The World in a Box
Art by Mike LeeWHEN MATT WELSH INTRODUCES A Microsoft-dependent computer user to a Linux box, he can't start by explaining that Linux is different from Windows. He has to explain that Linux is not Windows. Otherwise, "people will say, 'Oh, well, this is nice -- what version of Windows do I need to run it?'
"People have always equated Windows and the Intel-based PC," Welsh says. "They sometimes don't understand that it's an operating system -- that it's not part of the hardware. They don't even understand what an operating system is."
As a Ph.D. candidate studying Internet Scale Systems Architecture at UC Berkeley, Welsh has made it his life's work not just to understand operating systems but to imagine what they could be. One of the co-founders of the Linux Documentation Project, a collection of documents available on the Web that describes how to use, install and hack Linux, Welsh has been an evangelist for open-source-code software -- software a programmer can dig into and rewrite -- since he started using Linux as a Cornell University freshman in 1992. Now 24, Welsh has moved on to studying how to apply such software to the "next generation of services you'll find on the Internet." In other words, Welsh is looking for ways to put open and free, as opposed to closed and proprietary, systems on mobile phones and faxes, satellite dishes and television sets.
"It's one big global computer approach," Welsh says of his work. "We're building a generic architecture of Internet services that scales up to billions of users."
Other developers have built Linux kernels for the Palm Pilot and other PDAs, offering an alternative to the standard Windows CE system. "It's not very flashy right now," Welsh attests. "But the level of engineering work that went into it is astounding."
WELSH ISN'T SO CONCERNED WITH CONverting the home computer user running Windows 98 to a new software paradigm. "For the typical user, Word, Excel and Outlook are just fine," he allows. It's the Windows-dependent scientist, and especially the computer scientist, that has him worried, and that inspires his Web page, http://www.boycott-ms.org. "I've been seeing a large number of research projects in which Microsoft systems such as Windows 95 and Windows NT [Microsoft's network-server OS] are being adopted more and more of the time," he says. "It started to frighten me when I realized what would happen if one company were dictating the standard way of thinking in science." For starters, he warns, "We'd be stuck with a simplistic model of computing. The Internet would not be what users or programmers want it to be. It would be about what one company wants it to be."
Even worse, so would the society we live in. "The Internet is such a social mechanism," Welsh says, "and because of that it's defining the way people are collaborating with each other. It's changing the way communities and societies are built." If we're in the habit of locking up the blueprints of our software, we may soon decide to stem the flow of other vital information.
So far, that hasn't happened: "This is open technology, a system based on open ideas and open protocols," Welsh says. "As much as certain companies would like you to think otherwise, it's still everybody's Internet."
Welsh runs nothing but Linux in his own life: He writes letters in LaTeX, a text-processing program popular with engineers. He does spreadsheets and graphics in X-Windows, the Unix graphical user interface. He cables together racks of PCs and installs cooperating kernels on each one, creating one big supercomputer with Linux-based nodes on which he can fine-tune complex applications. But he isn't advocating Linux as the only way to smash the prevalent OS paradigm. "There are people doing much crazier things; people pushing out the envelope of what a computer is." Welsh cites the multimedia Nemesis operating system in development at the University of Cambridge -- a system that allows a user to manually allocate memory to each program -- and Sun Microsystems' Java Rings, computers with chips dedicated to Java applications, as examples. Designing such systems means "starting from scratch, really building from the ground up." It requires, Welsh maintains, "a non-Microsoft mode of thinking."
A non-Microsoft state of mind has never been so important. "We're at a critical point in the future of technology, because the next generation of computer scientists are going to be raised in a world that's almost entirely Windows-based. How are they supposed to conceive of something better? Today all of us drive automobiles that have gas pedals on the right. How can someone build a vehicular transport that doesn't have a gas pedal at all?
"Well, of course, no one can," Welsh admits, pausing for a moment to reflect on his metaphor. "Which is why it's important to remember that software is much more flexible than a car."
Get the This Week's Top Stories Newsletter