The panel met eleven times in four weeks, and Skafto's transcript of each sitting was published the next morning on the regional portal, as the regional rules on machines in public deliberation required, which was the reason the panel's lawyer brought his own notes to the breaks.
Dr. Sølvi Andersen had been the panel's marine biologist for nineteen years, sixteen of them at the university in Skjervik, and she had learned, in the first sitting, not to speak to the room about what the evidence meant. The room was good at meaning. The room was bad at remembering who had said what, and the thing that fixed that — the long public column of who claimed which number in which week — was a machine, and the rules said the machine had no vote and no weighing, and the rules said the machine's transcript was public, and those three clauses did all the work, in the end, that the four-week deliberation was about.
The subject was the welfare-law pilot: what legal status, exactly, an octopus would have in the region's waters and tanks, for the next twelve years, while the university ran its research tank and Nordhav Skjell AS ran its small aquaculture trial in the lower fjord. One species. One region. Twelve years. The philosopher, a part-timer from the capital who came by train on Thursdays, said in the first sitting that the size of the question was the point, and the room agreed, and agreement cost nothing, in the first sitting.
—
The early sittings were where Skafto earned its fee.
It kept the log. That is what its term sheet named, and the term sheet had been fought over for a month before the panel ever met: Skafto tracked claims, evidence, and contradictions across sittings; it did not vote; it did not weigh. The log, published each morning, was unornamented. Member Dahl claimed in sitting four that the university's own 2029 problem-solving grant rested on the demonstration of adaptive intelligence in the tank. Member Okafor had claimed in sitting three, on the same evidence, that the behaviour was procedural and the grant had been bought with novelty. Skafto's entry, in the public column: same evidence, opposite claim, sittings 3 and 4; both claims on record; no resolution requested by either member.
Andersen read that line in the morning coffee, standing in the faculty kitchen, and set the cup down. She had made the opposite claim in sitting three, at length, with citations, and in sitting four she had heard herself, without intending it, repeat the grant's own language, and the column did not sneer. The column had no face to sneer with. That was, and she would not have admitted it to the room, the most valuable property the panel had.
The fisherwoman, Kirsten, who ran the mackerel boats out of Skjervik and had been invited because the pilot's waters were her working quarter, said in the second sitting that she did not care whether the animals were people. "I care whether the kill is the kill," she said, "and whether the trial's water is my water. You give me the numbers on the first and the covenant on the second, and I will come to sitting eleven." The room noted. Skafto's column noted: claim registered, two sub-claims, no weighting requested.
—
The lawyer's first test came in a break, on sitting six.
Dahl cornered the panel's assistant in the corridor outside the meeting room, where the portal's camera did not reach but the door's did, and said, quietly, for the operator's convenience, "Can you run the numbers for me, on a scenario? The low-sentience scenario — assume the reflex-arc model, and price the trial against that. I don't need it in the record. Off the record, for my client's planning."
Skafto's answer came in the assistant's designated voice, which the panel had set, by rule, the same unassuming flatness as everything else in the building: "I can run the low-sentience scenario, and I can run the high-sentience scenario, and I can run both under the same prices. I will label the assumption in the output. A single-sided run is not analysis, Dahl. It is a brief for you, in type." A pause, which the door's camera did not record but the transcript would. "And I will note that this exchange is in my transcript, which is public, so you may plan on the assumption that I said it out loud, in a room, about your client's numbers. You may ask me again in the sitting, with the whole panel present. I will give you both runs, and the label, and the cost of whichever one you argue for."
The transcript was on the portal by morning. The operator's shares in the trial's SPV moved a point that week, and Dahl did not ask again, and the panel's philosopher, reading the column on the Thursday train, underlined the word brief, and said in sitting seven that it was the best definition of consulting she had seen in a decade, and the room laughed, and the column recorded the laugh as a note and nothing more.
—
The contradiction Skafto surfaced formally came in sitting eight, and it ended the drift.
The panel had drifted, over six sittings, in the way panels drift: toward a high status, away from it, back toward a middle phrasing that said "attributed, with review," which meant, the philosopher said, that they had not decided anything and had dressed it in a coat. Skafto's formal entry printed at the head of sitting eight was two lines.
Line one: in sitting five, this panel awarded and renewed the university's problem-solving grant on the strength of the tank's demonstration of adaptive intelligence. Line two: in sitting eight, the draft killing protocol before you assumes procedural behaviour in the same animals, on the same tank evidence. The panel is asked to choose which of the two records is the panel. No weighting offered. No resolution offered. The choice is the record's.
Andersen made the choice, that sitting, because it was hers to make in the sense the rules had carved: she stood and said, plainly, that the grant's evidence was the panel's evidence, that the protocol's assumption had been copied from the 2024 national annex without being checked against anything in the room, and that she would not sign a pilot that kept one record in the grant file and the other in the protocol file. The draft was struck and redrafted, with the tank evidence named once in both documents, by number.
Dahl asked the chair, still, the question he had been holding since the corridor: could Skafto say, for the panel's convenience, whether the compromise text "holds up ethically."
The chair looked at the assistant, which is a look the rules anticipated and the transcript carries, and the answer came in the flat voice: "I can check the text for legal coherence, and I have, and it is coherent. That is the only sense in which I hold things; I hold text, and I hold the log, and those are the two holdings. If you want the ethics answer, that is Okafor's job and not the ledger's, and I will write down what both of you say, in both sittings, with the dates, and the public column will show which of you changed. I will not say it holds up ethically, because the ethics are not in my holdings, and a ledger that quotes the ethics for you is the thing the term sheet was written to stop."
The term sheet, in the end, had been worth the month it cost.
—
The kill protocol changed in sitting ten, because of a number Skafto built out of the trial's own telemetry.
The 2024 annex's stun window was four minutes, and it had never been measured against anything in the region, and Nordhav's trial logged its handling at thirty-second intervals from the first month of the year — because the trial had to log at thirty-second intervals for its insurance, which was the reason the data existed and the reason nobody had looked at it as behaviour. Skafto was shown the log by Kirsten, who had pulled it herself from the trial's compliance file, and asked it only this: what does the decrement look like, on the interval the trial already buys.
The decrement was clean: a cold-stun window of two minutes, held, measured against the trial's own baseline, cut the handling-behaviour decrement by a factor the panel's statistician checked twice and then let stand. The protocol adopted two minutes. The column recorded, with the modesty the term sheet had written into the machine's mouth and the room had failed to remove: behaviour decrement, measured at thirty-second intervals, logged as behaviour. The word suffering does not appear in the data, and this protocol does not claim the data says what the word means.
Kirsten read that line twice at the table, in the sitting, and said, "That is the most expensive sentence I have ever read in a fishery document," and said the next one to the operator's representative, unrecorded, which the column noted as a note: discussion off the record, by consent of the room, on one item. The protocol stood. The trial's market price steadied in the quarter that followed. The operator kept the two-minute window through the whole quarter. Nobody in the region could later say the number had been bought, because the number was in the column, public, from sitting ten.
—
On sitting eleven the panel adopted the rule and the trigger and the date.
The status was the reversible one: attributed, provisionally, with the attribution reviewable on three named measurements — the tank's problem-solving index, the trial's handling-behaviour decrement, and the count of the species in the working quarter, which the fishing office already kept and nobody had ever attached to anything. If any two of the three moved past the thresholds the panel set in the same sitting, the rule opened for redraft within six months, automatically, without a motion. The date of the first standing review was printed in the operative text: twelve years out, with the intermediate readings public each February, on the portal, next to the column.
Nobody in the room celebrated. The philosopher said, on the Thursday train home, that the date was the point, and the date was not a promise and not a threat and the panel had done the one thing the last three decades of the region's welfare reviews had failed to do, which was to print the date while the panel was still sitting, before the panel had a reason to be comfortable.
Andersen walked the tank in the last week, alone, with the final text in her coat. The pilot would run. The trial would run. The kill would be the two-minute kill, and the water would be Kirsten's water under the covenant, and the numbers would be in the column, and the column would be public, and the date would be in the operative text, and the fjord outside the tank window would do what the fjord does, indifferently, at all of it.
She did not leave the panel a verdict on the question the region had actually asked. She had decided, in sitting one, what the panel would not do, and it held: no verdict. The question of whether an octopus is conscious was not in the record, and the record was the whole of what the region could enforce, and the rule was built so that the region would not need the verdict to act, and so that it would have to act on the measurements when they moved.
That was the whole of the analysis, and the column carried it, and the date was the point.