Before the Internet, networks were private affairs. You bought them from Novell or Microsoft or IBM. More accurately, you bought a set of network services you could obtain only from certain companies. You got file and print from Novell or Microsoft, messaging from Lotus, security from IBM and so on. None of those services would cooperate with those same services from other companies. That's not how “The System” worked. Every vendor wanted to be your only vendor—never mind that no vendor could provide anything close to all the services you needed.

The Internet broke that system by taking several network services and putting their protocols (i.e., HTTP for Web, SMTP for messaging, FTP for file transport and others) in the public domain, running on top of the equally public network protocols of TCP/IP. This gave us a high degree of interoperability for the first time. The Internet still does not perform many services at all, or performs them incompletely, so it did not put Novell and Microsoft out of business. The “Old Guys” still perform very well, but the Internet gave their businesses a whole new context—a public one. In a blink, their whole world changed, along with everyone else's.

It's still changing. When the Internet finishes settling in, almost everything in communications will depend on it, from embedded controllers to Palm Pilots to the entire phone system. But perhaps the Internet was just the sound of one shoe dropping—and the other shoe is Linux.

Linux seems to have the same calling: something free for everybody. The Internet seems to be its natural companion. Already the vast majority of web pages are served by Apache, running mostly on Linux systems. Are web services an enterprise beachhead for Linux? Or are they just a vertical application with little leverage?

Is there a big-time business model for any Linux vendor, besides attracting expensive investment bets from Microsoft OEMs and partners looking for alternatives? Are there other possibilities, all as radical and sensible as Linux itself?

How hard is it to imagine Linux kernels as the most basic building blocks of networked life? For Linux geeks, it goes without saying. For the rest of us it's a big stretch, especially when Linux still looks like a Swiss watch delivered in 400 separate parts. But so was imagining a networked world that wasn't a medieval mess of warring states dominated by the Empire of Microsoft, the Duchy of Novell and the Blue Kingdom of IBM.

We're still less than halfway through the shift from personal to social computing. Most households do not have PCs, and most that do are not connected to the Net. According to design critic and user advocate Don Norman, the two basic reasons for this are: computers are too complicated for many people, and the Net still lacks a plug-and-run infrastructure. He lays out a short-form prognosis in the title of his latest book, The Invisible Computer: Why Good Products Can Fail, the Personal Computer Is So Complex, and Information Appliances Are the Solution.

People are social. Telephony is equally social, because it lets people converse over simple appliances (incomprehensible cell phones and PBXes notwithstanding). Computing is social too, but only for a minority. There is still no computing appliance that's as social as the telephone. Will Linux deliver it?

I suggest that the first social computing appliance—let's call it the first SC—will cost less than $400, look friendlier than an iMac, get on the Internet with the ease of a phone call, and produce Microsoft Office-compatible files for those who want to use simple productivity applications.

Well, a prototype of that appliance can be built right now by slapping just about any Linux, any desktop and any office suite onto a commodity PC too slow for Windows 98, and a Windows newbie will probably know how to use it.

On the server side, the appliances are already here. The Cobalt Qube 2 is a seven-inch blue cube that anyone with an IP address can put on the Web in minutes and control easily over a browser. It runs Apache on Linux, but Cobalt doesn't mention that fact anywhere other than its data sheets. Why? Because it's an appliance. What's in it matters no more than what makes heat in a steam iron. All that matters is that it works.

Right now, the companies moving in the appliance direction are what economists call “fast followers.” Microsoft was once a fast follower of Apple, which is why Windows is a MacOS knockoff. Now Corel, KDE, Star Division and GNOME are fast followers of Microsoft, which is why their goods are not only effective clones, but in some cases (most notably Corel's) significantly improve on what they copy. Are they headed in the right direction? Not if they don't follow the user even more obsessively than they follow Microsoft. Don Norman says:

Currently, Linux requires far more compliance than Windows or Macintosh. Can that change? Only if some folks in the Linux community start to remember “the rest of us” abandoned by Apple fifteen years ago. If they do, the other shoe is sure to drop—big time.