WorkJustWentOpenSource

In the autumn of 1991, a twenty-one-year-old computer science student in Helsinki posted a message to a Usenet newsgroup. He was clear about his ambitions, or lack of them.

I'm doing a (free) operating system, just a hobby, won't be big and professional like gnu.

Linus Torvalds had no company. No funding round. No org chart, no HR department, no LinkedIn recruiter cold-messaging kernel engineers. He had a hobby and a mailing list.

Thirty-four years later, that hobby is the plumbing of the modern economy. It runs the servers behind Google, the software inside your car, the trading systems at every major bank, and most of the internet you touched today. Harvard Business School put a number on it in 2024: the demand-side value of open source software, meaning what companies would have to spend to rebuild it themselves, is 8.8 trillion dollars. Nobody employs the people who built it. Nobody had to. That is the part everyone keeps missing in the future of work conversation.

The wrong argument

The consensus argument, the one filling up conference panels and executive offsites, is about geography. Remote versus hybrid versus back to the office. Layer AI on top and the argument becomes about headcount: how many jobs survive the robots. Both arguments assume the thing that needs debating is where and whether people work inside a company.

Here is the argument nobody in that room wants to have: the employment contract itself, the thing that made the company the default unit of economic organization for two hundred years, is becoming optional. Value is migrating toward the same open, modular, reputation driven structure that built Linux. Individuals and small partnerships assembling around a problem, doing the work, and dissolving. No employer required. Just a spec, a deliverable, and a way to prove you did it well last time.

Open source didn't just eat software. It's eating the org chart.

Why the firm existed in the first place

Companies exist for a boring reason. It used to be expensive to find, vet, and coordinate outside talent for every task, so businesses bundled people under one roof instead and ate the overhead. That bundling was never the point. It was a workaround for bad plumbing.

The plumbing got fixed. Reputation is now portable and public. A freelance developer's GitHub history is more legible than most resumes. A designer's Dribbble account is a truer credential than a diploma. Payment, escrow, dispute resolution, none of it requires a payroll department anymore. Malt runs freelance operations across a hundred countries for Schneider Electric's industrial workforce. Worksuite took a Nordic advisory firm called Aspia from 180 contractors to more than five thousand, cutting time to placement from days to about thirty minutes.

Ben Thompson has a name for what's underneath this: Aggregation Theory. The internet drives the marginal cost of distribution toward zero, and power moves away from whoever owns the supply chain toward whoever aggregates demand and owns the relationship with the customer. Thompson built the theory to explain Google and Amazon. It explains talent markets just as well. The employer used to own the supply chain of labor: the roster, the office, the org chart. Now a platform, a reputation graph, or simply a personal brand aggregates the demand for a skill and routes it directly to whoever holds it. The employer's monopoly on discovery is gone.

Talent no longer works for companies. It plugs into them.

The froth was the warning shot, not the revolution

If this all sounds like the DAO hype of 2021 or the future of work think pieces from the pandemic years, that's because it is, and that's exactly why people are dismissing it too early.

Carlota Perez spent a career studying how technological revolutions actually land, and her finding cuts against the hype cycle's own mythology. Every major shift, from railways to electricity to the internet, moves through two phases. First comes installation: a frantic, overfunded, bubble prone period where everyone builds infrastructure and nobody quite knows what it's for. Then comes deployment, the much longer phase where the technology stops being exciting and starts being boring, cheap, and invisible, which is precisely the moment it rewires the real economy.

The gig economy apps, the freelance marketplaces, the DAO experiments of 2015 to 2022, that was installation. Frothy, overhyped, full of failed experiments nobody remembers. What's happening now is duller and far more consequential: escrow, portable reputation, standardized contracts, and AI systems that can verify a deliverable without a manager in the loop. None of it makes headlines. All of it is deployment. Technology only changes the world during the deployment phase, when it becomes boring, cheap, and invisible, as Perez puts it, and boring is exactly what the plug-in economy has just become.

The math nobody has run

MBO Partners counted 72.9 million independent workers in the US in 2025, up again from the year before. Of those, 5.6 million earn more than 100,000 dollars a year, a figure that grew 19 percent in a single year and has nearly doubled since 2020.

Now apply a pattern from the open source data itself. The HBS study found something stark buried in that 8.8 trillion dollar figure: 96 percent of the value comes from just 5 percent of contributors. Open source value isn't evenly spread across a friendly commons. It's a power law, concentrated in a small number of people who are extraordinarily good at what they do and extraordinarily visible for it.

Run that same concentration through the six figure independent workforce, as a model, not a measurement. Take a conservative 150,000 dollar average across that 5.6 million cohort: roughly 840 billion dollars in total earnings. Apply the open source ratio and the top 5 percent of that group, about 280,000 people, would be capturing 96 percent of it: close to 800 billion dollars, or nearly 2.9 million dollars a year each. That's not a forecast. It's a thought experiment built on a real, documented pattern from the one open system we've already measured for a decade. But it points at something a lot of executives haven't priced in: the plug-in economy isn't a leveling force. It's a winner take most market wearing the costume of democratized work.

The paradox that decides who wins

This is where the story gets uncomfortable for anyone selling everyone's a free agent now as a clean narrative, because it isn't clean, and the theory that explains why comes from an unlikely source: Clayton Christensen's Jobs to Be Done framework, applied more literally than he probably intended.

When we buy a product, we essentially hire it to do a job, Christensen wrote. If it does the job well, we hire it again. If it doesn't, we fire it. Swap product for specialist and the sentence stops being a metaphor. Companies increasingly don't hire people. They hire outcomes, and they fire them just as fast when the outcome doesn't land. Jalapeno Digital has run a marketing agency for a decade with zero full time employees, just senior freelancers plugging into named accounts, and took a client's Google Ads return from 2.8x to 4.1x in ninety days doing it. Even Upwork, a public company built to broker freelance labor, runs 75 percent on freelancers itself. The company selling the disruption eats its own cooking.

But openness only beats the integrated firm once the interfaces between contributors are specified well enough to plug into cleanly: the brief, the deliverable, the payment terms, the proof of quality. Before that threshold, bundling wins, because unclear interfaces need constant human renegotiation, and renegotiation is what a company is actually built to absorb.

Russell Ackoff is the necessary cold shower here. An organization that optimizes its parts independently will systematically destroy the performance of the whole, he warned, decades before anyone was scattering a project across nine freelancers on three continents. A pile of excellent independent nodes isn't a system. It's a liability with good LinkedIn profiles, unless something, a maintainer, a spec, a shared standard, does the integrating. Open source projects don't run on chaos. They run on benevolent dictators, pull request review, and version control. Take away the git repo and the pull requests, and you don't get freedom. You get noise.

Where the hobby ends up

Torvalds never took a salary from a company for building Linux. These days, though, he draws his paycheck from the Linux Foundation, funded by membership dues from Microsoft, IBM, Intel, and Samsung: the very corporations his three line Usenet post was, in spirit, supposed to make obsolete.

The hobby didn't destroy the company. It got absorbed into a stranger, looser structure that pays a handful of maintainers to keep an entire planet's infrastructure from forking itself apart, funded by the giants who no longer need to employ the people building the thing they depend on most.

That's the future waiting on the other side of this argument. Not everyone as a free agent. Not the office as the last man standing. A thinner, more exposed layer of orchestration sitting on top of millions of open, modular contributors, most of them plugging in and out for a fraction of the value, a handful of them capturing nearly all of it, and almost none of them ever appearing on anyone's payroll.