Linux Turns 35: Just a Hobby That Runs the World

Linux Turns 35: Just a Hobby That Runs the World

Thirty five years ago today, on August 25, 1991, a 21 year old student at the University of Helsinki posted a message to the Usenet newsgroup comp.os.minix. He wanted feedback on a small operating system he had been writing for fun on his new 386 PC. He apologized in advance for its modest ambitions. The message contained what has since become the most famous understatement in the history of computing, and the code it announced now runs roughly 90 percent of the public cloud, every Android phone on the planet, all 500 of the fastest supercomputers on Earth, and the flight computers of rockets that land themselves on barges in the Atlantic.

This is a birthday worth pausing on, and not only for nostalgia. Because in 2026, the idea at the center of that old Usenet post, that software can be free, inspectable and owned by no one, has stopped being a philosophical position for enthusiasts and has become a matter of state policy. Governments are rediscovering free software for a very unromantic reason: they have realized what it costs to depend on someone else. But first, the story.

A Bedroom in Helsinki, August 1991

In early 1991 Linus Torvalds spent his savings, plus a student loan, on a PC with an Intel 386 processor. He ran Minix on it, the small Unix like teaching system written by professor Andrew Tanenbaum, and grew frustrated with its limitations and its licensing. So he started writing his own kernel, initially as a way to learn the 386. By August he had something that could run bash and gcc, and he decided to ask the Minix community what features they would want. The subject line was modest: What would you like to see most in minix?

Version 0.01 went online in September 1991: a bit more than 10,000 lines of code that could not do much beyond booting and running a shell. Torvalds wanted to call it Freax, a mix of free, freak and Unix, because he thought naming it after himself would be arrogant. Ari Lemmke, who administered the FTP server at the Helsinki University of Technology where the code was uploaded, disliked the name and quietly put the files in a directory called linux. The most important branding decision in open source history was made by a system administrator naming a folder.

The Idea That Made It Unstoppable

The kernel itself was not the revolutionary part. Unix clones existed, and in January 1992 Tanenbaum famously told Torvalds in a public debate that Linux was obsolete, a monolithic design tied to one processor architecture. On pure computer science, Tanenbaum had a point. What he underestimated was everything around the code.

In early 1992 Torvalds relicensed the kernel under the GNU General Public License, the license written by Richard Stallman for the GNU project. The GPL carries a simple but radical mechanism: anyone may use, study, modify and redistribute the software, but improvements that are distributed must be shared under the same terms. Code can flow in from anywhere and can never be fenced off again. Combined with the young Internet, this turned Linux into something no company could replicate: a project where a graduate student in Helsinki, an IBM engineer and a hobbyist in Brazil could all improve the same kernel, and none of them could take it away from the others.

That development model scaled beyond anything the industry had seen. Recent kernel release cycles involve around 2,000 individual developers per release, employed by companies that compete fiercely with each other everywhere except inside the kernel tree. And when the tooling of the time could no longer handle the flow of patches, Torvalds solved that problem too: in 2005, in about two weeks, he wrote Git. The same person gave the world its dominant operating system kernel and its dominant version control system, and gave both away.

From Hobby to Geopolitics

For most of its life, the case for Linux and free software was technical or economic: it is stable, it is auditable, it costs nothing to license. In the last two years a third argument has moved to the front, and it is political. When your government, your courts and your hospitals run on software licensed from a foreign corporation, that corporation, and the state it answers to, holds a switch that can be flipped. This stopped being a theoretical concern in 2025.

The reaction across Europe has been concrete rather than rhetorical. The German state of Schleswig-Holstein is completing the migration of about 30,000 government workstations from Windows and Office to Linux and LibreOffice, reporting savings of more than 15 million euros in license costs in 2026 against a one time migration cost of about 9 million. Denmark's Ministry of Digital Affairs announced in June 2025 that it would phase out Microsoft software in favor of open alternatives, with minister Caroline Stage Olsen explaining that the country must never make itself so dependent on so few that it can no longer act freely. France's gendarmerie has run its desktops on Linux for years, Switzerland has legislated in favor of open source in government, and at the end of 2025 all 27 EU member states adopted a charter on European digital sovereignty.

None of these governments are acting out of love for the GPL. They are acting out of the same cold logic that Torvalds stumbled into as a student who could not afford a Unix workstation: software you control cannot be taken from you, repriced under you, or switched off above you. Free software turns out to be what political scientists would call a hedge, and what engineers have always called not having a single point of failure. The fact that the hedge also happens to be excellent engineering, maintained by thousands of the best developers in the world, is the part that took 35 years to build and cannot be bought at any price.

What This Means If You Write Software

There are two quiet lessons in this story for anyone who builds things. The first is about scale of ambition: Linux did not win because it aimed to win. It won because it was useful, open to contribution, and honest about what it was. The projects that changed computing this deeply, Linux, Git, the web itself, all started as tools their creators wanted for themselves and released without a business plan. The second is about durability: every proprietary platform of 1991 is a museum piece, while the hobby kernel is more relevant than ever. Openness is not just an ethical choice. Over long enough time horizons, it is a survival strategy.

And if you have never read the original thread, take five minutes today and read it in the archives. There is something genuinely moving about watching the most consequential software project in history begin with a student politely asking strangers what they would like to see, and promising he would try to implement it.

Published August 2026. This is an opinion piece and analysis, not a sponsored post. CodeHelper has no commercial relationship with the companies mentioned.