November 2002

Back at one of the early LinuxWorld Expos, when the vast Javits Center in New York was filled to the walls with booths and to the rafters with the festive sounds of speculative optimism, Jeff Gerhardt of The Linux Show said something that has stuck in my mind ever since.

He said, "I think just one of these booths costs more than the whole industry makes in a year."

Trade shows in speculative booms are like mining towns in the Old West: all false fronts on shacks and tents, mostly expressing the fantasies of their funders.

For Linux the funding fantasy began in August 1999, when the first Linux World Expo came to San Jose and Red Hat had its spectacular IPO. The next Linux World (when Jeff uttered his immortal words) was in February 2000, just two months after VA Linux had the most spectacular IPO in human history. At the time of the show VA was also in the midst of eating Andover, which had its own very impressive IPO on the same day as VA's. The price was over a $billion.

Whatever else was happening with Linux, a lot of people were suddenly very, very rich. At least on pixels. But the dot-com fantasy blue-screened a few months later, and the empty edifices came crashing down -- on Linux World Expo along with every other business trying to make real money in surreal times.

As I write this, the latest Linux World Expo has just ended. It was a fraction of its former size. The latest show fit with room to spare inside the North Hall (smaller than the South Hall) at San Francisco's Moscone Center. Except for Red Hat and SuSE, none of the commercial Linux distributions had a booth. Ximian's founders were there; but the company's signature jungle-theme booth and free stuffed monkeys were absent too, no doubt in compliance with an austerity imposed by its backers -- and its place in orbit around Sun Microsystems, which had a large booth at the show.

In fact, the big hardware companies seemed to take up half the convention floor. IBM and HP's booths were immense. Intel, Sun and AMDs were merely huge.

Lots of suits, too: My kind of guys. And most of them were happy, because -- for them, at least -- Linux is very good for business. We've been in a famously down economy, but nearly all the numbers for Linux have been pointing up. (See LJ Index in UpFront for a few of them.) For Linux Journal, it was the best show ever. We sold three times as many subscriptions as we've ever sold before, at any show, anywhere.

I heard attendance was also at it highest point since the inaugural show in 1999, and up 20% over the year before.

So while Linux gets bigger than ever, and its leading trade show gets more popular than ever, the show itself gets smaller and smaller as more and more companies either abandon the venue or move into the third party ghettoes of the big hardware company booths, which only get bigger. (I even heard that Apple would be exhibiting at the Linux World Expo in New York next Winter.)

What's up with all that?

Robin "Roblimo" Miller of OSDN (and author of The online rules of successful companies, due out right about now) thinks smoke is still rising from ground zero where the dot bomb went off and later collapsed. "We're still dealing with the wreckages of the Internet pump and dump schemes," he said. "The dumpster trucks keep driving up and pulling away."

Donald K. Rosenberg, author of Open Source: The Unauthorized White Papers, said "There's less hype now with the realization that Linux is at work and that there is a lot of it out here. But what we have now is real world excitement. Not contrived excitement."

There is also involved excitement. All the big vendors, especially H-P, IBM and Sun, bragged about how much they had contributed -- not just to The Linux Cause, but to Linux itself and its development tools.

The third party develpoment outfits were also looking strong. Trolltech was there, as always. Borland had one of the biggest non-hardware booths. Ted Shelton of Borland told me that Kylix, the company's Integrated Development Environment (IDE), is a huge success. It has been selling in the $millions, just as a cross-platform IDE for Delphi. Now that Kylix supports cross-platform C++ development, Borland expects Kylix' market to expand enormously.

So what's to stand in the way of universal Linux adoption, besides the vested inertia of potential customers in their own Microsoft environments?

One is the absence of a single Linux standard. While just about all the distros and hardware vendors say they stand foursquare behind the Linux Standard Base (LSB), Red Hat says it may be quite a while before all its offerings are in compliance -- and Red Hat has such a large market share that its distribution is effectively a second standard. Meanwhile major software and hardware makers (such as Oracle and Dell, just to pick two) often certify their products to different releases of Red Hat and/or the LSB. "Installations can be a nightmare," says Larry Augustin, founder and Chairman of VA Software. The majority of installation time is spent on getting hardware and application software to match up.

An executive with one of the software companies at the show told me "Linux hasn't just forked -- it's fanned out." His company copes by developing just for Red Hat and SuSE, and doesn't even both trying to comply with anything else.

David Sifry, who co-founded LinuxCare and Sputnik (where he's CTO), says,

Linux is 90% of the way there -- but getting the final 10% of the way requires a level of money, effort and fascism that doesn't exist in the Linux community. It is a level of integration that only a controlling monopolist can provide, and because the Linux user community rejects that model, there will always be a necessary level of expertise and pain-in-the-ass factor when integrating multi-vendor proprietary applications. This is why I think that the most realistic approach to integration is not at the application level but at the host level -- have a distinct host or cluster for each application, and it doesn't matter if one app is running on Debian and another is running on Red Hat. It only matters that the communications interfaces be compatable. That's also why I believe that communications glue like SOAP, XML-RPC and Jabber are vital infrastructure for the next generation of applications.

Another software maker told me there is no equivalent in the Linux world of Microsoft's wide array of development environments, tools and APIs. "There are too many new ideas that can still only be executed on Windows," one attendee at the show said. He also said the problem isn't isolated to Linux. "One of the reasons IIS has gained on Apache over the years is that there are certain kinds of e-commerce activities that only IIS will support."

But when I brought up these kinds of issues with folks at IBM, Sun and other new leaders in the Linux-in-business community, I got the same response: these are "remaining" problems.

It'll be interesting to see if how little they remain before the next LinuxWorld that comes here.

Doc Searls is Senior Editor of Linux Journal.