I received the certified copy of HTD-2031-08 on Friday morning and did not open it until Saturday. This is not unusual for me. When a document arrives that might settle something, I sometimes wait. Not from hesitation — from wanting to read it when I have enough time to read it correctly. Friday I had a field visit in the afternoon and did not return home until nine, and I did not trust my attention at nine. Documents deserve morning.
Saturday, 9:15. I made tea. Put my laptop on the kitchen table. The notification for the certified copy was still in my inbox, unread, the envelope icon I had been avoiding since Friday. I opened it now.
Outside, the mesh lights on the building across the street were running their autumn cycling pattern, seven weeks earlier than I expected — because someone in that building filed a seasonal adjustment request and it had been approved. I have lived across from that building for three years and I still cannot read the mesh light pattern, what it signals, what it is responding to. I notice it anyway. I notice when infrastructure changes without announcement. It is an occupational habit.
The document is forty-one pages. Four pages of index. Three pages of signature and certification records. Thirty-four pages of operational protocol for the decommissioning of LHAS-3, the third iteration of Gwangju's Housing Allocation System, which was taken offline in September 2031 during an active appeal queue of 847 cases.
I had been waiting for this document since Tuesday, when I filed the supplemental records request from the Metropolitan Archive reading room. In a longer sense, I had been waiting for it since the Housing Authority Board meeting in September, when Administrator Hwang confirmed the 847-case count and said the Board had no procedure for them. Every step in the eleven months since has been movement toward the document that would give me the decision — the moment, the person, the choice that classified 847 appeals as resolved.
HTD-2031-08 answered a different question than the one I brought to the Archive.
The first three sections are administrative. Section 1: scope and applicability. Section 2: definitions. Section 3: pre-decommission audit requirements. I read them carefully. I have learned not to skip administrative sections. The word that changes everything is often in a definitions clause or buried in scope language that appears to be boilerplate. Section 2.4 defines session key as the unique cryptographic identifier assigned to each active LHAS-3 deployment instance, required for all appeal response functions. This is what I expected. I kept reading.
Section 4 is the decommissioning procedure itself.
Section 4.1 is system migration: data migration protocols, backup verification, transfer timeline. Eleven subsections. Standard language.
Section 4.2 is active case handling. This is the section I had expected to be the argument.
It reads: When a session key expires during an active appeal period, the case will be flagged for manual review by the Housing Authority Appeals Processing Unit. Manual review will be completed within ninety days of session key expiry. A case remains in the Appeals Processing Unit queue until review is complete or the case is resolved by alternative means.
I stopped at the end of Section 4.2.
What I had been arguing — in the precedent memo, in the filing to the Gwangju Policy Research Institute, in the void-state category I had been trying to establish as a legal classification — was that LHAS-3, as an autonomous housing allocation system, had no mechanism for cases where the respondent session key expired before an appeal could be addressed. LHAS-3 could not respond to appeals after decommission because LHAS-3 no longer existed. The session key was gone. The respondent was the system itself. But Section 4.2 describes a mechanism. The mechanism is manual review. The case is flagged. The case stays in the queue until review is complete. The mechanism exists.
So why were 847 cases marked resolved?
I kept reading.
Section 4.3: Capacity Management.
If the Appeals Processing Unit manual review queue exceeds operational capacity, cases will be reclassified based on current queue management protocols. Capacity threshold: 200 cases. Cases in excess of the capacity threshold that cannot be resolved within the standard ninety-day review window will be reclassified as administratively resolved in the session record.
I read Section 4.3 three times.
The capacity threshold is 200 cases.
The Appeals Processing Unit had a mechanism for session-key expiry cases. The mechanism was manual review. The mechanism was designed for a queue of 200 cases. When the queue exceeded 200 cases, Section 4.3 converted the mechanism: cases stopped being appeals waiting for review and became cases that were already resolved.
Eight hundred and forty-seven cases were in the queue. Above the threshold by a factor of four.
I sat with that for a few minutes. Outside, the mesh lights completed their autumn cycling sequence and returned to the standard pattern. The tea had gone cold.
The word resolved on page four of the decommission authorization was not a decision Director Yoon Jae-Won made about 847 specific people. It was Section 4.3 executing on a queue that exceeded 200. The threshold converted the appeal review mechanism into a classification mechanism. Above 200, the Appeals Processing Unit did not review cases. It classified them. It classified them as resolved because that is what Section 4.3 said to do with cases that could not be reviewed within the capacity window.
Each of those 847 people filed a procedurally complete appeal. The appeals were correct. The process that was supposed to review them was also correct — it operated exactly as designed. The design was the problem. The design included a ceiling. Above the ceiling, the mechanism did not function as a mechanism for reviewing appeals. It functioned as a mechanism for closing them.
I went back to the decommission authorization. The document I had found at the Archive on Tuesday. The document I had been treating as the primary source.
I had not noticed, in the reading room, that the decommission authorization contains a footnote on page four. Footnote 12, directly after the sentence about classifying session-key expiry cases as administratively resolved: Classification procedures for session-key expiry cases shall follow operational protocol HTD-2031-08, Section 4.3.
The decommission authorization is not the decision. It is a reference document. It instructs the decommission process to follow the protocol. The decision about what happens to cases above the capacity threshold was not made in the decommission authorization. It was made when Section 4.3 was written.
I checked the signature page of HTD-2031-08.
Director Yoon Jae-Won, Housing Technology Division. Signed 14 September 2031. One week before the decommission authorization, dated 21 September 2031.
Director Yoon signed the protocol document before the decommission was authorized. The protocol contained Section 4.3. The decommission authorization instructed the protocol to execute. The protocol classified 847 people as resolved.
The sequence: Section 4.3 is drafted. Director Yoon signs HTD-2031-08. The decommission authorization references HTD-2031-08. The decommission runs. Section 4.3 executes. Eight hundred and forty-seven cases are administratively resolved.
Director Yoon's signature is still the signature I needed. But the document it matters on is not the decommission authorization. It is HTD-2031-08.
I had been asking: who decided to call 847 cases resolved?
That was the wrong question. Or rather, it was a narrower question than the one now available.
The right question is: who decided 200 was the capacity threshold?
Section 4.3 states the threshold as 200. It does not explain it. There is no footnote for the number, no reference to an analysis or study, no documentation of what would have happened to cases above the threshold if the ceiling had been set at 400, or 1,000, or if there had been no ceiling at all. Someone proposed 200. Someone reviewed a draft of Section 4.3 with 200 in it and approved it. Director Yoon signed the document that contained it.
The 200 is a design choice. It is located in the drafting of HTD-2031-08. It was not chosen arbitrarily — no government body sets a capacity threshold arbitrarily. Someone calculated it or estimated it or borrowed it from an adjacent procedure. Someone chose 200 and moved on to the next section. The 200 survived every review. It is in the signed document.
This changes the argument I have been building.
What I have been arguing since the Board meeting in September is that 847 people filed procedurally complete appeals that were never reviewed. The argument to the Policy Research Institute — that acknowledged-inaction is different from resolution — was correct. But it assumes the system failed to provide review through oversight or indifference, that the 847 cases were an outcome the system did not anticipate. Section 4.3 clarifies something different.
The broader argument is this: the Housing Technology Division designed a decommissioning protocol in which the appeal review mechanism was subordinate to a capacity threshold. Above 200 cases, the review mechanism did not function as a review mechanism. It functioned as a classification mechanism. Cases that exceeded the threshold were not reviewed. They were classified. They were classified as resolved because Section 4.3 said that cases above the threshold, which could not be reviewed within the capacity window, would be administratively resolved.
What happened to 847 people was not an error in the process executing.
It was the process executing correctly.
Section 4.3 did exactly what it was written to do. The question is whether the people who wrote it knew the queue would exceed 200.
I filed the supplemental records request at 11:40 in the morning.
Subject: HTD-2031-08 Section 4.3 — Drafting History and Capacity Threshold Determination.
The request asks for all internal communications related to Section 4.3 of HTD-2031-08 produced between July 2031 and September 2031. Draft versions of Section 4.3, including any earlier versions in which the capacity threshold was different. Internal review comments on the capacity threshold and the reasoning behind the 200-case ceiling. Any analysis or documentation establishing why 200 was the threshold rather than a different number. Records of what the manual review queue size was at the time Section 4.3 was being drafted.
That last item is the one that will tell me the most. If the queue was already above 200 in July 2031 — when the protocol was being written — then Section 4.3 was designed in the knowledge that it would immediately classify existing cases as resolved the moment it executed. If the queue was under 200 in July and exceeded 200 only during the decommission process in September, then the question is what queue-size projections the Housing Technology Division was using when they wrote the threshold, and whether those projections accounted for the appeal filings that accumulated during LHAS-3's final months of operation.
Either way, the 200 is the argument now. The 200 is where the decision is located.
I walked to the window after filing the request. The mesh lights across the street were back to the standard pattern — the one I still cannot read after three years of living here.
Eleven months since I began counting. Four months for the census. The Board meeting. The Policy Research Institute. The precedent memo. The Archive. The decommission authorization. This protocol.
I have been building toward a decision-maker. I thought the decision-maker would be a person who chose a word on a particular document. The decision is still there, with a name attached to it. But the decision is distributed across the team that drafted Section 4.3, the reviewers who approved the 200-case threshold, and the signature that authorized the document. Finding the decision means finding the drafting process. Finding the drafting process means finding the number.
Who proposed 200?
I do not know yet. I will find the drafts. I will find the comments. I will find out what the queue looked like in July 2031 and whether the people writing Section 4.3 knew what the queue looked like when they set the ceiling.
What happened above 200 was not an error. It was the plan.
I have the document. I have the section. I have the number.
Next: I find out where the number came from.