Linus Torvalds

Linus Torvalds: The Accidental Emperor of Software

On 25 August 1991 a twenty-one-year-old Finnish student posted a note to the comp.os.minix newsgroup that has since become the most spectacularly wrong prediction in computing: "I'm doing a (free) operating system (just a hobby, won't be big and professional like gnu) for 386(486) AT clones." Thirty-five years later that hobby runs the majority of the world's supercomputers, every Android phone, and most of the internet's servers. Torvalds has since written a second piece of software, Git, that took over how nearly all professional programmers store their work. Two for two is not a normal career.

A Statistician's Grandson

He was born in Helsinki on 28 December 1969 into the Swedish-speaking Finnish minority, and into a family of writers and academics: his parents Nils and Anna Torvalds were both journalists, his grandfather Ole Torvalds a poet, his other grandfather Leo Törnqvist a professor of statistics at the University of Helsinki. He was named for Linus Pauling, though he has always insisted the credit is split — "half Nobel Prize-winning chemist and half blanket-carrying cartoon character," the other half being Linus from *Peanuts*.

It was Törnqvist, the statistician, who put a Commodore VIC-20 in front of him around 1981. "The end result," Torvalds later said, "was that I got pretty familiar with the machine." He started in BASIC, moved down to machine code addressing the 6502 directly, then bought a Sinclair QL and rebuilt much of it, writing his own assembler and a Pac-Man clone called Cool Man.

He entered the University of Helsinki in 1988, interrupted his studies in summer 1989 for eleven months of mandatory service in the Finnish Navy's Nyland Brigade — emerging as a second lieutenant and artillery observer — and eventually took his master's in computer science in 1997 with the highest grade, *laudatur*.

Two Purchases

In 1990 he met Unix for the first time on a DEC MicroVAX running ULTRIX, and bought Andrew Tanenbaum's *Operating Systems: Design and Implementation*, the textbook built around MINIX, a deliberately small teaching Unix. On 5 January 1991 he bought an Intel 80386 PC clone. Between the book and the machine sat an obvious, enormous gap: MINIX was a toy by design and its licence kept it that way.

He wanted to call his kernel Freax — free, freak, and X for Unix. The name did not survive contact with the FTP server: Ari Lemmke, the administrator hosting the code at the university, named the directory "linux," and that was that. Version 1.0 arrived on 14 March 1994.

The Licence

The decision that made Linux was not technical. Torvalds' first licence forbade commercial use. In autumn 1991 he went to hear Richard Stallman speak at Helsinki University of Technology — his fellow student Lars Wirzenius had arranged it — and under that influence and pressure from contributors, he relicensed the kernel under the GNU General Public License version 2. GPLv2 required anyone distributing modified Linux to publish their changes, which turned every commercial user into a contributor whether they liked it or not. Competitors who would not share a lunch table now share a kernel.

From Author to Referee

Torvalds worked at Transmeta in California from February 1997 to June 2003, then joined Open Source Development Labs, moving his family to Portland, Oregon in 2004 to be near its Beaverton headquarters; OSDL merged with the Free Standards Group to become the Linux Foundation, which now pays him to work on the kernel full time. In 1999 Red Hat and VA Linux gave him stock options out of gratitude; when both floated, the paper value briefly touched around $20 million.

His actual coding share shrank as the project grew. By 2006 roughly 2 per cent of the kernel was his; by 2012 he described his job as "mostly merging code written by others, with little programming." What he kept was final authority over what enters the mainline kernel — and, through the Linux Mark Institute, the trademark.

Git

On 3 April 2005, after criticism for running kernel development on the proprietary BitKeeper, Torvalds wrote a replacement version-control system called Git. It was a side project to unblock his real project. It is now the substrate of essentially all professional software development and the foundation of a multibillion-dollar industry built on top of it. In late 2011 he started a third program, Subsurface, for logging scuba dives — he took up diving in the early 2000s — and handed it to Dirk Hohndel in late 2012.

The Temper

Torvalds' conduct on the kernel mailing list became almost as famous as the kernel. At Aalto University in 2012 he told Nvidia "fuck you" on camera, complete with raised middle finger, calling it the single worst company he had dealt with. "I'd like to be a nice person and curse less," he once said, "and encourage people to grow rather than telling them they are idiots. I'm sorry — I tried, it's just not in me."

That ended, at least formally, on 16 September 2018, when the kernel's Code of Conflict was replaced with a Contributor Covenant-based Code of Conduct. In the release notes for Linux 4.19-rc4, following critical questions from *The New Yorker*, Torvalds apologised, called his personal attacks "unprofessional and uncalled for," and took time off "to get some assistance on how to understand people's emotions and respond appropriately." He returned after 4.19 shipped on 22 October 2018.

Why Linus Is Called a Genius

The claim rests on two separable things, and only one of them is intellectual in the usual sense. The first is architectural taste at very large scale: the ability to hold a system of tens of millions of lines in his head well enough to know, on sight, which patch will make it worse. The 2012 Millennium Technology Prize honoured him for creating an open-source operating system that now underpins Android phones, television recorders and the world's supercomputers — but the kernel's survival across thirty-five years and thousands of contributors is a judgment achievement, not a coding one. He wrote 2 per cent of it and shaped all of it.

The second is Git, and Git is the stronger evidence. He designed it in a few weeks to solve his own problem, and the data model he chose — content-addressed objects, cheap branching, distributed by default — proved so much better than the alternatives that the entire industry converted. Producing one field-defining system might be luck and timing; producing a second, in a different domain, on a deadline, is not.

The honest counter-case: Torvalds invented no new computer science. Linux is a reimplementation of Unix, a design from 1969, and he has always said so; the GPL that made it unstoppable was Stallman's, and the userland tools were the GNU project's. Much of what distinguishes Linux is unglamorous — relentless pragmatism, the refusal of elegant microkernel theory in favour of what worked, and thirty-five years of showing up. He is a supremely gifted engineer and editor rather than a theorist, and he would be the first to say it.

Legacy

The recognitions pile up oddly for a man who calls himself unpleasant: the Millennium Technology Prize in 2012, the Internet Hall of Fame the same year, the IEEE Computer Pioneer Award in 2014, the Lovelace Medal in 2000, a Takeda Award shared with Stallman in 2001, 17th place on Time's list of the century's most important people, and asteroid 9793 Torvalds. He became a US citizen in 2010. The deeper legacy is a working demonstration that a global commons, governed by one stubborn maintainer and a licence, can out-engineer any corporation on earth.

Achievements

• Notable work: Linux kernel
• Notable work: Subsurface
• Affiliated with Transmeta Corporation, Open Source Development Labs and Linux Foundation
• Educated at University of Helsinki and Svenska normallyceum
• Worked as programmer, software developer and writer