Chapter 7 · 1991–1994

Helsinki

'Just a hobby, won't be big and professional like gnu.' It runs most of the planet now.

In the spring of 1991, a 21-year-old computer science student in Helsinki bought himself a PC. Linus Torvalds wanted to learn the machine down to the metal, and the operating system it shipped with wouldn’t let him. The teaching system he used at university — a miniature Unix called Minix — was deliberately kept simple, and its license kept people from improving it. So he started writing his own kernel. For fun. To learn.

That August he posted to the Minix newsgroup, and the post is now scripture, so quote it straight: “I’m doing a (free) operating system (just a hobby, won’t be big and professional like gnu).” He asked what features people wanted. He apologized in advance for how modest it would stay.

Hold the two halves of the last chapter in your head and watch them meet. Stallman’s GNU had spent eight years building everything except a kernel — compiler, tools, libraries, all free, all waiting. BSD, the obvious kernel, was about to spend two years frozen in a lawsuit. And here comes a kernel with nothing around it, written by a student who cheerfully bolted it to the GNU toolshed because the tools were right there, free, under a license that said yes. A kernel-shaped hole met a kernel missing everything else. The fit was so clean it looks planned. Nobody planned it.

What happened next had no precedent. Torvalds put the source code where anyone could take it, and strangers started sending fixes back. Ten. Hundreds. Eventually thousands of people who had never met, improving one program together across a network — and that sentence is the only one this book will spend on the internet, which deserves its own book and can’t have this one. By the mid-nineties, Linux — the name a friend picked; Torvalds thought it sounded egotistical — was outrunning systems built by corporations with thousand-person engineering staffs.

Why did the hobby win? Three inheritances, all from earlier chapters. Cheap hardware: chapter 3’s doubling had put real computers on student desks, and Linux ran on exactly those. Free tools: chapter 6’s GNU meant a whole development environment cost nothing. And the culture: chapter 4’s tapes had taught two generations that you read the source, fix it yourself, and mail the fix onward. The Lions book photocopiers and the Linux patch-senders are the same tradition, one license era apart.

There’s also the licensing what-if, and honest books admit their what-ifs. If the BSD lawsuit had never happened, the free Unix the world standardized on would probably have come from Berkeley, and Linux would be a footnote. Torvalds himself has said as much. History’s biggest operating system exists partly because AT&T’s lawyers picked the worst possible moment to sue.

So where did the hobby end up? That’s this chapter’s interactive, and the answer is: look around you. The counter below walks through the room you’re sitting in — phone, router, TV, car, thermostat, the server that just delivered this page — and tallies how many of them are running Linux right now. The number surprises almost everyone. The operating system that escaped Bell Labs as tapes, got enclosed, got rebuilt free, and got a kernel from a Helsinki bedroom is, by any honest count, the most widely deployed piece of software in human history.

And it was still missing one trick. All those Linux machines ran one program pile per box. The idea that would change that — the fencing trick from 1979, chapter 4’s footnote — was about to grow up. Next: slicing the machine.