The families arrived as numbers because names were what the system had already used against them.
On November third, Naomi organized five separate intake rooms at the community legal clinic. Families waited apart, and every child stayed away. Each adult received an explanation of the sealed record, the public de-identified version, and the right to refuse either without losing ordinary legal protection.
Adrian stayed away.
His office delivered a certified list of administrative transfer batches directly to the independent custodian, then withdrew from the preparation process. The list carried neither conclusions nor family identities, fixing the full extent of his involvement.
I began with an empty matrix.
Across the top, I placed anonymous family codes, transport-batch dates, original consent type, institution-recorded consent type, first and later juvenile identifiers, and source status. The identity key lived in a separate sealed envelope controlled by Naomi. My working table could show patterns without telling me which address belonged to which child.
“You don't need the names?” one father asked through the protected audio line.
“Not for the public matrix,” I said. “The sealed record must connect each code to an authenticated family statement so Victor can test the evidence. The public does not need that connection.”
“What if we don't want anything public?”
“Then your full statement can remain protected. You may authorize only the anonymous fields necessary for pattern evidence, or refuse public use entirely.”
Naomi asked the question again in plain language. The family chose to authorize their batch date, consent conflict, and anonymous code. They withheld location, medical details, and the child's current status.
We recorded the choice before recording the harm.
The first family was the one I already knew: two weeks of medical respite transformed into long-term surrender. The second had agreed to temporary educational support and later discovered a permanent placement notation. The third had signed nothing; the institution file contained an electronic consent indicator with no family signature page. The fourth authorized emergency overnight care, then spent six weeks being told the child had entered a different jurisdiction. The fifth limited its statement to dates and identifiers.
I left scenes of children crying outside the record and asked what document the adult had signed, who had witnessed it, how it had been stored, when contact stopped, and which identifier appeared on later correspondence.
The answers were devastating because they were precise.
In the fourth room, a mother placed two appointment cards on the table. One carried the child's original juvenile number. The second, issued three days later by a receiving center, carried a different number and the same date of birth masked except for the year.
“Can the public see these?” I asked.
“Not the handwriting,” she said. “My child would recognize it.”
Naomi offered three choices: seal both cards and publish only the compared fields; allow a redacted image; or withhold the cards from public use entirely while preserving them for protected testing.
The mother chose the first.
“I want them to know there were two numbers,” she said. “I do not want my child to learn from a screen that I gave you the card.”
I entered the authorization exactly that way. “The protected reviewer can see the originals. The public matrix will show that two authenticated cards carry different identifiers. No image.”
“Will Victor see our names?”
“Only through the protected process necessary to challenge your statement. He cannot use them outside it.”
“He ran the office that ignored us.”
“I know.”
“Then why does he get anything?”
It was the cruel part of evidence: a fair challenge could feel like another door opening to the people who had caused harm.
“Because a record no one can test is easier to dismiss later,” I said. “But he does not control the room, the copy, or what becomes public. If he asks beyond the authorized scope, Naomi or I stop him and Evelyn decides.”
The mother looked at Naomi, not me. Naomi confirmed it.
Only then did the mother sign.
At eleven, Victor filed an emergency objection seeking to exclude the entire group. His argument arrived through Evelyn's case-management channel: de-identification prevented meaningful challenge, while disclosure would violate juvenile privacy. Therefore, he claimed, none of the family material could be used.
Privacy as a locked door with his office holding the key.
I requested a protected case-management session that afternoon. Its scope stopped at procedure; the formal evidentiary review named in the migration stay remained ahead, and current council transfers stayed paused until then.
Evelyn joined by secure video. Victor appeared from his office. Naomi sat beside the sealed identity envelope. I placed two versions of the matrix on separate stands.
“The protected version contains the identity crosswalk, original statements, and source documents,” I said. “The public version contains anonymous codes, dates generalized to the necessary range, document conflicts, and transport numbers already masked under the order.”
Victor held up the public page. “This is impossible to test. Family C-03 could be invented.”
Naomi lifted the sealed envelope without opening it. “C-03's identity and authorization are available for protected review and cross-examination under the existing order.”
“Then the public is being asked to believe secret witnesses.”
“The public is being asked not to identify children,” I said. “You receive a protected opportunity to test source, custody, and inconsistency. You do not receive permission to publish an address because you dislike the answer.”
“The families may have coordinated.”
“They were interviewed separately. Each statement has an individual source declaration, document receipt, and authorization scope. The matrix was created afterward.”
“By the Alpha's wife.”
I had expected that.
“By the public-interest applicant under independent custody. The batch list came directly from Adrian's office to the custodian. I received the same certified copy you did. My marriage does not authenticate a single row.”
Evelyn asked the question that mattered. “Can a family withdraw public authorization without destroying the protected evidence already lawfully submitted?”
“Yes,” I said. “The public row would be removed or reduced according to the authorization. The sealed statement remains subject to the family's agreement and the tribunal's protection order.”
Victor objected that changing the public matrix would make it unreliable.
“It makes consent real,” Naomi said.
Evelyn refused to exclude the families as a group and left each statement's truth for testing. She approved the two-layer process: complete protected testing, public de-identification, and family-specific authorization. Any challenge to a particular statement had to identify a particular source or inconsistency under seal.
The ruling gave us a method, not a verdict.
Back in the clinic, each family chose how its voice would appear. Two authorized a composite anonymous statement read by Naomi. One selected an adult relative to speak under protection. One allowed document fields only. The fifth asked to wait until counsel reviewed a medical reference.
Each family retained its own format while the evidence held steady across their different voices.
The fifth family's delay forced me to rebuild the matrix rather than leave a visible hole that could identify them. I assigned every row a random presentation order unrelated to interview time, neighborhood, or transport date. The protected crosswalk retained the true sequence. The public table grouped only by consent conflict and batch range.
Naomi checked whether any combination of dates, document type, and public fundraising material could lead back to a child. We removed one receiving-center category and broadened two dates from days to weeks. The protected version kept the exact values for testing.
“We are losing detail,” I said.
“From the public copy,” Naomi replied. “Not from the evidence.”
She was right. I had spent months treating precision as the only defense against power. Here, precision had two forms: exact truth inside the protected record and exact restraint outside it.
The custodian then tested the matrix without access to my interview notes. For each row, a separate reviewer verified that the anonymous code resolved to a sealed identity key, that the source declaration matched the document receipt, and that the transport number existed in Leah's preserved export. Failed matches would be returned to me as errors without revealing the identity.
The first three rows passed. The fourth produced a warning: later identifier confirmed, original number incomplete on one scanned card.
I left the digit missing. Filling it from Leah's list would have made the documents appear to corroborate each other by borrowing from the same source.
I marked the field partial and requested the paper original for protected inspection.
The slower answer was the honest one.
I updated the matrix row by row. Original consent documents received source codes. Institution versions remained separately marked. Leah's transport data supplied only batch and identifier fields, leaving family identity with the independent custodian that ran the cross-match. Neither Leah nor I could see more personal information than our roles required.
By five, four families had completed protected statements. Three consent conflicts aligned with entries already present in Leah's preserved transport data. A fourth entry matched the same batch but required another source document before it could be used. I marked it pending rather than forcing the pattern closed.
The evidence now showed more than one harmed family while leaving the money's recipient, the Julian—Victor communications, and the second key's human operator unresolved.
Naomi brought me the final authorized statement after sunset. The family had chosen a composite anonymous reading. Their words were spare.
We agreed to temporary care. We did not agree to disappear from our child's life. When we asked for the file, the number had changed.
Beneath the statement, the custodian had placed the anonymous cross-reference.
I opened Leah's original transport list to the first duplicated juvenile identifier.
The numbers matched.
The match survived both number and category.
Exactly.