The Assumptions

---

Lena swam her lengths. Every Wednesday morning she went early to the Hallenbad Oerlikon to do her laps. She still felt panic when she lay horizontal in the water and lost the ground beneath her feet. She had learned to swim late; her relationship with her body’s coordination had always been disrupted: dyspraxia. It had taken three years before she dared to lie in the water, to let herself be carried. The weekly laps were proof, for her, that her will could win over her body. She shared the lanes with the older women and men. The young, fast swimmers reeling off their kilometres didn’t interest her. The clock read 6:47. Thirteen minutes remained if she wanted to catch the bus and the S-Bahn and arrive at her desk just before 7:40.

She caught them. She always caught them.

On the S-Bahn she took out her Leuchtturm notebook — dot grid, forest green, bought in a Papeterie that had once belonged to her parents — and read what she’d noted the previous evening. “History of Regulation,” a title for a continuing education course she might offer. But apart from the title and two bullet points, there was nothing. The notebook was full of these: half-starts, quarter-starts, openings that never became what they could have been. She closed the notebook and looked up. Bahnhof Hardbrücke had just passed and they were pulling into the Hauptbahnhof.

The number 11 tram (she would never adjust to the renumbering; in her mind, the 11 still ran to Rehalp) took her to Höschgasse. Amstein & Keller occupied two floors of a Seefeld building on Dufourstrasse. The house had been torn down and rebuilt at some point in the sixties. It housed a collection of consulting firms, law offices, and trust companies. For a location closer to the lake, where the real money worked, Amstein & Keller hadn’t quite managed. But the espresso machine was good.

After depositing her bag at her desk, she went to the coffee machine. Amstein came out of his office, a thin folder in his hand. Typically a mandate he had already accepted.

«Öppis Interessants,» he said, handing her the folder with the reverence that told her the mandate was very well compensated. «Genfer Stiftig. Klimaszenario-Modellierig — Infrastruktur- und Regulierigsuswirkige bi plus vier Grad. Sehr umfangrichi Unterlagä.» He paused. «Genau dis Gebiet, Lena.» Something interesting. Geneva foundation. Climate scenario modelling — infrastructure and regulatory impact at plus four degrees. Very thorough brief. Right in your wheelhouse, Lena.

«Wänn isch das cho?» When did it come in?

«Letscht Wuche.» Last week.

She registered this the way she registered most things Amstein did: precisely, silently, without response. He had accepted the mandate, agreed the fee, and assigned her the work, all without consulting her. The sequence was so familiar she no longer perceived it as a pattern. It was simply how things moved through the firm — decisions made above, work distributed below, the membrane between them thin enough to see through but never thin enough to reach through.

She opened the folder. Fee structure, timeline, scope document, client contact. A Geneva-registered foundation with a French name she didn’t know and board members she couldn’t find on LinkedIn. The website as nondescript as usual. None of this was unusual. Switzerland had more foundations than dentists, and most of them preferred not to be found.

The data package arrived by secure transfer that afternoon. She opened it expecting the usual: a scenario brief, baseline assumptions, references to IPCC pathways. What she found was something else.

---

She pulled up the IPCC’s AR6 pathways on a second monitor and began cross-referencing. The foundation’s atmospheric composition diverged from every published scenario. Not randomly — directionally. CO₂ concentrations 14 percent above SSP5-8.5 at 2060, double the levels of the year she was born. Humidity curves that assumed positive feedback loops absent from any peer-reviewed study she could find. Temperature gradients specified not as ranges but as values, as though the atmosphere were a room whose thermostat had already been set.

She spent an hour mapping the divergences. They were consistent, which was itself unusual. Client assumptions were typically wrong in scattered, idiosyncratic ways — a CEO’s pet theory about carbon capture, an investment committee’s optimism about electrification timelines. These assumptions were wrong in the same direction, by consistent magnitudes, across every variable. Wrong the way a translation is wrong when the original is in a language you don’t recognise.

She drafted a memo. Factual, measured, the kind of document she had written several hundred times: The atmospheric composition assumptions in the project brief diverge significantly from published IPCC scenarios. Recommendation: alignment with SSP5-8.5 as upper-bound baseline, or client clarification on proprietary modelling methodology.

The response came back in less than an hour. Not from the contact who had sent her the data, but from a different email address, same domain, no signature block.

Thank you for your thorough review. Please use our assumptions. The divergence is known to us.

She read it twice. The sentence was strange in the way that only perfect sentences can be strange. Smooth, as if written by an AI. There was nothing wrong with it, nothing to object to, and yet. The divergence is known to us. Not “We believe our model is more accurate.” Not “Our projections are based on proprietary research.” Known to us. As if the divergence from published science were a known quantity. A feature, not a flaw.

She adjusted the models to the client’s assumptions and continued her work. This was what she did. Clients had assumptions; she modelled their implications. She had learned not to voice what she thought. Too often she had not been heard. She had spent twenty-two years learning to be professional.

But she saved the original memo in a folder she named Abweichungen.

---

By seven o’clock the office was empty. The radiators from the sixties beneath the windows were silent. By late October they should have been running — the oil heating in the basement normally kicked in mid-month, and the tenants on the floor below complained every year about the smell in the first days. This year no one had complained. The heating had no reason to run. It was warm enough without it, and nobody asked why.

She had opened the window to the courtyard, where nothing grew except a cherry laurel that should have dropped its leaves two weeks ago. The air that came in was warm and carried a smell she couldn’t place — not the lake, not the scents from the holistic health practice across the courtyard, something mineral and faintly sour, like a thermal spring in a place where none belonged.

She closed the window and turned back to the data.

The dataset was chronologically ordered, clean, like everything in this package was clean. Atmospheric readings, sorted by date, from 1950 to 2085. She had cross-referenced the recent past against the published scenarios and traced the divergences into the future. Now, in the quiet office, she went back to the beginning of the time series.

The atmospheric projections were specified to two decimal places.

In twenty-two years of regulatory consulting she had seen climate data in every format — briefings for politicians, rounded to the nearest integer; scientific papers, cautious to one decimal place; industry models, hedged with confidence intervals wide enough to park a Transporter in. Two decimal places was not a projection. It was a claim of certainty that no one in climate science would make, because the systems were too chaotic.

The entries for 1950 had two decimal places.

She paused. 1950. Atmospheric CO₂ measurements had been taken with absorption spectroscopy then, at a handful of stations, with instruments whose accuracy was at best ten ppm. No one could have measured 310.28 ppm in 1950. The instruments didn’t exist. The stations didn’t exist. And yet there the numbers stood, precise as yesterday’s, as if every decade had been measured with the same instrument.

She pulled up historical data from NOAA and Copernicus and laid them alongside the client’s time series. For the years where real measurements existed — from Mauna Loa 1958, from the European stations in the seventies — the client’s data matched observed values. Not approximately. Exactly. As though they hadn’t been calculated but read off.

And for the period before, when no one could have measured so precisely, they were equally precise. The same certainty. The same resolution. As though someone had measured the atmosphere of 1950 with instruments that didn’t exist then.

Then the future. From 2027, the values were identically formatted. The same precision. The same structure. The same implicit certainty. No confidence intervals. No ranges. Values.

She sat for a while in front of her computer, the columns of numbers on the screen.

In her Lic thesis she had written about ecclesiastical land registries — the Grundbücher or property inventories of medieval St. Gallen, where the Church had recorded not only what it owned but what it intended to own, entering properties into the register before the transactions were complete. The register itself was the instrument of authority. An entry created a fact. Documentation preceded reality.

She had received a five-and-a-half for that thesis. Her professor had said it was excellent work, but as a woman the six had been withheld from her. It was one of the reasons she had not pursued an academic career, had decided — without it really being a decision — to let a consulting firm pay her to read documents. What she had learned in six years at university. She could always go back to academia later. That had been twenty-five years ago.

She looked at the dataset again. Past values that matched reality — even where real measurements should not have been possible. Future values formatted as though they, too, were recordings. The atmospheric composition assumptions were not projections. They were not scenarios. They were not even aspirations.

They were specifications.

Someone had written a Grundbuch for the atmosphere. Had entered into the register the conditions they intended to create, the way a medieval bishop entered a land grant for a parish that did not yet exist, because the entry itself was the first act of construction. And the historical entries matched because they had not been measured but entered retroactively — a complete ledger, backdated without gaps, as any proper register demands.

She minimised the spreadsheet. Stood up. Walked to the window she had closed and opened it again. The warm air came in, and with it the mineral smell — stronger now, or she was more alert to it. Below, the cherry laurel held its leaves in the dark. October should have taken them.

The divergence is known to us.

Of course it was known to them. In a project, you are always aware of the gap between what exists and what you are building. That is the plan.

She closed the laptop, put on her light jacket, and left the office. The stairwell smelled of floor polish and, faintly, of the same mineral sourness that had come through the window. She descended three flights without deciding whether she believed what she had just understood.

---

When she got home, Noah was already asleep, tired after training. The door closed, the hockey bag blocking the hallway. She moved it aside, as she moved it aside every week. She opened the door, checked that he was breathing, as she did every time. She could hear him breathe — steady, calm.

She went to the kitchen and began cooking.

A shakshuka takes a good forty-five minutes if you don’t rush it. She fried the merguez she had bought fresh the day before in Kreis 5, and cut peppers, onions, and garlic, which she then sautéed in oil in the already hot pan. The kitchen filled with the smell of spices as she continued cooking. Elaborate meals for one had been, even before Noah’s time, a welcome chance to focus, to make something with her hands that wouldn’t sit around afterwards, that served not only her physical but her psychological wellbeing.

While the shakshuka simmered, she sat at the kitchen table and opened the Leuchtturm notebook. She turned past the course she might offer, past the draft of an essay on regulatory capture she had started in March and abandoned in April. She turned to a blank page.

The pen was in her hand. The page was empty. Outside, through the kitchen window she’d cracked open, the Oerlikon night was warmer than October should permit. She could hear the S-Bahn trains at the station in the distance, the sound of arrival, of departure, the silence after. The air that came through the window into the flat was heavy. Not unpleasant. Just present in a way that air shouldn’t be. Like water when you swim — surrounding, touching, reminding you it is there.

She sat for a long time with the pen and the blank page and the warm air. She thought about the Grundbuch. About the 1950 values, more precise than they had any right to be. About the future values radiating the same confidence. About the politeness of the email and the cherry laurel still holding its leaves in the courtyard on Seefeldstrasse. About the radiators that weren’t running because they didn’t need to run. About the gap between knowing a thing and doing something with the thing you know. She asked herself what she could do, or what she could not.

She did not write anything.

After a while she closed the notebook, served the shakshuka on the blue plate she liked, and ate standing at the kitchen counter, because the table was covered in things she had been meaning to put away since Sunday.

The shakshuka was good. She had done everything right.

She washed the plate, dried it, put it back in the cupboard. Checked on Noah — still sleeping, still breathing, the steadiness of him the only sound in the flat that asked nothing of her. She brushed her teeth and went to bed.

Through the bedroom window, left slightly open because she needed air to sleep, the Oerlikon sky was cloudless and mild. Somewhere south, she thought — without choosing to think it — the conditions described in the data package were already closer to reality than the IPCC knew. The specifications were being met. The atmosphere was being adjusted. Not quickly, not violently, not in a way that would make anyone do anything other than switch on the weather forecast after the evening news and hear that it was agnähm. Steadily. Precisely. To two decimal places. By whoever had written the Grundbuch for a planet they intended to own.

She lay in the warm dark, listened to her son breathe, and did not sleep for a very long time.