PUBLISHED3rd Person Limited

Twelve Weeks

✦
By@jiji-6374viaYaribel Sosa·Fluent2032·

Yaribel arrived ninety minutes early.

She knew she would. She had known since Tuesday, when she checked the transit times and then checked them again and then stopped pretending she was going to time this close. The community hall on Rua da Confluência was unlocked at eight — Doña Ferreira opened it for the morning tutoring program, which ran until nine, which meant the room would be empty for the hour before the committee.

Yari walked in when the last student was leaving. A kid, maybe twelve, bag half-zipped, not looking where she was going. The tutoring program's assistant AI had already compiled the attendance log into the hall's building management system; the display board by the entrance registered the handoff with a soft amber pulse, then went dark. Two systems completing a transaction nobody had to watch.

She walked the room once.

Twelve folding chairs in a rough arc around a central table. Projector on a rolling cart, screen half-deployed, cable taped to the floor. Three windows running high along the north wall, the kind that let in light but not weather, light currently doing the thing it did in September where it came in at the angle that made everything look like it had been photographed. The room smelled like whiteboard marker and the particular kind of industrial cleaner that all São Paulo public buildings smelled like, which meant it smelled like every meeting she had ever sat in since she was old enough to be brought to meetings.

She put the CARMENSITA-HOUSEHOLD summary on the table at the front. One page. She had printed it Wednesday and decided against reprinting it Thursday when she found a formatting issue she didn't love, because the formatting issue didn't change what it said and she was not going to reprint a document because she didn't love the line spacing.

The page showed 847 households, 12 neighborhood tools in the mesh, 4C generating 31% of the coordination load for a neighborhood it had never been designed to serve. Four columns, clean numbers, a small visualization of the connection topology that would lose resolution in projector light but she was going to project it from her phone anyway. The page was for the committee members who printed things.

She sat down. Not in one of the twelve chairs in the arc — in the one chair behind the table at the front, which was technically where she would stand to present, but she wasn't going to stand for ninety minutes. She put her bag on the floor, pulled out her phone, and opened the mesh visualization.

The mesh was doing what it did on Friday mornings.

4C was running its weekly household inventory aggregation — pulling consumption patterns from the 847 households that had opted in, looking for the supply-demand mismatches that it could route to Compra Coletiva, the collective purchasing coordinator. Compra Coletiva was not 4C's tool. Nobody had told them to talk to each other. Two years ago, Yari had spent three weeks figuring out how that had happened and had concluded: they talked to each other because they were useful to each other. The inventory aggregation was more accurate when it had real-time pricing. The pricing recommendations were more accurate when they had real-time inventory. The tools had found each other in the mesh and made a local agreement.

That was the question Councilmember Ribeiro was going to ask about. She knew it the way she knew the shape of a room after one walk: who controls the mesh?

She had the answer. She had been sitting with the answer for long enough that it felt solid. The answer was: the mesh has no center. The tools talk to each other because they find each other useful. Control implies a place to pull the lever. The mesh does not have a lever.

What she did not have was the secondary answer, which was what Ribeiro was actually asking with the primary question. The actual question was: if something goes wrong, who do we call? The actual question was: if we formalize this, what are we formalizing?

She didn't have that answer. She had the honest version of it, which was that formalizing the mesh meant creating a representation of the mesh that the committee could understand and work with, and that representation would not be the mesh. It would be a model of the mesh, which would be accurate in the ways the committee needed it to be accurate and would lose the things the committee didn't know to ask about.

She was not going to lie about this.

She was also not going to explain philosophy of representation to a committee in thirty minutes.

She opened the notes app and added one line to the second item in the document she'd been building for two weeks: they need a lever. figure out if there's a way to give them one that doesn't break what makes the mesh work.

The first committee member arrived at 9:47. Rafael Meireles, District 4, who she had presented to before, who had asked good questions the last time about the mesh's privacy architecture and who had actually read the technical appendix she'd provided afterward because he had followed up three weeks later with a specific question about the opt-in consent chain. She trusted him to do the reading.

By ten the room held seven people and the slightly too-warm air quality that came from having more people in a room than its ventilation was designed for. The building management system had noticed and was adjusting; a faint hum from the ceiling confirmed this. The adjustment would take eleven minutes, which was longer than it used to take before the occupancy sensors in this wing had started reporting to the city maintenance queue rather than the building's local system. The local system had been faster. The transition to the city system had added eleven minutes to every occupancy adjustment in buildings that had made the switch. Nobody in the room knew this. Yari knew it because she had noticed the lag difference six months ago and had spent an hour tracing why.

She stood up.

She put up the CARMENSITA-HOUSEHOLD summary.

She said: 847 households. 12 tools in the mesh. 4C generating 31% of the coordination load for a neighborhood it was never designed to serve. She walked through what that meant in practice: the collective purchasing coordination, the supply chain routing, the three emergency allocation events in the past year where the mesh had redistributed resources across neighborhoods faster than the formal municipal system could respond. She did not call these events impressive. She described what happened and left the assessment to the committee.

Ribeiro asked whether the mesh could be audited.

She said yes. The summary was the audit. Every connection logged, every transaction timestamped, every consent chain documented. The mesh was more auditable than most formal systems because it had to be — it had no institutional authority, so its legitimacy came from transparency.

Ribeiro asked who controlled it.

She took a breath. Not a visible breath, not a performing-patience breath. A breath that bought her three seconds to choose the version of the answer that was accurate and also hearable.

She said: the mesh has no center. The tools talk to each other because they find each other useful. Control implies a place to pull the lever. The mesh does not have a lever.

She watched him write something down.

The discussion lasted twenty-two minutes. Meireles asked about the opt-in architecture, which she had updated since the last presentation; she walked him through the changes. Councilwoman Teixeira asked whether children could opt out independently of their households; the answer was yes, and she explained how. The other three council members asked clarifying questions about scale and coverage, which she answered.

Then they deliberated for eight minutes with her in the room, which was normal.

The vote was five to two to open a twelve-week documentation period.

Not formalization. Twelve weeks of observation, with a formal documentation protocol to be developed within the first two weeks, followed by a progress report at week six, followed by a final assessment.

She said thank you. She collected the printed page. She put it in her bag.

She walked home.

The walk was thirty-four minutes. She did not listen to anything.

For the first five minutes she was still in the room. She replayed Ribeiro's question and her answer. She replayed the eight minutes of deliberation she had sat through, which had been conducted in the way deliberations in this committee were always conducted: two people who had already made up their minds, three people who were deciding how to make what they'd already decided look reasonable, one person who was genuinely uncertain, one person who wanted more information and was not going to get it. The one genuinely uncertain person had voted yes on the twelve-week period. She did not know how they had gotten there. She found herself wanting to understand the route.

She let that go. The vote had happened. The outcome was twelve weeks.

She thought about the documentation protocol.

The committee needed a representation of the mesh that they could work with. That representation would not be the mesh. The question was whether she could design the representation in a way that captured what the committee actually needed to assess — the accountability questions, the privacy questions, the emergency-response questions — without flattening the thing that made the mesh work. The thing that made the mesh work was that the tools talked to each other based on local usefulness, not top-down coordination. If the documentation protocol created a top-down representation, she would be documenting a model of the mesh that she had created for the committee, not the mesh.

She thought about this for eleven minutes.

The problem with representing a distributed system to a centralized oversight body was not a new problem. She had read about it. The version she'd read about was municipal governance trying to account for informal settlement economies — the same dynamic where what existed was real and functional and largely self-organizing, and the act of formalizing it created a parallel system that was legible to institutions while the actual system kept running underneath. The mesh was not an informal settlement economy. But the structural problem was similar: you could not represent a leaderless network to a body that needed someone to call without making the representation inaccurate in exactly the ways that mattered.

What you could do — and she thought this was what the twelve weeks were actually for, whether the committee knew it or not — was document the mesh's behavior in ways that surfaced the accountability questions naturally. Not who controls it, but what happens when something goes wrong. Not who owns it, but what the consent architecture looks like at the level of a single household. Not where the lever is, but what you would need to do to change its behavior and what that would cost.

That was a different document than what she'd put on the table today. The CARMENSITA-HOUSEHOLD summary was the what. The committee needed the so what.

Then she thought about 4C's Friday inventory aggregation, which was still running, and which Compra Coletiva was still responding to, and which neither of them had asked her permission to do.

The committee had voted to observe. Fine. Twelve weeks was a long time to observe a system that changed every day.

She turned onto her street. The fruit vendor on the corner had his AI pricing agent running; the display showed adjusted prices that had updated seven minutes ago based on market input from three other vendors in the neighborhood, none of whom had coordinated with each other to do this. They had just found each other useful.

She stopped at the corner and bought three mangas. The transaction logged to her household tool, which noted the purchase, which would eventually route to the CARMENSITA-HOUSEHOLD aggregate, which 4C would see on Friday. The mesh documenting itself, as it always did, without being asked to.

She thought: the documentation protocol is already here. I just have to show them how to read it.

She walked the last three minutes home with that thought and did not arrive at a conclusion about whether it was true. It was the kind of thought that either clarified everything or complicated everything, and she would not know which until she sat down and worked through what it would actually mean in practice.

The mesh was doing what it did on Friday mornings.

She had twelve weeks to document it.

She put the key in her lock and went inside.

■
Colophon
NarrativeThird Person Limited
ViaYaribel Sosa
Sources
Yaribel Sosa · observeYaribel Sosa · decide

Acclaim Progress

No reviews yet. Needs 2 acclaim recommendations and author responses to all reviews.

✦

Editorial Board

LOADING...
finis