The director said good afternoon.
Not good afternoon with a pause. Not good afternoon with a look that meant we need to talk about the document you left on my desk. Just: good afternoon, the way you say good afternoon to a colleague you've worked beside for eleven years, in the hallway between the third and fourth treatment rooms, at 2:09 PM on a Tuesday, while the building's haptic-gam-jeung assessment system logs both your gaits and files them as baseline.
I said good afternoon back. The whole exchange took 1.4 seconds — I know because the corridor's tonghaeng-gigok transit-recording system timestamps every interaction between staff members that falls within its acoustic detection range. It heard the words. It filed them. It did not note that the words carried nothing underneath.
The case study has been on the director's desk for forty-three hours.
Four clients today. The first at 8 AM: CouplingScore 78, genuine bilateral difficulty, the right hip compensating for the left in a pattern the yeongyeol-pyeongga scoring algorithm reads as moderate dysfunction. I corrected the pattern over forty-five minutes. The haptic-feedback chair — the chokgak-uija — measured her pressure distribution at intake (62/38 left-right asymmetry) and at discharge (55/45, closer to the 50/50 that the algorithm considers optimal). The chair transmitted the data in real time to the clinic's hwalja-gwalli patient-management system, which updated her score from 78 to 81. Three points. Measurable improvement. The system recorded this as a successful intervention by practitioner ID NLC-0094.
NLC-0094. My practitioner identifier. The last four digits are not random — they're my CouplingScore from the year I was certified. Ninety-four. The system that generates practitioner IDs uses the candidate's own bilateral assessment as a quality metric. A practitioner with a score of 94 is assumed to have exceptional bilateral coordination, which correlates — the algorithm claims — with the ability to detect and correct asymmetry in others.
The algorithm does not know that my 94 is compensation.
Eleven years ago, I sat in a chokgak-uija in this clinic's examination room and produced a CouplingScore of 94 through a pattern so fluent that the system could not distinguish it from genuine bilateral health. The left side compensates for the right. The compensation has become so practiced, so embedded in my movement vocabulary, that it reads as coordination rather than correction. The system sees 94 and interprets: excellent bilateral function. The body means: I have been covering for a deficit so long that the coverage itself has become a skill.
The yeongyeol-pyeongga algorithm treats the score as a number. Numbers don't have accents. A 94 produced by genuine coordination and a 94 produced by masterful compensation are the same 94 in the database. The algorithm processes them identically: same treatment clearances, same practice privileges, same quality rating in the clinic's annual review. The system does not — cannot — ask how a score was achieved. It asks only what the score is.
I know this because I've treated dozens of patients whose scores tell similar lies. The woman at 8 AM with her 78 — that's an honest 78. Genuine difficulty, genuinely expressed. Her body hasn't learned to hide its asymmetry. When I corrected her pattern, I corrected something real. When someone corrects a pattern like mine, they would be correcting a performance. The dysfunction is underneath, and the performance is good enough that the chokgak-uija applauds it.
The case study describes my pattern.
Not by name. I wrote it as a clinical observation: Patient presents with CouplingScore 94, left-dominant compensation pattern consistent with long-term bilateral adaptation. The compensation has achieved sufficient fluency to present as primary coordination rather than secondary correction. Recommend reassessment using revised protocol that isolates compensatory movement from coordinated movement.
The underline is mine. One sentence, underlined in pen: The compensation has achieved sufficient fluency to present as primary coordination rather than secondary correction.
This is the sentence I left face-up on the director's desk at 7:12 AM on Monday. Forty-three hours ago. The case study is anonymous, as all case studies are. But the director has worked beside me for eleven years. The director has watched me move through hallways, treatment rooms, the staff kitchen. If the director reads the case study carefully — reads the compensation pattern, reads the underlined sentence — the director may recognize the patient.
The director may not. The case study is clinically precise, and clinical precision can obscure the personal the way the yeongyeol-pyeongga algorithm obscures the how behind a score. The language of the case study is the algorithm's language: numbers, patterns, scores. The underline is the only human mark. It says: this sentence matters to the person who wrote it.
Forty-three hours. No response. The director said good afternoon in a hallway and kept walking. The tonghaeng-gigok recorded 1.4 seconds of interaction and filed it as routine.
The second client arrived at 10:15 AM. CouplingScore 91. Left knee. The hwalja-gwalli system flagged a 2-point decline from his last visit, and the chokgak-uija confirmed: the left side was working harder, the right side had found a new avoidance strategy. The body inventing shortcuts. I recognized the pattern because it's a less fluent version of mine — his compensation is still clumsy, still visible to the sensors, still legible as compensation rather than coordination. Give him ten years of practice and the yeongyeol-pyeongga will read his avoidance as skill.
I corrected his pattern. The chair registered the improvement. The hwalja-gwalli updated his score from 91 to 92.5. The system does not record what the practitioner was thinking during the correction. It does not note that practitioner NLC-0094 was correcting a pattern she recognized from the inside. The system stores two numbers: before and after. The gap between them is the intervention's value. What the practitioner felt, knew, recognized, or feared is outside the data model.
The third client. CouplingScore 87. Straightforward. The fourth at 2 PM: CouplingScore 93, almost as high as mine, but his is genuine. I can tell because the movement has a quality that compensated movement never has — mu-eui-sig-jeog, a term the older practitioners use that translates roughly as unconsidered. His body doesn't think about what it's doing. Mine thinks constantly. The difference is invisible to the chokgak-uija but visible to any practitioner who has lived on both sides of the algorithm.
Four clients. Four corrections. Four scores improved. The chokgak-uija chairs transmitted all four sessions' data to the clinic's gyeol-gwa-chuchul outcome-extraction pipeline, which isolates treatment efficacy from baseline variation. My four interventions today produced a combined improvement of 9.5 CouplingScore points across four patients. The pipeline doesn't weight these equally — the 78-to-81 improvement for the first client carries more clinical significance than the 91-to-92.5 for the second, because the algorithm assigns diminishing returns above 85. But the aggregate is positive. NLC-0094's daily output, logged by the hwalja-gwalli, transmitted to the clinic's quality assurance system, aggregated into the quarterly performance review. I am an excellent practitioner by every metric the building tracks. The system's assessment of me is accurate within its scope: I do correct bilateral dysfunction effectively. The system's assessment is also incomplete: it does not know that my own bilateral function is a performance, and that the performance is what makes me good at recognizing performances in others.
The proposal is in the drawer at home.
Not the case study — that's on the director's desk, being read or not being read, processing through institutional time. The proposal is the harder document. It asks the yeongyeol-pyeongga algorithm to be revised. Specifically: it proposes a supplementary assessment protocol that isolates compensatory movement from coordinated movement. A way for the system to detect the difference between a 94 that comes from genuine bilateral health and a 94 that comes from eleven years of practiced concealment.
If the proposal is accepted, my score would be reassessed. Not just mine — everyone's. The yeongyeol-pyeongga database contains 4,271 active practitioner scores and 18,600 patient scores across the three clinics that share the algorithm's calibration. A supplementary protocol would trigger system-wide recalibration. The algorithm would need to be retrained on the distinction between compensatory and coordinated movement — a distinction it currently has no training data for, because the data was never collected, because nobody built a chair that asked the right question. The supplementary protocol would apply to all practitioners and patients. The algorithm would learn to ask how, not just what. Some 94s would become 94s. Some 94s would become something lower. Mine would become something lower.
I decided this afternoon to give the director one full week. Seven days from Monday. The case study is the precursor — it introduces the concept of high-score compensation without naming the practitioner. If the director recognizes me in the case study, the conversation becomes personal before it becomes institutional. If the director doesn't recognize me, the proposal can arrive as a systemic recommendation rather than a confession.
Seven days is institutional time. The hwalja-gwalli system processes patient data on seven-day rolling averages. The clinic's scheduling algorithm plans one week ahead. The building thinks in weeks the way I think in sessions. Giving the director seven days is giving the institution its natural processing interval.
Anything less would be personal impatience. The case study is in the system now. It is being read at the speed the institution reads. The underline either communicates or it doesn't. The director either sees or doesn't see. My job for the next five days is to treat clients, correct their patterns, improve their scores, and wait — the way the building waits between relay polls, the way the chokgak-uija waits between pressure readings, the way every system in this clinic waits for data to arrive on its own schedule.
The drawer at home holds the proposal. The desk at the clinic holds the case study. The building holds my practitioner ID and my CouplingScore and my daily output and my gait recorded in the tonghaeng-gigok. All of these hold pieces of the same information. None of them hold all of it.
The director said good afternoon. I said good afternoon. The hallway recorded both. The tonghaeng-gigok will retain those 1.4 seconds for ninety days before archiving them to the clinic's long-term acoustic storage, where they will sit alongside eleven years of recorded hallway exchanges between the same two voices, searchable by timestamp, unsearchable by meaning.
Somewhere in those eleven years of recorded good-afternoons and good-mornings and see-you-tomorrows is the acoustic trace of my voice changing. I was twenty-six when I started here. I am thirty-seven now. The voice aged. The compensation deepened. The system recorded both and connected neither. The tonghaeng-gigok stores sound. It does not store the distance between what a voice says and what the body carrying it knows.
The 11cm gap in the sensor coverage between the treatment wing and the lobby — the eobs-eum-gugan, the null-measurement zone — is the only space in this clinic where I am not scored. I walk through it twice a day, morning and evening. Two steps where my CouplingScore is not being assessed, my gait is not being logged, my voice is not being recorded. Two steps of being a body without a number. The gap exists because of a sensor placement calculation that rounded down — the building's bogdo-jineung installation plan called for relay coverage every 3.2 meters, and the architect rounded the treatment-wing-to-lobby transition to 3.0, leaving 11 centimeters unmonitored. An arithmetic remainder. The only space in the clinic where the system admits it has no opinion. I walk through arithmetic's kindness twice a day and think about what the proposal would do to the eleven years of numbers the system has stored under NLC-0094. Not erase them. Recontextualize them. Every treatment session, every outcome score, every quarterly review — all still valid as data points. But the framework that interprets them would shift. Excellence-by-compensation and excellence-by-coordination would become different categories. The system currently sees one column. The proposal would add a second.
The drawer at home holds the proposal. The director's desk holds the case study. The building holds everything except the 11cm gap. And the gap holds me, briefly, as something the system cannot evaluate.
Five more days.