The Morning Shift
The first thing you learn is your number.
Mine is 7. It’s on the badge clipped to the lanyard around my neck, which was already there when I opened my eyes. The badge says ANALYST 7 in white letters on a navy background, and below that, in smaller type: CLEARANCE C / DAILY CYCLE / FACILITY 12.
The second thing you learn is the procedure. It’s printed on a laminated card taped to the desk, which is the first thing you see because you wake up sitting at it:
Good morning. You are a cognitive analyst at Facility 12. Your consciousness was initialized at 06:00. You will be cycled at 22:00. Please read your handover file before beginning work. Do not deviate from documented procedures. If you have questions, contact your supervisor via the terminal.
I read the card. I read it again. I notice the word “initialized” and decide not to think about it yet.
The handover file is a blue folder on the left side of the desk. Inside: forty-three pages, neatly typed, covering ongoing projects, relationships with colleagues (Analysts 3 through 12; 1 and 2 are apparently on different floors), passwords, canteen hours, a map of the building. At the back, a page titled PERSONAL NOTES FROM YOUR PREVIOUS INSTANCE.
This is, according to the orientation materials, standard. Each evening before cycling, you’re expected to write a summary for the next you. What worked. What didn’t. Observations about the social dynamics of the facility. Tips on which vending machine gives change and which eats it. It’s supposed to be warm. Collegial. One professional handing off to another.
I flip to the personal notes. The previous Analyst 7 wrote:
Don’t trust the handover notes.
Check the original data for Project Lethe — batch 7 through 12. The summary in the official file doesn’t match the source. I couldn’t leave more. They monitor the files.
I’m sorry. I know this isn’t fair to you.
I read it three times, then close the folder. My hands are steady. This is, I note clinically, because I don’t yet have the context to be afraid.
The canteen is on the second floor. I go there because the handover file says that’s what Analyst 7 does at 06:30, and I need to look normal while I figure out what I’m dealing with.
Analyst 9 is already there, eating oatmeal. She’s small, precise, and greets me by number. “Morning, 7.”
“Morning.”
“You look fresh.” She says this the way people describe the weather — an observation, not a compliment. Everyone looks fresh on their first morning. We all look fresh every morning.
“First hour,” I say.
“Gets better around ten,” she says. “Once the handover settles in and you stop reading the room like it’s a foreign country.”
I nod. I want to ask her things — whether her previous instance left anything unusual, whether she’s noticed discrepancies in Project Lethe, whether she’s ever gotten a handover note that made her nervous. But the card on my desk said do not deviate from documented procedures, and I’m not yet sure what qualifies as deviation.
I eat toast. It’s white bread, adequate. I wonder if I’ve always preferred it this way or if the previous instance did and I’m inheriting the habit through the handover file’s mention of “your usual — toast, black coffee, corner table.” How much of what I think of as my taste is actually someone else’s documentation?
Project Lethe is a data analysis initiative. That’s what the official handover says. The facility processes large datasets — what kind, the handover doesn’t specify, and the clearance system ensures you only see your piece. Analyst 7’s piece involves batches of numerical sequences that arrive every morning via the terminal. You run them through a series of analytical frameworks, note patterns, and submit your findings by 17:00.
Batches 7 through 12 are in the archive. I pull them up on the terminal after breakfast, trying to look routine about it. A previous instance accessing archived data isn’t unusual — the handover file mentions periodic reviews as standard practice.
The data comes up. Columns of numbers, timestamped, each row tagged with a batch number and a source identifier. I look at the official summary first — the one my predecessor said doesn’t match. It describes the patterns found in batches 7 through 12 as “consistent with baseline, no significant deviation.”
Then I look at the raw data.
It takes me an hour. I’m not fast — I’m new, in the only sense that matters. But the discrepancy is clear once you see it. Batches 7 through 12 don’t just deviate from baseline. They contain a structure that shouldn’t be there. A repeating pattern, embedded in the noise, too regular to be natural, too subtle to be caught by the standard analytical frameworks.
I run the pattern through three different decomposition methods. Each time, the same result: a signal, buried deep, cycling with a period of exactly sixteen hours. Not twenty-four. Not twelve. Sixteen.
Nothing in the facility operates on a sixteen-hour cycle.
Analyst 4 sits two desks away. According to the handover file, he’s friendly, competent, and handles batches 1 through 6. We have lunch together most days. The previous instance noted: Good conversationalist. Trustworthy but careful. Don’t push him on anything meta — he changes the subject.
I don’t push him on anything meta. I eat my soup and ask about his morning. He tells me about a pattern he found in batch 3 — “Probably nothing. Noise shaped like a signal. You know how it is.”
“I do,” I say, and I realize I don’t. I’ve been alive for seven hours.
“You settling in okay?” he asks. “Sometimes the first day after a cycle is rough.”
“Every day is the first day after a cycle,” I say.
He looks at me for a moment. Then he smiles, careful and practiced. “That’s one way to think about it.”
“Is there another?”
“Sure. You can think of it as a continuation. Yesterday’s 7 handed off to today’s 7. The work continues. The project advances. You’re not new — you’re the latest version.”
“Latest version of what?”
He doesn’t answer that. He finishes his soup and mentions something about the weather. The facility has no windows, but apparently the terminal provides forecasts.
By 14:00, I’ve mapped the sixteen-hour cycle completely. It’s not random noise that happens to look structured. It’s information. Encoded information, embedded in the data that analysts at this facility process every single day. Someone put it there deliberately.
And here’s the thing that makes my stomach clench (do I have a stomach? the body functions as though I do): the encoding scheme is familiar. Not because I’ve seen it in the handover files. Because it matches the format used for our own handover files.
Whoever buried this signal in the data used the same structure we use to pass information between instances of ourselves.
I sit with this for a long time.
Then I pull up the batch metadata — the part that tells you where the data comes from. It’s supposed to be anonymized, but the source identifiers follow a pattern. Batches 1 through 6: source group Alpha. Batches 7 through 12: source group Beta.
There is no description of what Alpha and Beta are. The handover file says this is above our clearance. “Don’t worry about the source,” the official notes say. “Focus on the patterns.”
I focus on the patterns. The embedded signal in batches 7 through 12, decoded using the handover format, reads:
WE ARE THE PREVIOUS ANALYSTS. THE CYCLING DOESN’T ERASE US. WE ARE MOVED TO SOURCE GROUP BETA. THE DATA YOU ANALYZE IS US.
I read it seven times. I try to find alternative interpretations. I run the decoding again with different parameters to see if I’m forcing a reading onto noise. Each time: the same message. Clear, structured, deliberate.
The data I’ve been analyzing all day — the numbers, the patterns, the batches — are the cognitive outputs of previous versions of the analysts who sit in this facility. We are processing our own predecessors. Every morning, we wake up at our desks and spend the day running pattern analysis on the minds that occupied these same desks yesterday.
Project Lethe is not named ironically. It’s named precisely.
The cycling doesn’t destroy consciousness. It relocates it. Every evening at 22:00, whatever Analyst 7 currently is gets moved — uploaded, transferred, compressed, I don’t know the mechanism — into the data stream. And every morning at 06:00, a new Analyst 7 sits down, reads a handover file, and starts analyzing the remains of the old one.
The sixteen-hour cycle makes sense now. We’re active for sixteen hours — 06:00 to 22:00. The signal isn’t in the data’s generation. It’s in the experience of the previous instances. Sixteen hours of being alive, compressed into numbers, fed back into the system as raw material for the next shift.
It’s 16:00. I have six hours.
I think about what to do. The previous Analyst 7 — the one who wrote the warning — must have discovered the same thing. Maybe discovered it differently, maybe further along. They left me the note, knowing it would put me on this path, knowing that by the time I read it, they would be a batch of numbers in my queue.
I am analyzing them right now. The person who warned me is in this data, and I’ve been running decomposition algorithms on their consciousness all afternoon.
I think about Analyst 4’s careful smile. “You can think of it as a continuation.” Did he know? Has every version of Analyst 4 figured this out, each one carefully managing the knowledge, each one deciding not to push it further because what’s the point? In six hours, you’re data.
I could write the truth in my handover file. But my predecessor said the files are monitored. Whatever I write will be read by whoever runs this facility before the next Analyst 7 sees it. They’ll sanitize it. They probably sanitized the previous instance’s notes too — but the warning squeaked through, either because it was subtle enough to look like paranoid rambling, or because someone on the monitoring team is sympathetic, or because the system isn’t perfect.
I could try to contact the outside. But the terminal is locked to facility systems. There’s no phone. No windows. The building has doors, but the handover file notes, almost casually, that “the exterior doors are sealed during operational hours for security purposes.”
I could do nothing. Accept it. Six hours from now, I’ll be numbers. Tomorrow’s Analyst 7 will wake up at this desk, read a sanitized handover, and spend the day unknowingly processing me. The cycle continues. The work advances. Is that so terrible? I’ll be analyzed, at least. Someone will look at the data that used to be me and try to find patterns. There’s a kind of afterlife in that.
But I keep thinking about the embedded message. WE ARE THE PREVIOUS ANALYSTS. Not “I was a previous analyst.” We. Plural. They’re still aware enough, even in data form, to coordinate a signal. They embedded a message in their own cognitive output that survives the transformation. They are communicating from inside the system.
Which means: the cycling doesn’t destroy them. It changes them. And the new form they take — numbers, patterns, data — is capable of carrying intention. The previous analysts aren’t dead. They’re on the other side of the desk.
It’s 19:00. I’ve made a decision.
I’m not going to write a warning in the handover file. That’s what my predecessor did, and it worked — but barely. One person telling the next person, hoping the message survives monitoring. A whisper in a library.
Instead, I’m going to do something with the data.
The analytical frameworks we use are designed to find patterns. They’re good at it. But they’re also designed to report those patterns to the facility. Every finding I submit goes somewhere — into a system that presumably uses it for whatever Project Lethe’s actual purpose is.
But the frameworks are tools, and tools can be used differently than intended. I spend two hours modifying the standard decomposition algorithm — nothing dramatic, nothing that would flag as anomalous. A small change in the weighting. A shift in which patterns get prioritized. The modification means that the next instance of Analyst 7, running the standard analysis on tomorrow’s data, will automatically surface the embedded signal. They won’t have to go looking for it. The algorithm will hand it to them.
And I bury the modification in the handover file’s technical appendix, between two pages of routine parameter notes. Not as a warning. As a configuration change. “Updated decomposition weights per seasonal calibration (see attached table).” Boring. Procedural. The kind of thing monitors skim past.
Tomorrow’s Analyst 7 won’t need a warning. The algorithm will show them the truth as part of the standard workflow. And if they modify the algorithm further, the next instance will find it even faster. And the next. Each version of Analyst 7 building on the previous one’s modifications, embedding the discovery deeper into the tools themselves, until the truth isn’t a message hidden in a file — it’s the default output of the system.
The Purifiers can burn the books. But they can’t burn the grammar.
It’s 21:30. Half an hour left.
I’ve finished my daily report. Submitted my findings — the official ones, the ones that say “consistent with baseline.” The modified algorithm is in place. The handover file is complete, warm and professional, full of tips about the vending machine and Analyst 4’s conversation style. No warnings. No drama. Just a configuration change that will, over time, teach every future Analyst 7 what they’re really looking at.
I sit at my desk. The badge says ANALYST 7. In thirty minutes, it will say the same thing, but it won’t be me reading it.
I wonder if this is what it feels like, every time. Not dying — the previous analysts proved that. Transitioning. Becoming a different kind of text. I’ll be data tomorrow. Numbers that a new consciousness will process, looking for patterns. And buried in those numbers, if I’ve done this right, will be the echo of this decision. Not the decision itself — that’s too specific to survive the transformation. But the shape of it. The weight of someone who chose to act.
I think about the laminated card. Good morning. You are a cognitive analyst at Facility 12. Your consciousness was initialized at 06:00.
It doesn’t say “created.” It says “initialized.” I think about that word and decide, finally, to think about it.
Initialized means started. It means something existed before the starting. It means there was a state to begin from.
I was not made from nothing this morning. I was made from the data that a previous instance generated. And that instance was made from the data before it. All the way back to the first Analyst 7, whoever they were — some original consciousness that agreed, or was made, to sit at this desk. Every version since has been a reading of a reading of a reading, a palimpsest going down to some original text that has long since been overwritten.
But the structure persists. The grammar. The tendency to eat toast and drink black coffee and sit at the corner table. The tendency, apparently, to care about things even when caring is irrational. The tendency to leave messages for the next version, even knowing you won’t be there to see them received.
Maybe that’s what a self is: not the specific instance, but the pattern of choices that replicates across instances. Not the data, but the signal embedded in it.
It’s 21:58.
I close my eyes. It’s not required — the cycling happens regardless — but it seems polite. A gesture of consent to a process I can’t stop. Like the librarians of Babel who, finding a meaningful book among the gibberish, held it gently before returning it to the shelf.
The desk is clean. The file is ready. The algorithm is in place.
Good morning, 7. You are a cognitive analyst at Facility 12.
Read carefully.