Chapter 27: The Wrong Count
Nine was not a lie; it was a fence built around the dates that mattered least.
On Tuesday morning, Priya and I sat in her WCT office with Dana beside us while an authorized WMG custodian shared a read-only definitions panel.
The panel supported a one-page WMG count memo delivered after Grant's explanation.
Its heading said only nine current external-stability records existed.
The number matched Friday's sample by accident; it did not describe the same group of records.
Priya placed both documents side by side without pretending WCT controlled the company database.
Grant was not in the room, but his written standard and objection were attached to the session log. Daphne's limited authorization allowed the custodian to explain the count, not export identities or underlying cases.
The definition of current carried most of the answer.
It meant a designation opened under Version Four and still active on the extraction date.
Records created under Versions One through Three were outside the count even if their labels had later migrated.
Employees marked former were also excluded from the displayed population.
That rule removed Jonah before anyone considered whether his old designation remained connected to later records.
“So there are only nine under the definition they chose,” Priya said.
“The sentence is true under that definition,” I replied.
Calling it false would have given WMG an easy correction and left the fence standing.
The number was useful only as a snapshot of one version and one employment status.
I asked the custodian to display the protocol version directory.
Four versions appeared with effective dates, retirement dates, and repository identifiers.
Version Four covered the shortest period and the smallest active population.
The panel returned no counts for the retired versions.
It also hid whether an older label had been renamed, closed, or carried forward.
Priya asked what a broader sample would cost.
The custodian said archived queries required separate approvals, restored index tables, and additional redaction review.
Expense was real; it was not a reason to describe a narrow denominator as the universe.
Two WCT trustees had already asked whether expanded review would keep donor work frozen through spring.
I acknowledged that their nonprofit budget concern belonged in the record.
Then I proposed a scope that could answer the counting question without exposing a single person to us.
The new sample would span every protocol version used during the relevant period.
It would include active and former employment status rather than treating departure as disappearance.
It would count internal-issue triggers, later stability labels, approval roles, and status transitions.
No total would be treated as proof of motive, retaliation, or harm.
The reviewer would compare process fields before deciding whether any deeper sample was justified.
Priya revised the language on her screen instead of turning my request into a speech.
She asked whether the first production needed names.
“No.”
It did not need complaint narratives, addresses, medical information, or communication content either.
It did not authorize Claire Whitmore to receive source files.
Any later production would go directly from the proper WMG custodian to a lawfully selected and authorized independent reviewer. Priya signed the expanded sampling requirement as a WCT fund-risk condition.
She transmitted it to Daphne's WMG board channel as a request for matching company authorization, not a command to WMG staff, preserving the line between entities.
The action widened the proposed review while creating no present access, reviewer, or conclusion.
The political cost arrived before the technical answer.
A WCT trustee wrote that I was converting a memorial dispute into an unlimited corporate project.
Another asked how many training grants would be delayed to finance archived restoration.
Their questions were fair enough to survive my dislike of them.
Before ending the session, the custodian displayed the retired field-name index.
Most labels described ordinary safety or communications support.
One discontinued field had no Version Four equivalent.
Its name was reference stabilization.
The name did not prove its purpose, but the current count excluded former employees like Jonah.
Dana had authorization to review Jonah's D07 file under restricted conditions the next morning.