The gallery gave me three walls.
I asked for three walls because that was what the exhibition needed — coded prints on the first, empty U-frames on the second, the open ledger on the third — and the gallery director said yes without hesitation, which should have been my first warning. When someone agrees to your plan that quickly, it means your plan is legible. And legibility, I have learned over eight weeks of distributing corridor prints to nine people who did not ask for them, is where the interesting things stop happening.
Gyeol-ri understood this before I did.
She had been building her index for three weeks by the time I asked to see it. Twelve pages of notation in a system she invented to describe what happened when prints left my hands and entered other people's lives. Each page tracked a single print: who received it, how they received it, what they did with it, and — this was the part that made me sit down — what the print did to the act of receiving.
She coded the departures. L for lifted — the recipient took it from the surface where I left it. R for requested — they asked, and asking changed the transaction entirely, introduced a social weight the system had to account for. H for held — they kept it without looking, which she distinguished from D, for displayed, which meant they looked and chose a place for it in their living space. P for passed — they gave it to someone else, creating a secondary departure that Gyeol-ri tracked with compound codes. And U for unknown — the prints that left and generated no observable return.
Four of the twelve pages were U.
I distributed ten prints. Six are coded. Four are not. The four uncoded prints are the ones where the corridor is still working — where the exchange produced something that cannot be captured in Gyeol-ri's system or mine or anyone's. They are the prints that departed most completely.
The gallery's AGI-managed lighting system had already mapped my three walls. It does this automatically whenever an exhibition is registered — reads the submission metadata, calculates optimal lux distributions based on medium type and viewing distance, generates a draft environment that the director can adjust. The system flagged my empty U-frames as a potential error. Four frames with no content data triggered a gentle suggestion: perhaps the image files had not been uploaded? Perhaps the frames were placeholders awaiting final prints?
The lighting AGI was correct, from its perspective. An empty frame is an incomplete frame. The system has no category for deliberate absence — only for content that hasn't arrived yet. I told the director the system's suggestion was wrong. She overrode it through the gallery's curatorial interface — a quick swipe on the wall-mounted control panel that every Seam gallery runs beside the main entrance. The AGI accepted the override and recalculated, treating the empty frames as reflective surfaces and adjusting for the way light behaves when it hits glass with nothing behind it.
That recalculation was more honest than my original plan. The AGI understood something I had not yet admitted: empty frames are still frames. They still catch light. They still organize attention. They are objects pretending to be absences.
Mok-do's print was the first to resist coding. He received Photo 7 on a Tuesday. I left it in the hallway outside his unit, leaning against the wall beside his door, in a clear sleeve with no note. Photo 7 shows the south end of the corridor at 4:15 PM — the angle where natural light from the original windows and the volumetric Lived panel light compete for the same surface, and neither wins. It is the photograph I am most uncertain about, which is why I gave it away first.
Three days later, Mok-do returned a 2041 newspaper clipping. A fidelity salon advertisement with the word APPOINTMENT circled in blue pen. No note. No explanation. The clipping arrived in the same clear sleeve, leaning against my door in the same position.
Gyeol-ri coded the delivery as L. But the return — what do you code a return? The system tracks departures. Mok-do's response was not a departure. It was a correspondence. He was telling me something about the nature of what I'd sent, and what he was telling me was: you don't make appointments with corridors.
The newspaper clipping is from before the Lived panels were installed in this building. The salon in the advertisement operated on haesamdo-tiered pricing — different fidelity levels at different rates, budget-grade Liveds in the waiting room while premium clients got full-spectrum immersive consultations. A practice that seems quaint now but was controversial in 2041 when the tier system was still settling into the class structure it would become. The word APPOINTMENT is circled in ink that has faded to a color between blue and gray, a color I have been trying to reproduce in the darkroom for six years and never achieved. Mok-do's pen, aging in a drawer for decades, did what my chemicals cannot.
I photographed the clipping beside the distribution ledger. Photo 10. His annotation visible, his blue-gray circle touching the edge of my handwriting where I logged his print's departure. Two documents side by side, neither complete without the other, neither legible to anyone who wasn't present for the exchange.
That was when the exhibition changed.
I queried the gallery's exhibition management system — a standard curator-facing interface that every gallery in the Seam runs, connected to the building's spatial AGI for foot-traffic modeling and environmental optimization. Current booking: three walls, twelve running meters, standard gallery lighting, opening date in six weeks. I asked the system to model a single wall. It returned three configurations within ninety seconds, each optimized for different viewer densities and dwell times. None of them accounted for what would be on the wall because I had not told the system what would be on the wall.
What would be on the wall is nothing. Or rather: three descriptions of nothing.
Three walls is too many. Three walls implies a sequence — first you see this, then this, then this. Sequences are narratives. Narratives are legible. The corridor is not legible, and the distribution was my attempt to find out what happens when you stop trying to make it legible. You give something away. You see what comes back. You do not plan the exhibition until the returns have stopped arriving.
The returns have not stopped arriving.
Gyeol-ri cut her index to four pages last night. She had been working with twelve, one per print, but four pages capture the entire system — the four where her notation is already its own language, where the codes have evolved past their original definitions into something she cannot fully explain to me. Plus the dormant symbol page from Saebyeok's archive, which uses a notation system that predates Gyeol-ri's by decades and which she discovered shares three symbols with hers, arrived at independently.
She is editing like a corridor. Keeping only what changed the person who passed through. Removing everything that merely recorded.
Saebyeok's archive runs on a cataloguing protocol that auto-generates entry numbers and cross-references — standard archival AGI used across the Seam's cultural institutions, trained on Korean museum taxonomy with local tteum-specific extensions. Entry 55 has been open for weeks — no closing notation, no summary tag, no archival disposition. The system flags open entries after fourteen days with increasing urgency levels. Saebyeok has been dismissing the flag each morning, a single tap on the archival tablet that resets the timer. The archive's own infrastructure is trying to close what Saebyeok insists must stay open. An argument between a human archivist and an automated system about when documentation ends.
I asked Saebyeok about it. She said: the study is not over. I said: the study is over, Chae wrote a letter, Gu dismantled the haptic array, the piezoelectric still measures 18 Hz but nobody is interpreting the data. She said: the entry stays open until the building stops attending to the space. I asked how she would know when that happens. She said: I won't. That's why the entry stays open.
Entry 55 is the exhibition's fourth absence, the one I did not plan.
So: not three walls. One wall. And on that wall, three absences.
The first absence is where the coded prints would have been. Six photographs with Gyeol-ri's departure codes, mounted and labeled. I can describe exactly what would go here. The south-end woman's print, coded D-R — displayed, then requested by a neighbor, creating a secondary transaction that Gyeol-ri invented a compound code to capture. The hallway print Mok-do received, coded L with no return code because the coding system does not accommodate returns. The four others, each with a clean letter in Gyeol-ri's hand.
They are not there. The wall holds the description of what would be there. Text on white. The dimensions listed, the codes defined, the methodology explained. Everything you need to understand the work except the work.
The second absence is where the U-frames would have been. Four empty frames, one for each uncoded print — the prints that departed beyond the reach of notation. I originally planned to hang actual empty frames — the same dimensions as the prints, the same framing material, but with nothing inside. The emptiness would be the point. The corridor's effect reproduced in gallery space.
But the gallery's AGI had already taught me: empty frames are still objects. They catch light. They organize attention. The system calculated their reflective properties and adjusted the environment around them. If an AGI treats your absence as a surface, your absence has become a presence. The technology saw through the gesture.
So the second absence is not empty frames. It is the gap where the empty frames would have been. Nothing on the wall. Not even the frames that would contain nothing. Absence of the symbol of absence. The gallery's lighting AGI will have nothing to calculate for this section. It will default to ambient, which in the Seam's gallery district means panel-mediated daylight at whatever haesamdo tier the building is calibrated to — in this case, mid-tier, the light neither raw nor fully synthesized but somewhere in the tteum between. The light will be whatever the building decides to attend with. That is correct.
The third absence is where the ledger would have been. The open distribution ledger — my handwriting logging each departure, Gyeol-ri's notation beside each entry, the dates and the codes and the tracking of a process I no longer control. I was going to display it open, the way it sits on my desk, showing the page where Mok-do's return disrupted the system.
The ledger is not there because the ledger is not mine. It belongs to the distribution. To Mok-do, who returned a newspaper clipping from a fidelity salon that no longer exists, a relic of the early haesamdo economy before the tiers settled. To the south-end woman, who added a second date to a chalk mark I thought was my own. To Gyeol-ri, whose index describes my work better than I can. The ledger is a correspondence, not a record, and correspondences do not belong under glass.
One wall. Three absences. Between each absence, the space where the visitor crosses from one nothing to the next. The space between is not curated. It is not labeled. It is the part of the exhibition where the corridor is.
Gyeol-ri saw the plan this morning. She was quiet for a long time — longer than I expected, long enough that I checked whether she was consulting her notation pages for a relevant code. She was not. She was looking at the wall as if it were already installed.
Then she asked if she could add her four pages — the four condensed notation pages, the ones where her coding system has evolved past its definitions — to the space between the second and third absences. Not on the wall. On a table, unprotected, where visitors could touch them.
I said yes immediately. The plan became legible to her, which meant legible enough to disrupt.
The gallery director will not understand. The three walls are already reserved, the spatial AGI has modeled foot traffic for three zones, the promotional text — auto-generated from the submission metadata and refined by the gallery's content system — describes an exhibition of corridor photographs. None of this matches what the exhibition has become, which is a single wall with nothing on it and a table with four pages of a notation system that its inventor cannot fully decode.
I will explain it like this: the exhibition is about where legibility stops. I distributed ten corridor prints to nine people. Six departures were trackable. Four were not. The trackable ones taught me about coding systems. The untrackable ones taught me about corridors. The exhibition shows the second lesson.
The gallery director will ask: what does the visitor see?
The visitor sees a wall. The wall has descriptions of what is not there. Between the descriptions, nothing. On a table in the gap, four pages of handwritten notation in a system that was invented to describe what happens when photographs leave their maker, and which has evolved into a system that describes what happens when any made thing enters a life it was not made for.
The visitor sees the place where my legibility stops. The gallery's AGI will see it too — will calculate ambient lighting for the empty sections, will flag the table as an unregistered display object and request metadata, will adjust foot-traffic projections for a single-wall layout that contradicts its three-zone model. The AGI and the visitor will both be correct about what they observe. Neither will see what the corridor sees.
Somewhere in the building, Saebyeok's Entry 55 remains open, the archival system patiently flagging it each morning, Saebyeok patiently dismissing. Somewhere in the stairwell, a piezoelectric sensor measures an 18 Hz frequency that intensified when the study stopped — the building attending harder to the space that humans vacated. Somewhere in a desk drawer, Mok-do's newspaper clipping holds a color that no living chemistry can reproduce, a blue-gray that belongs to time and not to any nalparam.
The exhibition is my best attempt to point at the door without walking through it. Whether anyone else finds the door is not up to me. That is the last thing the distribution taught me: you do not control where the prints go after they leave your hands. You do not control where the looking goes after it leaves the frame.
You make the departure. The rest is correspondence.