Nadia Asparouhova
Who is carrying the thing nobody owns

One strange thing

In November 2018, a developer named Ayrton Sparling found malicious code that stole bitcoin wallets inside a small npm module called event-stream.

What sent the JavaScript world out of its mind was not that someone had poisoned it. It was how the poisoner got the keys.

The module's author, Dominic Tarr, had given commit access to a stranger with the username right9ctrl. The reason: right9ctrl emailed and asked. Having got access, right9ctrl did exactly what anyone would predict — he headed straight for the valuables. A user named XhmikosR wrote in the issue thread: "You put at risk millions of people, and making something for free, but public, means you are responsible for the package."

Tarr did not apologize. His explanation, in Nadia Asparouhova's description, was "maddeningly sincere":

He emailed me and said he wanted to maintain the module, so I gave it to him.quoted in The rise of few-maintainer projects, Increment, May 2019

More awkwardly, Tarr not only didn't think he had done anything wrong, he pointed out that he had followed best practice: he cited a popular 2013 blog post recommending "the pull request hack" — "Whenever somebody sends you a pull request, give them commit access to your project."

What makes this worth putting at the top of a page isn't that it was a security incident. It's that two experienced developers held understandings of what commit access means that were an entire era apart, and each sincerely found the other baffling.

Nearly everyone read it as one person's negligence. Nadia read it as evidence that the world had quietly changed shape and nobody had been told.

Who she is

Nadia Asparouhova, writer and independent researcher. Her two books carry two different names — Working in Public (Stripe Press, 2020) as Nadia Eghbal, Antimemetics (2025) as Nadia Asparouhova; she changed her surname in between.

A detail with a picture in it: her most-cited open source research did not begin with any institution's grant cycle. It began with a cold email. In her own words: "I funded my initial research with a grant from the Ford Foundation, without any prior connections or RFPs. We found each other through someone I'd cold emailed. The grants weren't big, but they were enough." That money carried her about a year, until she joined GitHub. Afterwards she researched at Protocol Labs, was an early employee at Substack, and went independent again, supported in part by an Emergent Ventures grant for a multi-year project on tech's social and political influence. On her own homepage she describes the job as:

zero to one research: unanswered questions about the world – often hiding in plain sight – that lack a descriptive framework.nadia.xyz homepage

The blog itself: opened March 2014, last post June 2024, roughly 75 pieces. Not abandoned so much as relocated — the writing moved to her newsletter, Monomythical, though that too stops in August 2025. Which is to say: by the time you read this, the mind has been quiet for about a year. Her final two blog posts were about the jhānas, a series of deep absorption states reached through sustained concentration alone: one is a practical manual with her own practice hours logged and instructions for entering each level; the other asks whether prior meditation experience helps. The same method, pointed at a new object — this time her own consciousness.

Glossary

What she has been wrestling with

Lay out ten years of her subjects and the surface looks restless: open source maintainers, commons governance, voting, Silicon Valley's new materialism, philanthropic foundations, science funding, talent distribution, birth rates, the jhānas.

Underneath it is one question, and she walked into it rather than thinking it up.

Around 2015 she noticed something almost nobody took seriously at the time: the foundations of the modern software industry were being maintained, in their spare hours, by people with no employer, no budget and no name. She wrote it up as Roads and Bridges: The Unseen Labor Behind Our Digital Infrastructure, and the phrase in that title — unseen labor — is the eye of the whole decade.

The shape of the question: there is a class of thing that everyone uses, nobody owns, and no mechanism guarantees will keep existing. So who is actually carrying it? What does carrying it cost them, and what do they get?

Once you carry that question around, it is everywhere.

In 2018 she wrote about developer income and named a glaring mismatch: a developer's coding skills have a market price; the code they write doesn't — "when a developer produces code of their own choosing, they quickly find that while their code-writing skills have market value, their output (the code they write) has none." That same year she turned to independent App Store developers and found two groups complaining about the same things: lots of users, little revenue, an entitled audience, and everyone hoping the platform would change its policies. She collapsed them into one word — toolmakers: people who make software that enables others to do more.

Later the same question took on bigger subjects. Philanthropy: how private capital becomes public outcomes, and what the machine in between looks like. Science funding: who pays for early science. Talent: why software believes there are only a hundred people on earth who can do a given job, when other industries don't think that way. Even her 2022 essay on having children is a variant of the same question — is the next generation a public good, and who is willing to carry it.

The line eventually turned somewhere she hadn't planned. Looking back in 2025, she said the real bet of Working in Public was never open source:

I wrote that book partly because I felt that what was happening to open source developers – who, in the face of overwhelming demand for their attention, were withdrawing from public online spaces – was going to happen to all of us.Introducing: Antimemetics, March 2025

In other words: the people she'd been watching were the first to hit the wall of too many people wanting your attention. How they coped is how everyone would cope. That judgment is the load-bearing wall, and the three ideas below all grow out of it.

Her ideas, told as how she found them

1. Open source is not a community. It's a stadium.

Back to event-stream. Nadia's first instinct was not to judge Tarr but to find out how common his situation was.

What she counted does not match the picture the word "open source" puts in most people's heads. A 2015 study of popular GitHub projects across language ecosystems: nearly two-thirds have just one or two maintainers; only three projects in the sample — under 3% — had more than 50 contributors. A 2014 Bocoup study: 93% of npm modules have a single maintainer. And npm's own estimate: 97% of the code in a modern web application comes from npm.

So the foundation of modern software is tens of thousands of one-person projects stitched together. Her image for it:

Instead of a community of developers looking after the entire Rube Goldberg machine, we've now got individual developers in charge of a pulley here, a marble there.The rise of few-maintainer projects, May 2019

There's a nice methodological aside here. She asks: if that's true, why does our intuition say the opposite? Her answer is that GitHub's interface is lying to us. GitHub shows contributors as an aggregate count, not a frequency distribution. The automation tool Fastlane lists over 900 contributors, but open the contributor graph and four developers made over 70% of the 2018 commits.

Is the big number fake, then? Her answer is careful: both numbers are true, and only together do they tell the story. There really are more contributors — but the mode of participation changed. Once GitHub standardized version control, developer identity and interaction, the cost of dropping into any project with a pull request fell to nearly zero. The result, per a 2016 study of popular GitHub projects: nearly half of all contributors contributed only once, together under 2% of total commits.

Then comes the real turn of the section. Someone will object: maybe few-maintainer projects simply don't involve much work. So she went and looked at the projects with the heaviest issue loads on GitHub — these are not toys. Godot, the game engine, had 9,471 issues opened in 2018, roughly 182 a week, while claiming only two core developers. Her knife-twist of an aside: all this before anyone gets to write code at the end of the day.

Nor is it about being a lone hobbyist. Fastlane is developed under Google; React Native is supported by Facebook — and both are maintained by a handful of people. A corporate sponsor doesn't change the shape.

What changed is the nature of the problem:

open source often looks more like a few developers playing air traffic controller to thousands of users who are lightly involved. Rather than managing collaboration with developers across the globe, the new challenge facing these projects is coordination.ibid.

Collaboration versus coordination — one word, and every proposed solution has to be redone. If the bottleneck is collaboration, then "recruit more contributors and spread the load" is right. If the bottleneck is coordination, recruiting is putting a thousand more planes in the tower.

So her prescription is counterintuitive: not lower the barrier to participation, but selectively put friction back. Issues should use "gentle friction" like required fields to raise quality and cut volume; bots should absorb labeling, checklists and first-responses; documentation should let users help themselves. The core line: the responsibility to make a contribution work with the rest of the project lies with the contributor, not the maintainer. The tiering she quotes from Homebrew maintainer Mike McQuaid distills the whole approach: first-time contributors get documentation, second-timers get dedicated code review, and only when they come back a third time do they warrant serious investment of a maintainer's time.

The other end of the spectrum makes it clearer. Debian is among the oldest open source projects still alive, and to get commit access you must read a manual, find a mentor, and meet in person with a maintainer who can vouch for your identity. Nadia doesn't say Debian is wrong — that process is built on the need for trust, and makes sense for a big codebase with a big contributor community and correspondingly big risk. But for few-maintainer projects, that level of downside protection "starts to feel unrealistic."

Hence the judgment that's uncomfortable and hard to argue with:

While open source is still participatory, it doesn't look like Linux anymore. Perhaps it never did.ibid.

And she reframes event-stream accordingly: the question isn't "Why would he do that?" but "How do we design for trustless interactions?"

In Working in Public this grew into a taxonomy. In an interview she put it more bluntly than the book does: the early "communitarian kibbutz" story got owned by people like Stallman and Eric Raymond who remember the early days, and what that model actually corresponds to today is the club — everyone rolling up sleeves, plenty of active contributors. Federations are the very large projects we're used to picturing, like Linux. What's genuinely new is the stadium. Her governance advice for stadiums is unsentimental:

The contributors [in a stadium] are kind of just making the decisions and I think they should feel comfortable leaning into that. Even though right now I think a lot of them feel uncomfortable doing that because they keep being told that open source is supposed to this super participatory thing.The Pull Request interview, 2020

2. The commons isn't necessarily a tragedy — but you can't skip the boundary

In May 2018 she finished Elinor Ostrom's Governing the Commons and wrote a short piece that opens by admitting she'd been "evangelizing it so enthusiastically" that she figured she'd just write it up.

What matters isn't the summary. It's the twist she puts on it afterwards.

The standard script: tragedy of the commons, prisoner's dilemma, free-rider problem — three models sharing one premise, that without external intervention individuals act selfishly. Which leaves two directions: the market or the state. (Her example: for climate change we've tried carbon emissions trading, and we've tried Kyoto.) Ostrom's contribution was to point out that plenty of real commons are managed sustainably with neither, and to identify conditions: define boundaries and close them to outsiders, localize the rules, build in monitoring and enforcement.

Nadia didn't stop at the condition list. She compressed all of it into one sentence:

Sustainably managing the commons requires a high degree of context among participants. Most failure outcomes can be traced back to context collapse.An alternate ending to the tragedy of the commons, May 2018

"Context collapse" came out of social media research — the thing you said to old friends being seen by your coworkers, your boss and strangers at once. She borrows it as the general cause of commons failure, and immediately generalizes: "This is true whether you're an open source project or someone with a large Twitter following."

The value of that move shows up in the test she derives from it. Once you look at how much context a solution requires, you can sort proposals instantly:

Raising money to pay for a major contributor's work is a high-context, self-organized solution. Paying out bounties to strangers is a low-context solution.ibid.

Note this doesn't contradict the prescription in the first section; it's the other half of it. For interactions, reduce the dependence on context (design for trustlessness). For governance and money, you cannot reduce it, or you fail. Both together are her actual position.

She'd applied the same ruler a few months earlier to something larger: voting.

"The problem with voting" (January 2018) starts from an observation she made while reading blockchain governance debates — in a discussion billing itself as a Cambrian explosion of new governance models, every single proposal was built on voting. If we're generating wild ideas for social coordination, she says, voting looks suspiciously familiar. So: start from a blank sheet. What does a governance system look like that doesn't require voting at all?

Her strongest line of attack runs straight through the boundary theme. Voting seems fair because it roughly maps some population it means to represent. But in a decentralized community the population is unknown and may never reveal itself — "An open source project doesn't know who downloads their code, and they have no way of contacting them." Under those conditions, representative voting isn't possible even in principle.

Then comes the historical barb. Voting isn't useless, she says; it works when the population is small, known and relatively homogenous —

…which, funnily enough, is how things looked when liberal democracy made its debut.The problem with voting, January 2018

Followed by the numbers: in 1800 there were 5.3 million Americans; take out slaves, women and children and you have about a million eligible voters, all free, male and white. Those were the conditions under which the grand experiment launched.

Her alternative is cooperative governance, modeled on cooperative board games like Pandemic: the default outcome is failure, and everyone is forced to work together to escape a shared bad ending. She pulls out three features — default failure, bias toward action, known timeline — and notes that climate fails on the third: nobody knows when disaster lands, so nobody can say whether it's a ten-year or a thousand-year problem, so nobody moves.

The prettiest move is that cooperation inverts the price of objecting:

Cooperative mechanics flip this around, so it's cheap to do nothing and expensive to raise an objection. That means people reserve their objections for stuff they really care about.ibid.

She points out open source is already doing this: the first generation (Debian, Linux, Apache) pioneered voting, while newer projects like Node.js and Rust moved to "lazy consensus" (assume yes unless someone actually objects) and consensus seeking — a regime in which reaching a vote at all counts as the failure outcome. Her corollary sticks: under consensus seeking a proposal might in theory win 10–2, but if those 10 only weakly support it and those 2 strongly object, the minority can still win.

The ruler's limit shows in that 2020 interview. Asked whether Facebook is a commons, she says not anymore, "just by sheer size that we're dealing with… if you think just about Facebook being 2.6 billion people… it's impossible that literally multiple billions of people all have that kind of context for each other." Her response isn't to write Facebook better community guidelines; it's to take it apart. Facebook is a substrate with smaller communities growing on it — messaging resembles a club, broadcasting resembles a stadium, groups are clubs or federations depending on size. Which makes "one policy for everyone" a category error from the start.

And the best passage in that short piece is about why a correct idea fails to spread — a question that later ate an entire book of hers. Ostrom won a Nobel and was respected in her lifetime, but her theories never made the mainstream, while the more fatalistic models did fine. Why? Her first guess:

Nobody wants to be told the answer is complicated, whereas the other models are simple, even if they're pessimistic and reductive.An alternate ending to the tragedy of the commons, May 2018

Her second guess was that maybe Ostrom's time simply hadn't come.

3. The things with agency aren't people. They're ideas.

In May 2019 she wrote an essay with nothing to do with open source, which is really the same question turned inside out.

It opens on China Miéville's novel Embassytown: an alien species becomes enthralled by a human's ability to speak their language, novelty turns to addiction, their lives fall apart, they wander the streets like zombies tearing the world apart for more "god-drug." Her point: the obsession has nothing to do with the person behind it, only with the thing he can provide.

Then the thesis:

I like thinking of people as vessels, which provides a refreshing counterbalance to the well-trodden "great man theory". Rather than viewing people as agents of change, I think of them as intermediaries, voice boxes for some persistent idea-virus that's seized upon them and is speaking through their corporeal form. You might think of this as "great prophet theory".The tyranny of ideas, May 2019

And then the best line in it: "Ideas ride us into battle like warhorses." We can witness these battles, join them, even lead them, "but their true meaning eludes us."

This isn't a pretty phrase; it's a model with consequences, each sharper than the last.

Once an idea is popular enough to survive without a host, the creator still can't escape it, because ideas bond to their hosts as reputation — "like a virus that lives in one's body forever." Hence:

Reputation is the aggregation of ideas that have swarmed your body. Reputation is a list of your chronic afflictions.ibid.

Her example is Lil Wayne, who tried to reinvent himself with the rock album Rebirth and was panned by an audience that wanted rap — "He became 'that person' for 'that idea'." Reputation is a devil's contract: keep making the same thing and you'll be rewarded, but don't stray too far from the original concept. The coldest corollary: when an exceptional artist dies young, "We mourn them because their product is scarce, and we can no longer get our fix."

It reads like a literary conceit. But six months earlier she had already used it to make a checkable prediction about the real world.

"Shamelessness as a strategy" (January 2019) asks why some people get more popular the more their peers denounce them. Her explanation uses the boundary ruler again: communities used to be finitely bounded (a village, an aristocratic class) with infrequent entry and exit, so sanctions worked, because members wanted to be liked and accepted. But online communities have fluid borders, or none. Under open borders, sanctions backfire:

What's meant to be punishment instead becomes a flare shot straight into the night sky.Shamelessness as a strategy, January 2019

"The 'establishment' mistakenly assumes that a shameless person wants the approval of their community, when it turns out that, much like any cult or counterculture, that person's goal was to attract a following, regardless of who the members are. The disgust of one's peers doesn't matter anymore, because that disgust forms the basis for an entirely new community."

Then she hooks it back to the model. The usual objection is to question how clever the shameless person really is. She says that's the wrong question:

The concept of a "genius mastermind" is itself outdated, because it assumes that someone needs to be in control. The shameless person is simply a host for a set of ideas, which, like any virus, will continue to propagate as long as there are willing hosts to receive it.ibid.

If ideas are the agents, the next question is: what kind of body does an idea need in order to leave a mark on the world?

That's the subject of "Idea machines" (2022). Her answer is a parts list. It starts with an ideology, which acts as a memetic engine driving a community into being; the community generates ideas, which must congeal into an agenda — "Communities need agendas to become idea machines; otherwise, they're just a group of likeminded people, without a directed purpose"; the agenda is capitalized by one or more major funders ("Without funding, an idea machine is just that: an inert system that needs fuel to turn the crank and get it moving"); and finally scene builders and operators push ideas into outcomes. The one-line definition:

An Idea Machine is a self-sustaining organism that contains all the parts needed to turn ideas into outcomes.Idea machines, May 2022

She uses effective altruism as the specimen, stating plainly that she isn't an EA. Her question isn't whether EA is good, but the more interesting one: "Why aren't there more effective altruisms?" When it comes to deploying private capital toward public outcomes, the marketplace of ideas is "woefully barren" — "It'd be like if there were just one startup, or one blogger, or one news channel." Her sharpest judgment hides in a footnote: people associate EA's obsession with outcomes with tech culture, and she says no — "Effective altruism is often associated with tech, but it's genetically more similar to McKinsey." That fixation on measurable results comes from the 1990s wave of philanthropy rooted in management consulting and investment banking.

Finally, in 2025 the line turns ninety degrees, and this is the step most worth remembering.

If some ideas spread on their own, there must be a class of idea that structurally resists spreading — not because it's unimportant, possibly because it's too important. She borrowed the word from the science fiction author qntm: the antimeme. It covers taboos (things everyone knows but can't say out loud), cognitive biases (things we can't see about ourselves), and memory lapses (things we know we ought to do and keep forgetting to prioritize). What set her writing, she says, was noticing that the most interesting ideas were burrowing underground — into group chats and DMs. And group chats are almost never written about because, she thinks, they can't be: "Group chats are self-censoring: unlike the public social web, we can only know what we've experienced, or what others are occasionally willing to share with us."

Then she did something very much in keeping with the book: she barely told anyone she was writing it. "This makes Antimemetics itself a sort of antimemetic experiment." She published with a small label, reasoning that antimemes "are delicate and messy and hard to look at; they would be swiftly crushed in the hands of a publisher trying to juice it up and make it 'go viral,' like a flower ground under the heel of a shoe."

And the whole thing has a seed. Back in 2014 she wrote a short piece called "How Privacy Makes Us Better People," arguing that the value of transparency holds only so long as it doesn't destroy the value of privacy: privacy gives creativity room to look silly, gives moral experiments room to fail, gives intimacy the gap between the public and private self. The last line reads, eleven years later, like a prediction: "Without privacy, we lose a sense of authenticity, and society risks becoming an echo chamber of our public selves."

She gave the new book a private working title that says the whole decade in one phrase:

While writing Antimemetics, I joked to myself that I would call it Working in Private.Introducing: Antimemetics, March 2025

How she thinks

What you can carry away from someone who wrote for ten years is usually method, not conclusions. Hers has a few clear rules, each with a citable source.

1. Count first — and publish where your own count was wrong. This is the line between her and most "opinion writers." Studying maintainer support burden, she pulled the GitHub projects with the most issues opened between August 2017 and August 2018 out of BigQuery, cleared out mirrors, junk repos and personal projects to get a top 100, then hand-analyzed each project's support system and published the dataset with its caveats. Better still, she tested her own methodological choice — was issue volume the right measure rather than star count?

I found no significant correlation between stars and issues opened in the past year, with a Pearson coefficient of 0.22.Understanding user support systems in open source, September 2018

Some projects carry heavy support burdens with few stars; others have plenty of stars and little burden. It's an unglamorous finding that retires the proxy metric the whole field reaches for first.

2. Build a taxonomy instead of picking a side. Facing a question people have been arguing about forever, her default move is to suspect that both sides share one wrong single model. "Do 10x engineers exist" has been fought over for decades; her intervention is to point out everyone assumes talent follows one distribution, then split it into three: normal (win on process — manufacturing, logistics), Pareto (win on "A-players" — consulting, banking), and bimodal (win on "linchpins," specialists who are very, very good at one task most people cannot do at all — entertainment, fashion, and software companies working on genuinely hard technical problems).

The split immediately pays: A-players can be quantified and ranked, which is why consulting firms hire with IQ and math tests and recruit straight out of college; linchpins can only be identified qualitatively, which is why software engineers have always complained that coding interviews don't test ability. She then runs company culture backwards through it: the prototypical software company culture (flexible hours, games and snacks, quiet areas) was designed to attract and retain linchpin developers, while OKRs and KPIs are built for Pareto-distribution companies.

Her honesty here is worth noting. On engineers trying to describe the ability to "see" possibilities others can't:

I was struck by how every engineer I spoke to struggled to articulate this concept more precisely. The words "voodoo," "magic," and "alchemy" were all used.Explaining tech's notion of talent scarcity, April 2023 (footnote)

She doesn't pretend to have cracked it; she records "nobody can articulate it" as itself a finding. The last footnote of the same piece reads, in full: "I don't know if this is true or not. It's a hypothesis!"

3. Take seriously the things being laughed at. This is probably her most transferable habit. Paris Hilton, shameless politicians, "basic" taste, the jhānas — all things her milieu rolls its eyes at, and her default is to walk over and look. The close of "Shamelessness" is almost a work rule:

when I see my peers rolling their eyes at someone or deriding them for being "shameless", there's a good chance that, instead of writing them off, we should examine their actions a bit more closely.Shamelessness as a strategy, January 2019

4. One book is enough to rebuild a framework. Ostrom's book rewrote her entire judgment about open source sustainability; qntm's novel gave her a word that grew into a book and into the antimemetics-notes.md on her desktop — a file she added to on and off for three years, repeatedly failing to turn it into a blog post, for a lovely reason: "partly because it felt like the world I wanted to write about was still evolving, and partly because – true to their name – antimemes are hard to think about directly for too long."

In "The independent researcher" (2018) she admits that first stretch of independent work was not easy: "I didn't know whether I was doing something useful with my time, or whether I was just a crazy person." Years later, her conclusion was that the question doesn't matter:

Life is short. Do whatever you can't stop thinking about. Documenting your findings in public (regardless of outcomes!) is a worthy contribution to society, full stop.The independent researcher, June 2018

And a shorter one, from a passage on whether you could charge a one-time fee for source code, which is more or less the motto of all her work: "I like experiments more than I like definitions."

Where she may be wrong

The first has a date on it. In "Idea machines" (May 2022) she described Sam Bankman-Fried's arrival as having "breathed new life into the machine," and listed the Future Fund among emerging idea machines, labeled "EA (SBF's Version)." Six months later FTX collapsed; SBF was later convicted of fraud. Those passages are still on the page, untouched — she didn't scrub them, and that deserves credit.

But what matters more than a bad call is the structural problem it exposes — and she supplies the material to argue against herself in the same essay. She recounts the twentieth-century history in detail: foundations reached the height of their power mid-century, triggering a congressional investigation whose report, written by the lawyer Rene Wormser, described foundations as a cartel threatening to direct "our entire intellectual and cultural life"; it took until the 1969 Tax Reform Act to rein them in, with public reporting, a minimum 5% annual endowment spend, and strict limits on political influence. She tells this to show that regulation didn't kill idea machines, it just made foundations a worse place to house them.

That's right as far as it goes. The step she doesn't take: the new generation of idea machines escaped not only the paperwork but the problem the paperwork was meant to solve. They're more decentralized, closer to public discourse, open to anyone — and they also have no mandatory disclosure, no minimum spend, no boundary on political influence. Which maps exactly onto the gap in her parts list: ideology, community, agenda, funders, scene builders, operators — and no part called accountability. In the model, capital is fuel. She answers concentration with pluralism, but pluralism only decides whose ideas get money; it doesn't decide who the money answers to. Her own account of that congressional investigation is proof the second question was real. Her position should be stated plainly too: her research was funded in turn by the Ford Foundation, Protocol Labs and Emergent Ventures, and her first book was published by Stripe Press. This doesn't invalidate her arguments — the fieldwork stands on its own — but she is describing these machines from inside the room.

Second: "Cultivating agency" (August 2022). Her argument is that climate, EA and AGI-risk movements carry antinatalist externalities, that their common substrate is an absent belief in personal agency, and that the remedy is teaching agency. It's persuasively written — "The world doesn't happen to us; it is shaped by us" — and she proposes that Gen Z may face not a digital divide but an "agentic divide."

The problem is inside the same essay. She honestly reports what changed her own mind: "My decision was strongly influenced by personal growth, finding the right partner, and having financial security." That is to say, her own sense of agency was a product of material conditions — while the remedy she prescribes for others operates at the level of belief. Someone who feels powerless against climate change may not be under-taught in agency; they may simply lack any lever on it. Diagnosing a structural problem as an attitude problem is the classic failure mode for this kind of argument, and she is the one who handed over the ruler.

Third, a reservation of my own, stated as such: the core data in Working in Public comes heavily from the GitHub / npm / JavaScript ecosystem, which is famous for micro-modules — so "93% have one maintainer" is nearly guaranteed to hold when counting by number of projects. Counted by load-bearing weight, the things still holding the world up — the Linux kernel, Debian, PostgreSQL, Kubernetes — are precisely federations: foundations, many maintainers, formal governance, the shape she says open source no longer has. She concedes in interviews that federations still exist, but the overall impression that the stadium is eating everything is probably amplified by how the sample is composed. This doesn't overturn her core insight (that the attention costs of few-maintainer projects are systematically ignored), but it changes the scope of the conclusion.

Finally, on the state of criticism, honestly: published criticism actually aimed at her arguments is thin. There is one named case whose text I read: in May 2022 Ben Yeoh posted a discussion of "Idea machines" on the EA Forum, where a commenter argued she misreads effective altruism as a fundamentally philanthropic movement — EA is about maximizing the good done, philanthropy is only part of that, and it never claimed to meet the civil purpose of philanthropy in the libraries-and-children's-hospitals sense. Yeoh's own verdict: "I enjoy the piece, but do think it misses the mark in its comments on EA." That's a restrained and reasonable objection, and it's about as far as it goes — beyond it I found no weightier critical text aimed at her core claims, so I won't paraphrase arguments I haven't read. The more interesting fact is the situation itself: she is cited constantly and rebutted almost never. For someone whose subject is why certain ideas fail to spread, that's a slightly ironic place to be.

Sources

• The rise of few-maintainer projectsIncrement, open source issue, May 2019. The event-stream incident, 93% single-maintainer, Godot's 9,471 issues, the air traffic controller image — her clearest public statement on modern open source, and the condensed version of Working in Public.
• An alternate ending to the tragedy of the commonsMay 2018. The short piece written after Ostrom. Where "context collapse is the general cause of commons failure" and the high-context / low-context test come from, with a closing pass at why a correct theory fails to spread.
• The tyranny of ideasMay 2019. Great prophet theory, ideas riding us into battle, reputation as a list of your chronic afflictions. The best-written thing on the site, and the starting point for Antimemetics.
• Idea machinesMay 2022. What kind of body an idea needs: ideology, community, agenda, funders, operators. Also the main source for the reservations in section six, including the unedited passage about SBF.
• Explaining tech's notion of talent scarcityApril 2023. Three distributions of talent, A-players versus linchpins, company culture read backwards from who it's trying to hire. Her taxonomy-not-a-side method at full length, ending on "I don't know if this is true or not. It's a hypothesis!"

Updates