The Signature

He arrived at 7:11am. The office floor was empty. The queue system hummed on the wall-mounted screen; the cases had updated overnight, the Priority 1 flag still there, still his.

He did not take his coat off. He sat down at his terminal, set his bag on the floor beside the desk, and opened the wetland case.

His annotation was where he had left it. He read it from the beginning.

It was better than he remembered. Cleaner. The section on the confidence interval held up — the argument about where the 15.3% concentrated, the data gaps he had named specifically, the eight-year outcome window and the year-nine case that wasn't in the projection. He had been precise about that. He read the section about Banshtala. The woman with the smoker. I watched her for thirty seconds and I do not know what she was reading in the canopy or what it told her. He had not tried to make it more than it was. He had written what he observed and what he could not observe and he had said that the difference mattered. He had said it plainly and without apology. He read the line: You do not have a variable for what leaves with the people. It still said what he had meant it to say.

He read the final section. What he was prepared to be wrong about.

He read it slowly because it was the section he had known he would read again.

He opened the model's recommendation. He had read it twelve times. He read it a thirteenth.

The net habitat gain projection: core system, 2045–2060, the specific numbers in the third column of the table. The ecosystem resilience assessment, the confidence band around the recovery trajectory. The buffer zone viability at 31%, the trajectory without intervention, the argument for managed surrender — not abandonment, the documentation was careful about the distinction, the word managed doing real work in that phrase. The argument was sound. It had always been sound. He had not been deferring because the recommendation was wrong. He had been deferring because he had needed to understand what he was signing in full, and he had not been ready to understand it, and now he was, or as ready as he was going to be.

He closed the model recommendation.

He opened the signature field.

He signed at 7:34am.

The system processed for a moment. Then: Decision recorded. Implementation protocol initiated. Notification to regional coordination: pending.

The case closed. The Priority 1 flag was gone.

He sat for a moment with the terminal in front of him and the empty floor behind him and the queue on the wall screen, which now showed two Priority 1 items, which meant two new items had arrived since yesterday, which was the queue and the queue did not stop.

He put on his coat.

In the lift he checked his personal device. The notification was already there — faster than he had expected, the coordination system automated, no bottleneck between his signature and the next protocol.

Buffer Zone Transition Plan BR-7 formally initiated. Coordination timeline: 24 months. Implementation lead: Regional Authority Dhaka. Affected communities: notification scheduled 48–72 hours. For queries contact: caseref-BR7@authority.gov.bd

He read it twice. The notification was correct. It contained the information it was supposed to contain. The language was exact and had been reviewed and would be used in the 40,000 notifications that would follow it, the 40,000 that would say your community is affected in the passive construction that distributes consequence across a system and lands in a specific household and sits on a specific table and is read by a specific person, whose name the automated message did not contain and had not been designed to contain.

The lift reached the ground floor.

Outside, the Dhaka morning. February still, the air close and carrying moisture from the east — a weather front from the Bay moving inland, the kind of cloud cover that made the city feel lower than it was, the towers of the northern districts blurred at their tops. The street was already full. The automated freight vehicles in the dedicated lane, the pedestrian crossing, the food stalls open and doing business, a school cohort moving in a directed group across the junction — uniformed, orderly, the system running.

He stood on the pavement outside the building. He did not need to be anywhere for forty minutes.

He put his device in his coat pocket.

He thought about the annotation. He thought about whether he had said what needed to be said — not to the model, not to the record, but in the sense of whether it was an honest account of what he had understood. He thought he had. He thought the section on the year-nine outcome data was the most important part and he had not rushed it. He thought the direct address — you do not have a variable — had been the right choice, not because it changed anything, but because the record should say that too: that there was someone who noticed. He would probably be wrong about some of it. He had said that. The record would show that he had said that. He would read it again in two years, when the first coordination reports came in, and he would find out which parts he had been wrong about, and he would know he had named the right things as uncertain and the record would show that he had known.

He would not know her name then either.

He reached into his coat pocket and found his device and then did not take it out. He left it there. He stood for another moment with his hand around it and the Dhaka morning in front of him and the freight vehicles in their lane and the school cohort gone now, turned somewhere he couldn't see.

He took his hand out of his pocket.

He started walking.