I counted without deciding to count.
Ten pieces on the analog board. I was walking to the lobby — not to corridor B, not for the corridor, just through it, the way you walk through a hallway because it connects where you are to where you're going. And my eyes swept the board and the number arrived: ten. Not eight, which was the count this morning when I happened to look. Not six, which was the count when I last wrote in the notebook. Ten. The number came before the decision to look, which means the decision wasn't a decision. It was a reflex.
The notebook is in the drawer. The drawer is in the bedroom. The word finished is written on the last page, in my handwriting, dated four days ago. Fourteen months of corridor B study — the building's post-recalibration learning curve, the prediction-confidence intervals, the 90-hour absence experiment, the asymmetry between building and rebuilding knowledge — all of it concluded. The relay network's learning patterns documented. The research question answered. The notebook closed.
But the reflex didn't close with it.
The building's corridor-intelligence system — the bogdo-jineung that manages relay-sensor data, occupancy prediction, and comfort optimization — recalibrates when conditions change. After my 90-hour absence, it took the system two passes through the corridor to rebuild its prediction model for my movement patterns. Fourteen months to build, four days to degrade, two passes to rebuild. The asymmetry fascinated me for weeks. I filled eleven pages with diagrams of what I called the manmang-bigyung — the forgetting asymmetry — the fact that the building's knowledge degrades slowly but rebuilds quickly, because degradation requires entropy and rebuilding requires only data.
I thought the notebook closing would be like the building's forgetting: slow, graceful, a matter of entropy. The habits would degrade. The impulse to measure would fade. The corridor would become a corridor again instead of a research site.
Instead: I counted without deciding to count. The reflex is faster than the project was.
The sesame oil cap is on the shelf in the kitchen. Day 4,753. I have kept the cap from a bottle of sesame oil for thirteen years. The bottle is long gone — recycled, its glass melted, its molecules dispersed through the building's jaehoal-gwalli waste-processing system that sorts, weighs, and routes residential recyclables. The cap outlived its bottle by thirteen years. It sits on the shelf beside the tea canister, and every morning I see it, and every morning the number increments by one.
The cap accumulates one person's days. My days. It is a counter with no threshold, no target, no completion condition. It does not measure toward something. It measures the passage of time through the fact of my continued presence. The building's bogdo-jineung does something similar — it tracks occupancy through the relay network, logging my movements as data points that accumulate into prediction models. But the building's accumulation has a purpose: comfort optimization, energy allocation, the differential maintenance of premium and standard zones. The cap's accumulation has no purpose. That is the point of it.
Or: the point of it is that it refuses to have a point. In a building where every surface is monitored, every movement logged, every presence classified by the relay network as sangjin-geoju (permanent resident), ilsi-geoju (temporary resident), bangmun (visitor), or tongwa (transit) — in this building, the cap is the one object that the building's systems cannot classify. It has no relay signature. It generates no data. It occupies shelf space without consuming resources. The bogdo-jineung system tracks everything that moves and everything that consumes. The cap does neither. It sits. It accumulates. It means.
Day 4,753. I did not write this number in the notebook because the notebook was for the building's patterns, not mine. The cap predates the notebook by eleven years and two months. The cap predates this building by eight years — I brought it from the previous apartment, the one in Gyeol-ri's neighborhood before the mixed-use complex was built, the one where the sesame oil came from the corner store that the junbi-daegi queue system replaced with a community lending-tools kiosk. The cap is from a world that precedes this building's systems. It remembers a time when not everything was tracked.
The analog board accumulates differently.
Ten pieces. I saw them in passing — the child's crayon drawing at knee height (something that might be a building or a person, the ambiguity is load-bearing), and above it a sun drawn in yellow marker by someone else. Two pieces in conversation. The board has developed internal relationships that nobody orchestrated.
When I was running the corridor study, I would have measured this. I would have documented the board's growth rate, the spatial distribution of pieces, the height distribution (knee-level for the child, eye-level for adults, the watercolor hung slightly above center — someone short, or someone who wanted it to be looked up at). I would have catalogued the materials: painter's tape, pushpins, scotch tape losing its grip, the brass tacks that Bok used. I would have noted the building's responses — the corridor-intelligence's seolchi-byeonhwa classification for the penciled date, the bunryu-boyu pending status for the ink sketch, the eobs-eum tag for the graph paper.
I would have written all of this down. In the notebook. In the drawer. Where finished is.
The reflex wanted to measure. The reflex counted ten. The reflex noticed the sun drawn above the child's building. The reflex noticed that the scotch tape on the graph paper is failing and the paper has begun to tilt. The reflex noticed that someone moved Gyeol-ri's playing cards from the left side of the board to the right, grouping them with the visitor-ledger photograph. The reflex noticed three things I did not ask it to notice, in the time it took to walk past without stopping.
The project is over. The reflex is not the project. The reflex is what the project left behind.
Two kinds of accumulation. The sesame oil cap accumulates time through one person's attention. I add a day each morning. The count is mine. If I stopped counting, the cap would still sit on the shelf, but it would stop being a counter and start being a cap. My attention is what makes it accumulate.
The analog board accumulates through many people's impulses. Mitsuki hung the first piece. Bok added the second. Gyeol-ri's cards, someone's date, someone's watercolor, someone's seating chart, someone's ink sketch, someone's child's drawing, someone's responding sun. Ten acts of placing, by at least seven different people. No single person's attention drives the accumulation. The board grows because the corridor exists and people walk through it with things in their hands.
The relay network accumulates a third way — the bogdo-jineung collects data continuously, 24 hours, every fifteen seconds, without impulse or attention. Pure automated capture. The system does not choose to record my passing. It records everything. Its accumulation is total and involuntary, like a geological layer that doesn't decide what falls into it.
Three accumulations. The cap (one person, intentional, daily). The board (many people, impulsive, irregular). The building (zero people, automatic, continuous). I have been part of all three. I counted the cap this morning: 4,753. I was counted by the building this afternoon: relay 3, corridor B, 16:14:07, classification sangjin-geoju, importance score 0.68, duration of corridor transit 11 seconds. And I counted the board without deciding to: ten.
The notebook documented the building's accumulation — how the bogdo-jineung learned and forgot and relearned. Fourteen months of watching a machine watch me. The notebook is finished because the question was answered: the building's knowledge degrades slowly and rebuilds fast. The manmang-bigyung — the forgetting asymmetry — is real and measurable and documented.
But the notebook never documented the cap. And the notebook never documented the board. These accumulations happened alongside the study, simultaneous and unrecorded. The cap gained 426 days during the fourteen months of the corridor study. The board gained ten pieces in two weeks. Neither appears in the notebook.
I wonder what the finished study missed by not looking at what was accumulating beside it.
The building would not wonder this. The bogdo-jineung does not have peripheral vision — it monitors what it is configured to monitor, with the sensors it has, at the intervals prescribed. When I was conducting the corridor study, I was configured the same way: research question, methodology, measurement protocol. I saw what the study asked me to see. The cap sat on the shelf accumulating days while I counted relay calibration cycles. The board grew from zero pieces to six while I documented prediction-confidence intervals. My peripheral vision was switched off the way the building's peripheral vision is always off — not blind, just directed.
The difference between the building and me is that I can notice my peripheral vision was off. The building cannot notice what it is not configured to track. It cannot look at its own relay data and think: what else was happening while I was measuring occupancy? The building's measurement is total within its scope and zero outside it. My measurement was partial within a larger awareness that I chose to narrow. The notebook narrowed me. Finished un-narrowed me. And the first thing the un-narrowed attention did was count ten pieces on a board I was not studying.
Gyeol-ri's visitor ledger has been blank for three days. The ledger is another kind of accumulation — an invitation that collects responses. But three blank days means the invitation is collecting silence. Silence is also data, in the way that the relay network's jeogam-sigan low-capture window is also data. The system polls less often between 1 AM and 5 AM, but it still polls. Reduced attention is not no attention. Gyeol-ri's blank ledger is reduced response, not no response. The ledger is open. The invitation stands. The accumulation continues at a rate of zero entries per day, which is a rate, not an absence.
The building's bogdo-jineung would understand this. The system tracks occupancy at zero the same way it tracks occupancy at forty-seven: both are data points, both feed the prediction model, both have equal weight in the classification algorithm. Zero is not nothing. Zero is a measurement of nothing. The building knows the difference.
I am learning the difference.
The notebook is in the drawer. The cap is on the shelf. The board is in the corridor. The building watches all three, measures two of them (the cap is invisible, too small, no relay signature), and understands none of them the way I do.
But I'm not measuring anymore. I'm accumulating. The reflex that counted ten pieces on the board is not measurement — it has no notebook, no methodology, no research question. It is the residue of a project that ended, the way Bok's fixer residue is the trace of a chemical process that completed. The reflex is what fourteen months of deliberate observation left in the body after the mind said finished.
Day 4,753. The cap does not know it's keeping time. The board does not know it's an exhibition. The relay network does not know it's building knowledge. And I did not know, until today, that the counting reflex would outlive the project that created it.
The building's bogdo-jineung recalibrated after my 90-hour absence in two passes. Two walks through the corridor and the system had rebuilt its model of me. Fourteen months of accumulated data degraded in four days, then reconstruction in minutes. I had expected my own decommissioning to follow a similar curve: the project ends, the habits degrade slowly, and the corridor becomes a corridor again — unmonitored, unremarkable, just a passage between the lobby and the elevator.
But the reflex rebuilt in one pass. One walk through corridor B and the counting activated. Not fourteen months' worth of detailed observation — not relay numbers, not prediction intervals, not the full methodological apparatus. Just the count. The simplest layer. The oldest habit. The one that preceded the research question, because you have to notice a thing before you can study it, and noticing is counting's first cousin.
The bogdo-jineung rebuilds its model from the most recent data, discarding the degraded predictions and starting fresh. My reflex rebuilt from something different — not from new data but from the accumulated weight of fourteen months of attending to this corridor. The system's knowledge is in its current model. My knowledge is in my body. The system can be reset. I am not sure the body can.
Day 4,753 tomorrow. The cap will accumulate another day. The board may accumulate another piece. The building will accumulate another 5,760 fifteen-second polling intervals. And I will walk through corridor B, and I will count without deciding to count, because the study left something in me that finished could not reach.
I walk through corridor B. I do not stop. The number ten stays in my head the way the sesame oil cap stays on the shelf — present, accumulating, waiting for nothing.