Independent Verification

Six days was not long enough for a case this old, and everyone in the room seemed aware of it, which did not make the room any less uncomfortable to sit in.

There were three of them behind the review table and one of her in front of it, which was the standard configuration for a reclassification hearing and did not, on its own, mean anything. What meant something was the way the review chief opened with the transcript of her original log rather than the waveform itself: Recommend reclassification. Signal exhibits structured repetition with variation inconsistent with mechanical origin or degraded beacon function. Read aloud, in a flat procedural register that was not unkind, it sounded exactly like what it was — a Principal eleven days into rank, four hundred and twelve files into a shift, filing a professional opinion that happened to arrive by way of a headset instead of a spectrum display.

"Talk me through the method," the review chief said. Not hostile. Careful, in the way people are careful when they have already decided the finding is probably right and are still obligated to make sure it survives contact with a harder question than is it true.

She talked them through it. The four seconds. The phrase that came back eleven seconds later, almost the same and not quite. She did not say it sounded like music to the review board, because she had learned somewhere on the way here — she was not certain exactly when — that a Principal who wanted a finding to survive review did not hand a table of three careful people a sentence that easy to dismiss. She said: periodic structure with non-identical recurrence, consistent with authored variation rather than mechanical repetition or signal degradation. Same finding. Harder to file under noise.

They put the raw waveform through independent analysis while she sat and did not fill the silence, which she had also learned somewhere, and which cost her more than talking through the method had. Twenty-two minutes. The confirmation came back exactly where she already knew it would.

"The classification stands as reopened," the review chief said. "For the record — your log entry did not claim intelligent origin. That distinction is the only reason this survived first pass. If you'd written it's a message, we would not be having this conversation; we'd be having a different one, about you." A pause that was almost, not quite, approval. "Keep filing the distinction you're actually entitled to. Not the one that would make a better story."

She said she would. She meant it. She had never wanted the better story; she had wanted the true one, which was a smaller thing and harder to be believed about, and she had just watched exactly how much smaller and harder in real time.

Walking out, she thought, not for the first time this week, about a hillside she was not supposed to think about on duty, and a rule that lived there: you don't ask why until you're ready to carry the answer. She had asked. The review chamber had made her carry it in front of three people who outranked her, in language flat enough to survive them. That was, she was beginning to understand, simply what carrying it looked like from the inside of an institution instead of a hillside.

The message from the Prime Ganglion's office was two lines, and Sohrel's central brain read a great deal into the fact that it was only two.

Noted: Administrator-Voice Sohrel's authorisation of independent verification, Sentinel file, reversing a classification standing three centuries. No further action at this time.

Noted, not approved. No further action, not no concern. Three centuries of holding the Halyx portfolio had taught Sohrel to read a two-line message from the Prime Ganglion's office with the same care Shepherd's review board had apparently learned to read a Ranger's log entry — for the shape of what was not being said, which was, on both sides of the Halyx Relay Station's liaison channel this week, considerably more informative than what was.

What was not being said: that an Administrator-Voice who reopened a filing that long, on evidence that thin, and turned out to be correct, had still spent something the Prime Ganglion's office intended to remember spending. Correct was not the same account as costless. Sohrel had known this before authorising the verification and had authorised it anyway, which the third dorsal arm's ganglion, humming faintly beneath the desk exactly as it had six nights ago, seemed to regard as sufficient grounds for something adjacent to satisfaction.

The central brain did not fully share the sentiment. The central brain was doing arithmetic the limb did not need to do: three centuries of correctly-drafted, defensible, procedurally sound answers, spent now against one filing that had needed the opposite of that instinct to ever be looked at twice.

A fourth internal request had arrived that morning, forwarded without comment from an analyst two administrations over, asking — again — whether Meridian's reef-cities might be consulted on the Sentinel's structure, given the parallel to claw-percussion communication that nobody working the case had stopped noticing. Sohrel's central brain drafted the same answer it had drafted three times before: procedurally sound, correctly worded, and no.

The limb-ganglion beneath the desk did not argue this time. It simply went still, the particular stillness Sohrel had learned, across three centuries, to read as the shape of a disagreement choosing not to spend itself on a fight it already understood the administrator would lose on its behalf regardless of who won the argument.

Sohrel filed the response. Some costs were worth spending on a Ranger's hunch that turned out to be a signal with structure in it.

Some were not yet worth spending on a reef-city that had not asked to be listened to any more carefully than the Sentinel had.

Not yet was doing a great deal of work in that sentence. Sohrel was, if anything, three centuries better than most at noticing exactly how much.