REBEL CODE: LINUX AND THE OPEN SOURCE REVOLUTION

 

Glyn Moody

 

Perseus, Cambridge, Mass., 2001

 

The Hidden Revolution

Ten years ago, Linus Torvalds was a shy undergraduate at a Finnish university, his Linux operating system did not exist, and few took seriously the idea that large amounts of quality software could be developed outside the closed framework of commercial companies. Today, open source and free software runs most of the Internet, and Linux is the fastest-growing operating system, prompting a sea change in how many companies do business. Glyn Moody charts this rapid transformation in his book Rebel Code: Linux and the Open Source Revolution and tells how chance and fortune combined with the Internet and the availability of cheap processing power to make the impossible pos-sible. IEEE Spectrum's Stephen Cass talked to him to find out more.

Why did you write this history?

The more I got into the Internet, the more I kept coming across linux and GNU [see Defining Terms, p. 80]....It struck me that it was tragic that all this history was dis-appearing because a lot of it was in [Usenet] newsgroups and a lot of these are starting to be wiped offhard disks....! was keen to try and get down on paper what really happened, because with time, we are going to find it harder and harder to find the original sources. I thought it was very important to try and pin this history down for the future.

How did you assemble your Information?

Largely through interviews. Otherwise I used the Internet. I managed to dig out the most amazing things, which are there for anyone to see but almost impossible to find. Often I'd follow up a lead, and halfway through a Web page, there'd be a link to something else or a reference. It was a kind of on- line detective work, which I found immensely exciting, because a lot of this stuff hasn't appeared before in the mainstream media. I felt a lot like an archeologist, unearthing a lost civi-lization with the para-doxical fact that most of the inhabitants were still alive, even though the civilization itself was hidden.

When did you see Linux become significant outside the hacker community?

The breakthrough really began at the end of'97, early '98. In my book I describe the key moments, like the Freeware Summit Conference in Palo Alto in April 1998 that Tim O'Reilly [the president ofO'Reilly and Associates Publishing] organized for key open-source developers.

Then attendance at [the Silicon Valley Linux users' group] by Oracle and Intel changed those companies' strategies for developing and distributing products- the effect of seeing a thousand wild Unux lovers together and their sheer energy con-vinced even the suits that their strategies [needed changing]....Netscape taking its Communicator 5 browser open source made people sit up.

One of the interesting things I discov-ered was how much that was an act of desperation, whereas at the time it looked like a tremendously bold move. They had-n't anything better to do, and they really had nothing to lose.

I was also impressed by how IBM came to back Apache server software in the middle of '98 when it decided to include Apache with all its e-commerce products. There was a lot happening in the background that most people haven't come across because it was all behind closed doors. But about '98 all of these events started happening to-gether... [and] that was the crit-ical moment, when it went from being the private secret of initiates to being a kind of public secret. It certainly wasn't known in the boardrooms, but I think savvy engineers started saying, "Linux [and open source] is a serious thing."

Why did Linux become the dominant open source operating system, when there where similar competinq projects?

The answer to that is complex. If a few factors had been missing, then none of this would have happened. Linus Tor-valds only created the Linux kernel. We'd still be waiting for the other 90 percent of the operating system if Richard Stallman and the Free Software Foundation hadn't started the GNU project [to create a free clone of the Unix operating system]. But luckily, Stallman had done 90 percent of the work and just happened to get stuck on the kernel so there was a hole waiting, which the Linux kernel filled.

So many other operating systems nearly made it but didn't. If you look at the 386BSD project [a PC version of BSD Unix], that could have been Unux, but the personalities of the leaders were wrong there. This is the crucial contribution of Linus in many ways; his personality just happened to be the right one. When peo-ple sent in fixes, he didn't say "Okay we'll include it in the next release in about six months' time." He fired off an e-mail say-ing, "Fantastic, I've put it in the code, and great work." This gave a tremendous buzz to people. He is a very gifted manager without ever having been trained as such, and he just made a lot of the right deci-sions instinctively

Other factors [also] happened by chance, like the fact that Linus had this Prince of Persia game on his computer. This meant that he wanted Linux to be able to share a hard disk with DOS. With 386BSD you had to scrub off all your DOS programs, which a lot of people were reluctant to do. Having a game on your computer determine the history of computing is crazy, but those tiny elements put together just made everything fall into place.

Why did PC users flock to a Unix-like oper-ating system that had its root in large multi-user systems?

Three main reasons. One was that Unix was always the hacker's system. In the early 'yos, nearly all hackers used Unix because it was freely available at universi-ties.... When they found they couldn't get access to the source code because AT&T [changed the license], they started feeling disgruntled and thought they'd like their own free version of Unix.

Another important reason is the fact that Unix was one of the few truly por-table operating systems.

The final factor, which is perhaps a bit more subtle but is crucial, is that Unix isn't a monolithic slab of code-it consists of hundreds of little utilities. And therefore it is much easier for somebody like Richard Stallman to sit down and write a done like GNU because he just writes one hundred little pieces, rather than needing to write one huge piece.

Also, other people can join in much more easily when there are a hundred clear subtasks, rather than one massive task. That's fed through into the structure of not just Linux, but most of the main open source and free software projects; they're highly modular.

 

Why is Linux so central to the open source movementi and what are some of the other large open source projects?

So many gifted people are producing such good software, it's unfair that they tend to get pushed into the shadows. Unux has the lion's share of attention because it's attack-ing Microsoft. But there are things like Peri and Apache and Sendmail and Bind [which run most of the Internet]. They're almost all highly modular because what's come to be called the open source methodology of producing software depends on that modularity. You've got to be able to have people working [anywhere in the world] on a little module and have it inter-face cleanly with all the other modules. The only way you can do that is by having a completely dean interface. Unix [and Linux] always had that from the start, so it made it perfect for the open source/free software methodology even though that hadn't been invented yet.

Is there a common factor among the key open source developers?

Humility. They really want to serve people. The/re the most modest people you can imagine. Sometimes they're a little bit pleased with what they've done, justifi-ably...but they have a tremendously deep respect for the user. Responsiveness is common to all these people; they listen to the users.

As the open source user base grows to include nonhackers, will that responsive-ness persist?

Yes-for example, linus says he doesn't care about user interfaces. He loves the kernel, but there will always be another layer to put on top, and those new layers will address the needs of the users. That's what's interesting about the KDE and Gnome [desktop environments] projects. The teams behind these projects specifi-cally say they're trying to give users an even better interface. The older hackers probably don't care very much about user interfaces, but the younger ones will, because that's where the action is.