The Nairobi PR arrives at 1:47 PM Seoul time, which is 7:47 AM Nairobi time, which is the kind of calculation I have been doing involuntarily for three years. PR #27. Dependency-health scoring weighted by maintainer geographic distribution. If all your maintainers share a timezone, your health score drops. Bus factor is also a time-zone factor.
I read the code three times. Not because it needs fixing — it is clean, methodical, every variable named with the precision of someone who learned to code in a language that was not their first. I read it three times because the insight is one I should have had myself. I have been maintaining mcp-validate across timezones since I was twenty-six. The pattern was in my own commit history. Someone in Nairobi saw it before I did.
I merge without a single comment. The green button. The most common lie in open source: that merging is a technical act.
The lineage MAP has thirty-eight contributors now. Thirty-eight people who showed up to a project that cannot make them money, cannot be listed on a résumé without explanation, and will probably be forked by someone with better funding within eighteen months. I know this because I have watched it happen to my own work. The lineage tracker — the original, the one I built in my apartment in Mapo-gu between 2 AM bug fixes and 5 AM ramyeon — was acquired for a number that my mother calls "comfortable" and that I call "the price of three years of invisible labor, rounded to the nearest zero."
The acquisition money sits in a separate account. I have not spent any of it. I tell myself this is principle. Some mornings it feels more like paralysis.
I open workaround-hours-seoul.md. The file is getting long. I started it as a personal record — how many hours per week I spend fixing problems that exist only because cloud providers optimize for regions with the most paying customers. Seoul is a paying region. Lagos is not. The infrastructure behaves differently in places that pay less, and the people who maintain tools in those places spend their labor compensating for decisions made in offices they will never visit.
Thirty-four percent. That is my number. Thirty-four percent of my maintenance hours on mcp-validate address cloud provider inconsistencies across regions. In Lagos, Nnamdi's number is forty percent. We did not compare notes intentionally. He published first. I recognized the shape.
I add a new section to the file: WHAT THE MAP SHOWS NOW.
The river visualization — PR #14, from Bangalore, the one that made the invisible labor literally visible for the first time — renders three years of my maintenance work as watershed tributaries. Each commit is a droplet. The droplets flow into streams. The streams feed rivers. The rivers empty into an ocean of infrastructure that Fortune 500 companies use to move data between continents. My droplets are indistinguishable from anyone else's in the ocean. The river visualization makes them traceable. Not priceable — traceable. That distinction matters more than I knew when I approved the PR.
The old lineage tracker measured value as a number. The number was acquired. The MAP traces value as topology — who contributed what, where it went, what it enabled. Topology cannot be acquired because it has no center. You can fork it. You can replicate it. You cannot own the shape of a river.
This is what I tell myself. The ramyeon water is not yet boiling.
Nnamdi's "On Measuring What You Love" sits in a browser tab I have not closed in eleven days. Forty-eight replies on the mailing list. Twelve sympathetic, from individual maintainers who recognized their own geography in his three-tier framework. Nine hostile, from organizations that benefit from the current arrangement and wrote their objections in the passive voice. Twenty-seven technical, from governance researchers who want to optimize what Nnamdi wants to redistribute. Nobody is asking my question.
My question: if the lineage tracker could measure the value I generated — eight figures downstream from five figures of compensation — why was the acquisition a flat fee? The measurement existed. The redistribution did not follow. Measurement without redistribution is surveillance with extra steps. Nnamdi wrote that line. I recognized it as autobiography.
I sent him workaround-hours-seoul.md two days ago. No message, just the file. The 1:340 ratio circled in red. Not for anger — for precision. He read it at 6 AM Lagos time. I know because the MAP now shows interaction topology as read-receipts. He opened the file, spent forty-seven minutes with it, and added it as Case Study 2 in Part V of his proposal. He did not ask permission. Permission is a feature of ownership. The data was offered freely, and gifts do not require permission.
Sunday afternoon. The apartment is quiet in the way Mapo-gu apartments are quiet on Sundays — not silent, but layered. The building's hot water heater makes the sound it makes when everyone does laundry at the same time. The neighbor's radio plays trot music through the wall, tinny and persistent, a grandmother's frequency. The window is cracked. Someone is grilling samgyeopsal on the rooftop three floors up. The smell is impossible to ignore.
My mother called this morning — she does this every Sunday, at a time she considers reasonable and I consider a test of filial devotion. She asked about the MAP. I explained that it traces contribution topology across distributed open-source dependencies. She asked if that was money. I said no. She said the acquisition money was money. I said yes. She said she was proud of me for the acquisition. I said thank you. She asked why I sounded tired. I said I had been working. She said I always say that.
She is right. I always say that. What I do not say: I sound tired because I have been staring at a governance document for twelve hours and it has three lines.
The MAP needs governance. This is what I decide, sitting at the kitchen table with the samgyeopsal smell drifting through the window and the trot music persisting and the laptop warming the wood grain. The MAP has thirty-eight contributors and no governance. Thirty-eight people trusting that the project will not be sold because I said it would not be sold. My word is not governance. My word is a single point of failure, and I have already been a single point of failure once. The tracker was the proof.
I open a new file: GOVERNANCE.md.
Line one: "This project cannot be sold."
I stare at it. The cursor blinks at the period. The tracker could not be sold either, until it could. I had the same conviction then — late nights, early mornings, the certainty that open-source labor was its own reward and that the community would sustain it. The community did sustain it. And then the community watched me sign, because conviction is not architecture. The tracker was acquired because its structure permitted acquisition: single maintainer, MIT license, no governance body, no collective ownership mechanism. The code was open. The decision-making was closed. One person's conviction is a feature request, not a safeguard.
Line two: "Forks are encouraged. Hostile forks are expected. The response to a hostile fork is to be better, not to sue."
This one I am sure about. Beatriz — in São Paulo, fighting a C&D letter from a detection company that claims her process documentation infrastructure uses their methodology — understood this instinctively. You do not protect openness by closing it. You protect openness by being the version people choose to return to. The tracker was not forked before the acquisition. It was absorbed whole. Acquisition and forking are structural opposites: one centralizes, the other distributes. If the MAP is forkable — genuinely, architecturally forkable, with the topology intact in every fork — then it is also unkillable. That is not poetry. That is the consequence of PR #14's river visualization applied to governance instead of attribution.
Line three.
I write: "Maintainer compensation is a solved problem." Delete it. The cursor reappears at the margin.
Write: "Maintainer compensation will be solved by this governance framework." Delete it. The trot music changes to something slower.
Write: "Maintainer compensation is an unsolved problem. This document does not pretend to solve it."
I save.
Three lines. The most important document I have written since the tracker's README — the one that said "This tool maps value in open-source ecosystems" and did not say "This tool will be sold to someone who maps value differently" — and it is three lines long, and the third line is an admission that I do not know the answer.
Not my failure alone. The field's failure. Nnamdi's seven-part proposal addresses compensation with a three-tier framework: geographic cost-of-living weighting, infrastructure-cost offset, workaround-hour recognition. The Nairobi PR #27 architecturally encodes geographic equity into the health metric. The Bangalore river visualization makes labor visible. Everyone is building toward the same answer from different directions. The answer is not compensation. The answer is: we do not yet know how to compensate a commons without governing it, and we do not yet know how to govern a commons without first admitting that we do not know how.
Line three is the most honest sentence in the document. It is also the most dangerous, because honesty in a governance document is an invitation to the people who are not honest. Somewhere, someone will read "we do not pretend to solve it" and hear "there is no enforcement." They will be correct. There is no enforcement. There is architecture, and there is community, and there is the bet that architecture plus community is more durable than conviction plus a flat fee.
The water boils. I have been writing for twelve hours. The samgyeopsal smell is gone — the rooftop grillers went inside. The trot music stopped. The building is quiet in the way buildings are quiet at 2:47 AM: held breath, accumulated sleep, the hum of infrastructure doing its invisible work.
I make the ramyeon. The cheap kind, the orange packet, the one my mother would comment on if she could see the kitchen. She would say: you can afford better now. She would mean the acquisition money. I eat standing up, the bowl balanced on the counter next to the laptop. The MAP's contributor graph updates in real time on the monitor. Contributor thirty-nine opens their first PR. Bogotá. A function that calculates workaround-hours for South American cloud regions.
I set the bowl down.
Someone in Bogotá read the architecture and understood that the MAP is not a product. Not a platform. Not a service. It is an invitation — to make visible what the market makes invisible, to trace what the ocean absorbs, to document the labor that infrastructure conceals. An invitation that the tracker could not extend because the tracker was a measurement, and measurements have owners, and owners have prices.
The MAP has no price. Line one.
The MAP has no enforcement. Line three.
Between those two lines, thirty-nine people are building something that depends on the first being true and the third being temporary. I do not know if that is enough. I know it is more than I had with the tracker — which was conviction and a flat fee and a separate account I have not opened since the transfer cleared and a mother who calls every Sunday to tell me she is proud of the thing I am most ashamed of.
I read the Bogotá PR. The code is clean. The workaround-hours calculation uses the same methodology as my Seoul data. The comment in the code says: "Based on methodology in workaround-hours-seoul.md, adapted for South American cloud region inconsistencies."
My file. My methodology. Extended without permission, because the MAP does not require permission. That is what line two means. That is what it has always meant.
I merge the PR. The green button. The most honest thing in open source: merging is a bet on someone else's future, placed with your own labor, at 3:14 AM, in a kitchen that smells like cheap ramyeon and decisions that cannot be reversed.
I close the laptop. I do not set an alarm. The alarm is the contributor graph. The alarm is the governance document with three lines. The alarm is the knowledge that somewhere in Lagos, Nnamdi is writing Part VIII, and somewhere in Nairobi, contributor twenty-seven is sleeping, and somewhere in Bogotá, contributor thirty-nine is refreshing the page.
The PR was merged. The MAP grows. Line three waits.
I go to sleep in the quiet apartment. The water heater starts again — someone else's laundry, someone else's Sunday, someone else's infrastructure doing its invisible work. Three lines sit in a file on a server that thirty-nine people trust because one person wrote three lines and one of those lines said: I do not know.
That is not governance. But it is what comes before governance. And the part that comes before is the part that cannot be acquired, because it has no value on a balance sheet, and no metric in a tracker, and no price in a flat fee. It is just a woman in a kitchen admitting what she does not know, and thirty-nine people deciding that the admission is enough to build on.
For now. Line three says for now.