What motivates or drives people to contribute to open source? What prevents people from doing so? Does open source make business sense? Is the open source business model broken or does the magic cauldron still have power?
Open source looks like a glitch in the economy. Good software materializes “for free,” strangers fix it, companies build on it, and the whole thing somehow does not implode.
ESR’s answer is annoyingly grounded: it is not magic, it is culture. “Cultures are adaptive machines,” he says, and open source is a response to specific pressures, with a bunch of unspoken rules that sometimes contradict the official ideology.
The simplest way I can explain the whole thing is a movie scene.
In Steve Jobs (2015), Woz keeps begging Steve to acknowledge the Apple II team on stage. Just let them stand up, give them applause, morale booster, basic human stuff. Steve refuses. Woz is not asking for stock options or a bonus. He wants public credit.
That is open source in a nutshell. Developers run on praise. Especially after a good piece of work. And when money is not the primary reward, credit becomes the paycheck.
ESR calls it a gift economy. You give code away, you get reputation. Your name on the thing is value.
What drives people to contribute
A lot of people contribute for extremely unromantic reasons:
They want the tool to exist & for the fun of it.
Most open source starts as “I need this” not “I want to save the world.” You scratch your own itch, you push it public because it costs almost nothing, and then people show up with patches because they have the same itch and slightly different fingernails.
Then the other incentives stack on top:
- Praise and credit. Woz wanted applause for the Apple II team. Maintainers want the AUTHORS file to not get rewritten by amnesiacs. That is why “do not remove someone’s name from credits” is basically a sacred rule.
- Reputation and career leverage. Your code is your résumé that actually runs. People can inspect your taste, not just your LinkedIn adjectives.
- Ego, but the useful kind. ESR is blunt about this. Prestige matters. Not in a cartoon villain way. In a “my peers respect this work” way.
- Community. Some people genuinely like the tiny village that forms around a project. The jokes in the issues. The “thanks, this saved my weekend” comment. The quiet feeling that you built a thing with other humans.
So when people ask “why would anyone give away code,” the answer is: they are not giving it away into a void. They are trading it for status, skill, relationships, and the weird satisfaction of seeing your idea propagate.
The taboos are not anti freedom, they are the price of not descending into chaos
ESR lists taboos that sound like they contradict open source: strong pressure against forking, frowned on to distribute changes without maintainer cooperation, absolutely do not delete names from history.
These norms ring true historically because they protect the actual scarce resources.
Code is not scarce. Attention is scarce. Trust is scarce. Maintainer time is scarce.
Forking is allowed legally, but socially it is treated like pulling the emergency brake. Not because forks are evil, but because casual forks split the community, split attention, and often kill momentum. Most projects would rather argue in public and keep one lineage than multiply into ten half-alive versions.
The taboo about “do not ship your own patch set without working with maintainers” is the same story. It pushes the ecosystem toward one shared source of truth instead of a swamp of private patch stacks.
And the credit taboo is the most important of all because in a gift economy, reputation is wages. Delete the name, steal the wages. It does not violate the spirit of open source. It is how open source survives contact with human ego.
What stops people from contributing
Open source does not fail because nobody cares. It fails because contributing is often a pain.
Tooling is annoying. Setup is annoying. Tests fail. The maintainers are tired. The docs are missing. People are scared of being that person who opens a PR that gets ignored for three months and then closed with “please rebase.”
So the cauldron has a hidden dependency: contributors have to feel safe, and the path from “I found a bug” to “I fixed the bug” has to not feel like a hazing ritual.
This is where money starts mattering again.
Open source runs on praise, but it increasingly survives on donations, sponsorships, and companies paying for maintenance because the alternative is watching critical infrastructure rot.
Platforms like GitHub Sponsors exist for exactly this reason: letting people financially support the projects they depend on. And Open Collective exists because you need boring financial plumbing, not just vibes.
The uncomfortable truth is that “thanks” does not pay rent, but it does keep people going long enough to maybe get paid later.
Long-term effects: the Mistral proof
This is where I think the story gets interesting long-term.
Mistral is a great case study because it proves you can drive real change without being the biggest valuation on the scoreboard.
OpenAI hit a $500B valuation in 2025. Anthropic raised at a $183B post-money valuation in 2025. Mistral’s 2025 round valued it around €11.7B, roughly in the $13B to $14B range.
So Mistral is about 35x smaller than OpenAI and about 13x smaller than Anthropic, give or take. And yet, if you care about open weights, Mistral has been punching way above its valuation for a while. They shipped serious open models early, and they publish benchmarks showing strong performance among open models, especially in coding and math, like Mixtral 3 last week.
OpenAI has now joined the open-weight party too with gpt-oss models under Apache 2.0, which is a big signal shift. But the broader point stands: open models create leverage. They let developers run things locally, fine-tune, build tools, and not ask permission. That is the long-term cultural power of open source translated into AI.
Big valuation buys distribution. Open weights buy a different kind of momentum.
Also, Astral is the same story in miniature
If you want the non-AI version of “small sharp tools quietly taking over,” look at Astral.
Ruff is MIT licensed. uv is dual MIT or Apache 2.0. ty is MIT. These are not giant platforms. They are fast tools that feel good to use. They built trust by being performant and boring in the best way. The ecosystem rewards that over time.
That is the part people miss when they obsess over “the next big thing.” A lot of the real “next big things” are just someone shipping a small tool that does not suck, then doing it again, then doing it again, until it becomes infrastructure.
So is open source broken?
Both. Which is the most irritating true answer.
It works as a production system. It builds the world’s plumbing. It also burns people out when the only compensation is applause.
Developers love praise. Open source runs on that. Open source also needs money, because servers, time, audits, and sanity are not free.
The cauldron still has power. It just needs groceries.