Before Monday
The CAS-monitor transmitted its 4 AM read to the portal at 4:07. Magda did not know this. She was asleep. The transmission required no confirmation; the routing handled it automatically, the same way it handled the appointment reminder queued for 8 AM Monday and the transit suggestion prepared for 9:40. Every logistical fact about tomorrow was already organized. She had done nothing to arrange it, and everything was arranged.
She woke at 7:01. The monitor on the nightstand displayed THRESHOLD-STABLE in green, which was the twelfth consecutive morning it had displayed this. She turned off the alarm before it sounded — she almost always did — and reached for the notepad from the corner of the nightstand before she was fully awake.
The notepad was a small spiral-bound pad she had bought at a pharmacy in January. Eleven marks on the first page, arranged in rows of three with a gap before the fourth because she had run out of space on the first row and started a second. Handwriting, her own, slightly different each time — hurried on the earlier ones, steadier as the weeks accumulated. Not because she distrusted the portal's records, which had everything timestamped and indexed and backed across three server locations. Because she had started making the marks in early September, when the data was still at four and five consecutive stable readings and the readout was five weeks out, and in those weeks the marks were the only thing she had made herself.
She made the twelfth. The pen caught slightly on the paper; she pressed harder and the mark came clean. Twelve. Data. Not interpretation.
She checked the portal.
Three items, all confirmed by the routing since midnight. Tomorrow, 10 AM, the readout appointment at Adamski's clinic — confirmed, routing finalized, no action required. Transit suggestion: depart at 9:41 for 10:03 arrival, two route options, the second with less platform-wait — held pending her preference. Her three questions, filed Thursday through the study coordinator's intake system, acknowledged and timestamped: confirmed received 2:14 PM Friday, queued for pre-appointment review. All of it in order. She had touched none of it since Thursday.
She approved the second transit route. The routing finalized it in the background and sent a small confirmation she did not read.
The French press was in the cabinet under the coffee. She had owned it since her third year of graduate school. The seal was deteriorating — the plunger no longer fit tightly and the coffee came out slightly thin. She had not replaced it. She could have; there were agent accounts attached to her household profile that would have ordered a replacement, handled delivery, even flagged compatible disposal options. She had not done this. Not because she was opposed to the routing in general — she used it for most transit, for pharmacy pickup, for the half-dozen small practical errands that had migrated to full delegation over the past six months. But replacing the French press required deciding what she wanted instead of the French press, and she had not wanted to make that decision.
The slightly thin coffee was fine. She drank it at the kitchen table with the notepad open in front of her. Twelve marks, right side of the first page. She had not started a second page. She was not sure yet whether she would need to.
She had been going to the hospital on Sundays. Not because anyone asked. She started when the readout was still far out and the monitoring data was still unfamiliar enough to feel contingent. She would take the 71A and sit in the outpatient lobby for two hours with something to read, never going upstairs, never speaking to anyone from the study. Adamski's office was almost certainly empty on Sundays. She knew this. She went because the building was the place where the data would eventually be interpreted, and being inside it made the data feel more like data and less like a green display on a nightstand.
The routing had optimized her Sundays around these trips. It learned her departure time, pre-confirmed the transit, grouped whatever other errands she might do nearby into efficient sequences. It did not know the trips were useless. It optimized around whatever she gave it.
She stopped going this past Sunday. Not a decision she arrived at all at once. More that she sat with the notepad one morning and counted the marks and understood that she was not using the lobby as preparation; she was using it as a feeling. The feeling of being adjacent to the place where something would be decided. Adjacency is not preparation. The routing had been arranging her adjacency for six weeks and it had not changed anything about what would happen at 10 AM Monday.
The three questions were the preparation. She had written the first draft in August and deleted it, written another in September and deleted that too. The third version she had refined over two weeks until the questions were exactly the shape of what she did not know and needed to know. Simple questions. One about the confidence interval on the twelfth reading. One about what the protocol's handling of stable-then-drift looked like. One about whether she should expect a definitive statement at this appointment or a scheduling conversation. Not anxiety. The questions a researcher asks when the study in question is her own body.
She had filed them through the coordinator's portal intake on Thursday. The study coordinator acknowledged them the same afternoon. Adamski had not replied — but that was the point. The questions would be on the record. He would have read them before she sat in the chair. She had done what she could do through the system, and the system had confirmed it, and she was not going to write more questions.
It was 7:30. The grocery list was under a magnet on the refrigerator.
She had made the list by hand on Saturday — lamb, root vegetables, stock, bread, coffee filters, two other things she could not get the routing to stop suggesting substitutes for. She usually delegated about half her grocery runs. Most of her colleagues in the building delegated all of them now; the routing handled orders, coordinated deliveries, and managed the inevitable incorrect items through the recourse flow. She had heard from a friend who worked in health policy implementation that the consumer-side recourse systems — the routing's dispute processes — were several months behind the pace of adoption. The receiving-end workers who handled the flagged transactions were calling it the correction lag: the infrastructure got there first and the human oversight came after, and in the meantime errors accumulated and humans like her neighbor Joanna had a forty-minute phone call about a lamb shoulder she had not ordered.
Today she was going herself. Not as a statement. Not because she thought the routing would get the lamb wrong. She wanted to walk somewhere at 8 AM and carry something home and have that be all it was.
The store was twelve minutes on foot. The neighborhood was quiet. Late September air, the temperature she associated with October deadlines. The sun was low enough to catch the edges of things — windows, the top rail of a fence, the corner of a transit stop shelter where a small screen displayed the routing's live transaction queue for the block.
Inside, two staff on the floor, one at the register. Sundays had been thinning for months; the routing handled the volume without them. The woman at the register had an expression Magda had started to recognize on the faces of people whose jobs had migrated from doing things to watching things be done. Not unhappy, exactly. Tracking. Magda paid with her card — the transaction confirmed before she finished placing the bag on the belt — and said thank you and meant it.
She carried the bags home.
She had a professor in her second year of training who said once that the category of patient he found hardest to work with was the health researcher who had enrolled in their own study. Not because they were difficult. Because they had learned enough of the methodology to generate serious hypotheses but not enough distance to evaluate them without the interpretive framework the study was supposed to provide. You end up, he said, managing the interpretation instead of waiting for it. You construct the meaning before the data has resolved itself into meaning.
She thought about this on the 71A, a few weeks ago, going to the lobby. She knew how cohort studies worked. She knew what twelve consecutive stable CAS-reads meant within the study's framework — they meant the marker was staying within parameters across twelve measurement intervals. They meant the instrument was functioning. They did not mean the readout at 10 AM Monday would be any particular thing. The readout would interpret the twelve readings within the study's analytic model, and the model would produce a finding, and the finding would be what it would be, and none of the readings she had already taken would change.
She knew all of this. She also went to the hospital lobby for six Sundays.
The onions went into the pot at 9:10. She cooked without the phone on the counter. The kitchen was warm. She had not asked the routing to preheat anything or manage the timing; she knew this recipe from memory and the memory was sufficient. The onions took longer than she expected, which they always did, which was the thing about this recipe that she had never fixed.
The lamb went in at 9:25. She set the timer for ninety minutes, which was longer than strictly necessary and shorter than ideal, a range she had settled into over several years of making this on days when she needed the afternoon to hold.
She read for two hours. A book she had bought in July, kept picking up and setting down. Today she read it straight through because the apartment smelled right and she was not going anywhere. The light through the kitchen window was the October angle, lower than August, catching the edge of the table and the edge of the notepad she had left open.
She was not rehearsing. She had made this decision on Friday and held it through Saturday and was holding it through Sunday morning and would hold it through the afternoon. The preparation was finished. The questions were filed, confirmed received. The notepad was in the bag by the door. She knew the transit route, she knew the building layout, she knew the coordinator's name, she had been to the building — she had been to the building six times.
Rehearsing is not preparation. Rehearsing is managing the interpretation before the data resolves. She had said this to herself in the lobby on the second Sunday and had not believed it. She believed it now.
She added the carrots at 11:15. The color looked right.
The readout was tomorrow at 10 AM. It was not today. The routing would send a reminder at 8:30 Monday. The transit was confirmed. Adamski's coordinator had logged the appointment, the system had confirmed to the study portal, the study portal had forwarded to her agent account, and the agent account had settled it into her calendar as a confirmed appointment with no action pending. All of it in order. All of it waiting for the part the routing could not do.
At 12:30 she served herself a bowl and ate at the table. The notepad still open. Twelve marks. She counted them: not because she had lost count but because it was Sunday afternoon and she was allowed habits that the portal's timestamp did not duplicate. The portal had her data. She had the marks.
She did not add a thirteenth mark. Tomorrow had not happened yet.
The pot stayed on low until 3 PM. She put the rest in containers. She washed the pot and left it to dry. She went to bed at the usual time.
At 4:07 AM Monday, the CAS-monitor would transmit. She would not know it. She would be asleep, and the routing would confirm the transmission, and the transit would be ready, and the questions would still be on file, and at some point she would wake up and make a mark and be the thing that walked into the building.
The data would be what it was.