather than an ongoing argument, a series of accommodations, a pile of sticky notes held together by a rubber band.
This is not a Luddite argument against cameras or AI. It is an argument about fidelity. The group was not opposed to documentation. Marta had been documenting them for fifteen years. What they were opposed to was documentation that transformed the thing it claimed to preserve.
The Third Meeting: The Decision
The third meeting was in September. The air conditioning had been fixed, but now the heat wasn’t working, so everyone kept their coats on. Daniel was invited this time. He sat in the circle, listened to the group’s concerns, and did something that surprised several people: he didn’t argue.
He said he understood. He said he’d been thinking about the difference between a document and a record. He said he’d been re-reading his own footage from the mural project and noticing how much of the real work—the arguments about color, the apologies, the meals—had happened when the camera was off. “I think I’ve been making films about the parts of community work that look good on film,” he said, “and missing the parts that actually hold it together.”
This admission changed the temperature of the room. It didn’t resolve the debate, but it made the debate feel less like a standoff and more like a negotiation between people who were all, in their own ways, trying to protect something.
What emerged was a compromise that no one had predicted. The group would allow Daniel to film the warm-ups. The physical and vocal exercises at the start of every rehearsal. The circle stretches, the tongue twisters, the game where someone says a word and the next person has to say the first word that comes to mind. These were, everyone agreed, public-facing in a way that the scene work was not. They were the part of the process that the group already shared with newcomers, with donors, with the occasional local reporter who stopped by for a feature.
But the scenes themselves—the ones where actors broke down, where directors changed their minds, where someone tried a line three different ways and landed on the version that made the room go silent—those would remain off-camera. No AI transcription. No script generator. No searchable archive. Just Marta’s notes, handwritten in the margins, with her small question marks where the questions belonged.
What the Camera Won’t See
I attended a rehearsal two weeks after the decision. The warm-ups were filmed. Daniel set up a single camera on a tripod in the corner, and for twenty minutes, the cast did their usual routine. They shook out their limbs. They hummed. They played the word association game. It looked, on the small monitor Daniel had set up, like exactly what it was: a group of people preparing to do something together.
Then the camera was turned off. Daniel stayed in the room, but he sat in a chair against the wall, notebook closed, hands in his lap. The cast moved into the first scene of the new play—an original work about a family trying to keep a corner store open while the neighborhood gentrifies around them. The actor playing the grandmother had written some of her own lines. The actor playing the developer had brought research about actual zoning changes in the city. The director stopped them twice to ask questions, not to give notes.
At one point, the grandmother actor flubbed a line and started to cry—not in character, but as herself, because the line was about losing a store, and her own father had lost a store in 1987. The room held still. Someone brought her a glass of water. The director said, “Take your time.” After a few minutes, she nodded, and they continued.
None of this was filmed. None of it will appear in the documentary. It will exist only in the memories of the people in the room, and in Marta’s note, which she wrote later that night: “R. broke during Scene 3. Line about the store. Recovered. Scene completed. No camera.”
The Archive That Chooses Its Own Shape
What the theater group understood, in the end, was something that many communities learn only after a breach: documentation is not neutral. The tools you use to capture the work shape what counts as the work. A camera privileges the visual, the performative, the well-lit. An AI transcription tool privileges the verbal, the searchable, the extractable. A script generator privileges narrative coherence over lived incoherence. None of these tools are bad. But none of them are innocent.
Marta’s library is a different kind of archive. It is incomplete. It is illegible to anyone who wasn’t there. It cannot be searched by keyword. It cannot be turned into a grant report without significant translation. But it is faithful to the thing it records. The sticky notes are still sticky. The coffee stains are still brown. The rubber band is still holding the unfinished third act together, and everyone still knows not to touch it.
The documentary will be finished this spring. It will show the warm-ups, the set construction, the costume fittings, the strike. It will be useful for grants. It will probably bring in some new members. But it will not show the scene where the grandmother actor cried. It will not show the argument about casting that lasted three meetings. It will not show the lemon bars.
And that, the group decided, is as it should be. Some records are for the public. Some records are for the people who were in the room. The skill is knowing which is which, and having someone in the corner taking notes who knows when to put a question mark in the margin and when to simply write: “No camera.”
Marta is still there. She still arrives forty minutes early with her plastic bin. She still three-hole-punches the new drafts. She still does not take bows. But everyone in the company knows: the script library is the real archive, and the keeper of the library is the one who decides what the community remembers.