The form got shorter again.
This was the current version. No one had told her what version number it was.
Folake had been counting. Not obsessively — she didn't keep a separate log, didn't mark each change in a notebook. She had been counting the way you count things you're not supposed to notice: peripherally, involuntarily, with the part of the brain that pattern-matches before the conscious mind can say stop.
Version 14 was the one in use when she started at the Witness station, two years ago. It had a cover page. It had a glossary. It had three sections: personal circumstances, signing context, and a separate acknowledgment for families with children under twelve. Forty-seven fields total. She knew this because she had counted them, two years ago, when the form was new to her and she was learning to walk families through it.
The current version — she had not been told its version number, but she knew it was not fourteen — had nineteen fields. The cover page was gone. The glossary was gone. The acknowledgment for families with children under twelve had been folded into a single checkbox: This signing may affect minors in the household. I understand this.
The family that had come in last had signed without reading. This was the third time this week. When the form was forty-seven fields, no one signed without reading — there was too much to sign, too many places to put initials, too many moments where the pace of the thing forced you to engage. Nineteen fields could be completed in four minutes if you moved quickly. The family this afternoon had completed it in three.
Folake submitted the records request on Wednesday morning, before the station opened.
She had considered submitting it informally — asking Coordinator Benwa directly, in the hall, whether she could look at the previous versions. Benwa was not obstructionist. He would probably have said yes. But informal requests existed only in memory, which meant they could be forgotten or misremembered, and she had decided that she wanted a formal record of having asked.
The request form for historical administrative documents required a stated purpose. She wrote: Comparative review of intake documentation for operational purposes. This was accurate in the way that accurate things sometimes fail to describe what you actually mean. What she actually meant was: I want to know what was removed, in what order, by whom, and whether anyone documented why.
She sent the request and opened the first family file of the day.
The records came back on Friday afternoon, three days after she submitted the request.
Versions 1 through 8 were labeled Prototype and had no author attribution. Versions 9 through 12 were labeled with a committee name — the Signing Process Standardization Working Group — and had revision notes attached, which was useful. Versions 13 and 14 had author attribution (Coordinator Benwa's predecessor, a woman named Adaeze whose departure had preceded Folake's arrival by four months) and revision notes that read like genuine documentation: Removed field 23 (emergency contact duplication) after analysis showed 94% of respondents listed same contact as field 17. No informational loss.
Versions 15 through the current version had no revision notes. They had version dates, nothing else.
Folake sat with this for a moment. Not with alarm, exactly. With attention.
The committee had documented its reasoning. Adaeze had documented her reasoning. The current version process had no documented reasoning, which could mean the reasoning was obvious, or that it was stored elsewhere, or that no one had asked for it.
She made a list of the removed fields. Not removed from 14 to the current — removed at each step, in order, from 15 onward. The pattern was not random. The fields that had been removed first were the ones that required narrative responses: In your own words, describe the circumstances leading to this signing. What questions, if any, do you have about the process? Then the fields requiring acknowledgment of specific risks. Then the glossary. Then the children's supplement. Last, the cover page.
The form had been simplified, iteratively, starting from the places where families had to use their own words.
She did not bring this to Benwa on Friday. She brought it to Benwa on Monday, after the weekend, after she had decided what she was actually asking.
He was a careful man. He listened without interrupting, which was a quality she had come to associate with people who had heard difficult things before and had learned that the first response mattered.
"Where are the revision notes for version fifteen onward?" she asked, when she had finished walking him through the list.
"I don't have them," he said.
"Do they exist?"
He paused — not evasively, she thought, but carefully. "I received the updates from the central office. The versions came with implementation instructions, not revision documentation."
"Did you ask for the documentation?"
Another pause. "I should have."
This was the thing about Benwa that she had come to rely on: he did not defend decisions he hadn't made or couldn't explain. He acknowledged the gap directly, which was either a quality or a limitation depending on what you needed from him.
"I'd like to submit a request to the central office," she said. "For the revision documentation on versions fifteen through current."
"That's your right," he said. "I'll co-sign it if you want."
She had not expected that. She said yes.
The co-signed request went out on Monday afternoon. She did not know how long it would take to receive a response, or whether a response would come. Formal requests sent to central offices had a way of moving in parallel with the work rather than intersecting it.
She continued working her station. The families came in. The current version of the form moved through their hands in three or four minutes. Some of them read it. Some of them didn't. The ones who didn't read it signed it the same way the ones who did signed it: pen to paper, name complete, date.
She had started, without intending to, keeping a count.
Not of the signings — she had always kept that record, it was part of the job. She had started keeping a secondary count: how many families, per week, completed the form without pausing at any field. Not reading it start to finish — she hadn't expected that since version twelve, probably. Pausing. A hesitation that meant the field had caught something, raised a question, cost the family something to answer even if they didn't ask it aloud.
The current version: three pauses per week, on average, across all families. She had five weeks of data now.
She did not know what version 14 would have produced. She knew that forty-seven fields with narrative responses produced a different pace than nineteen fields with checkboxes. She could not prove causation. She had the count, the version history, the list of removed fields, and a co-signed request for documentation that might or might not come back.
The form got shorter again.
This was the current version. No one had told her what version number it was. She was watching it happen, in the particular way that witnesses work: not preventing, not intervening, not yet. Gathering what could be gathered. Making it available to the record.
What you did with the record was a different question. For now, she was the record.
The second family of the week came in on Tuesday.
Father, mother, a teenage son who sat in the chair beside them with his phone face-down on his knee, which was the particular posture of a teenager who had been told this was important and was trying to be present without committing to it. Folake had seen this posture many times. She found it more honest than the adult postures that performed engagement while the eyes drifted.
The father picked up the pen before Folake had finished her standard introduction. She completed it anyway — not because she thought he needed to hear it, but because the introduction was part of the record too: she was required to confirm that it had been given.
The form before you is the current intake document for the Signed registry. Please read each field carefully. I am available to answer questions at any point. You may take as long as you need.
The father nodded. Signed. Passed the form to the mother, who signed. Passed it to the son, who signed without looking up from the form, which at least meant he had looked at the form.
Three minutes and forty seconds, start to finish.
Folake marked the session in her secondary count: zero pauses.
On Wednesday she received a response from the central office.
Not the revision documentation she had requested — a response to the request, which was different. The response said that the revision documentation for the current form series was maintained at the regional level, and that her request had been forwarded to the regional administrative coordinator for the New York district. A response from the regional coordinator could be expected within thirty business days.
Thirty business days was six weeks. She noted this and filed the response.
Benwa stopped by her station at the end of the day. He had seen the forwarded response — it had been copied to him as co-signatory.
"Six weeks," he said.
"Yes."
"That's not unusual for a cross-level documentation request."
"I know."
He stayed at her station for a moment longer than was necessary, which she had come to understand meant he was deciding whether to say something. She waited.
"The form changes were implemented about fourteen months ago," he said. "The instructions from central said the revision was part of a pilot program to improve completion rates."
Completion rate. She had not heard this framing before. She turned it over. A form with lower completion rates was a form that families didn't finish — walked out partway through, or returned it blank, or never came in at all. A form with higher completion rates was a form that families completed.
The current version had very high completion rates.
"Did the instructions say anything about comprehension rates?" she asked.
"No."
She nodded. He left.
She stayed at her station for another twenty minutes, after the last family of the day had gone, looking at the secondary count she had been keeping. Pauses per week. Five weeks of data. Average of three. She did not know what the number meant in isolation. She knew what it felt like to watch a family sign a form they hadn't read: the pen moving through the fields at a pace that had nothing to do with the words on the page.
Completion rate was a real metric. It measured something real. She did not doubt that.
She added a new column to her count: completion time, minutes. She would need more weeks of data before the number said anything useful. But she wanted it in the record.
The third family of the week came in on Friday.
Mother and adult daughter. The daughter was the one who had called ahead to schedule; she had done the paperwork, asked the right questions, arrived with the supplementary documents Folake had mentioned in the phone confirmation. She was the kind of person who read things.
She read the form.
Not quickly — she paused at field six (Nature of the signing circumstance, if voluntary), which was one of the surviving fields that had been in version 14. She did not ask a question aloud, but she paused there long enough that Folake could see her working something out. Then she continued. Finished. Signed. Passed the form to her mother, who signed without reading, but who had, Folake thought, been briefed by her daughter in the car.
Eight minutes.
Folake marked the session: one pause. Field six.
After they left she wrote in her secondary log: Field 6 pause — survivor from v14. Narrative-adjacent. She was not sure what she would do with this observation. She added it anyway.
The regional coordinator's response arrived in five weeks, not six.
It was two paragraphs. The first paragraph acknowledged receipt of the documentation request. The second paragraph said that the revision documentation for the current intake form series had been prepared in conjunction with a third-party process improvement consultant and was considered proprietary to the consulting engagement. A summary of the changes was available upon request from the signing program director.
She forwarded the response to Benwa.
He called her into his office that afternoon, which was unusual. He had a copy of the response printed on his desk.
"A third-party consultant," he said.
"Yes."
"That would explain why there's no internal revision documentation."
"It would."
He looked at the printout. "Do you want to request the summary from the program director?"
She had thought about this for the forty minutes between receiving the response and the end of the last family session. The summary would be a different kind of document than the revision notes — it would tell her what the consultant had found, not what had been decided and why. It might still be useful. It would definitely be another formal request, another response window, another document in the file she was quietly building.
"Yes," she said.
He co-signed again without being asked.
The count was now nine weeks.
Average pauses per session: two point six, adjusted for the third family who had read the form carefully. She had written a note beside that session in her log: Outlier — high-engagement family. Daughter drove the pace. Arrival with supplementary documents. She did not remove it from the average, because she was not trying to produce a clean number. She was trying to produce an honest one.
Completion rates were high. The form moved quickly. Families left with the documentation they needed, signed, in order.
She continued working her station. She continued keeping her count. The request to the program director was filed and would return in whatever time it returned.
What she had, after nine weeks, was a count, a version history, a list of removed fields, two co-signed requests, two responses, and a note to herself that said: Field 6 pause — narrative-adjacent. She had a colleague who co-signed requests without being asked to explain why. She had a secondary log with a new column.
None of this was evidence of anything, yet. Evidence was a different stage. This was the stage before: observation, documentation, the slow accumulation of the record that evidence would eventually require.
She was a Witness. This was, as best she could tell, what that meant.