Just What is "Linux Hysteria"?

New report purports to go "beyond Linux hysteria" but misses much of forest for trees.

Sudden success almost always produces a certain delirium--both for those profiting and those merely watching in awe. Sudden success also tends to produce new conventional wisdoms, upon which a whole new set of fictions can be erected. New World Orders, New Economic Paradigms, etc. were variants of some of the more popular ficciones of the nineties.

So, in many ways, a brief published by Forrester Research that attempts to look beyond the Hindenberg-like stock valuations of Linux companies like Red Hat (Nasdaq: RHAT) and VA Linux Systems (Nasdaq: LNUX) is a welcome voice in the continuing conversation on the fate of companies looking to make money on the open source operating system. But while the Forrester Brief presents a few valuable reminders about open source business models, its overall understanding of the Linux phenomenon is limited to Linux proper and fails to see the more important connection between Linux and open source.

One small step for Linux is one giant step for the open source movement. Regardless of what happens to Linux--much less any individual Linux company--the true story over the past several months has been the triumph of the open source philosophy over some of the more vulgar, ham-handed proprietary models that are already wearing away. Unfortunately, this triumph must be somewhat qualified, in the same way that the philosophy has become qualified. The success of the open source philosophy has allowed companies to take extracts of the basic premise and distill them for their own ends (which are occasionally "anti-open" in spirit, if not in letter).

Remember way back when the Ur open source business model was based on service and technical support? Given the current preeminence of Red Hat, a distribution company, the old bromide that it is impossible to sustain revenue growth by selling a free operating system seems to have been tossed out of the window (along with a host of financial pastimes such as fundamental analysis and due diligence). The Forrester Brief brings much of this back, noting "Red Hat and Corel realize that reselling a freeware product is a tidy little business at best--not a recipe for growth." In fact, although Red Hat has been attempting to branch out into services (another Next Big Thing for many tech companies is the attempt to make "services" responsible for as much as 25% of revenues), the company still makes its money on selling Red Hat Linux. Of course, "makes its money" is a bit euphemistic. While revenue growth for Red Hat is increasing at a fairly dramatic rate of 24% from the second quarter to the third, the company reported a net third quarter loss of $3.6 million.

One salve against the damage distribution companies may face when sales revenues begin to dip--at least as far as Forrester is concerned--is what the Brief refers to as "defensible assets". If this sounds at all ominous, it may be because of moves by both Red Hat and TurboLinux (one of the leading Asian-area distributions) to separate their versions of Linux from the rest of the pack. Not all defensible assets are as problematic as others--defensible assets being characterized as that value a company has that is unique, defining, and/or virtually impossible for another company to replicate. Any number of things can represent defensible assets, from market share to patents to sheer talent. This last item--a stable of brilliant, experienced workers--is one of the major assets of a Linux company like Cobalt Networks, which is among those the Forrester Brief believes will "thrive" in the coming years.

Another Linux company with "defensible assets" of a different order is TurboLinux. While TurboLinux 4.0 is little distinguished from many of the other top of the line Linux distributions from Caldera, SuSE, Red Hat or others, TurboLinux's clustering software represents one of the more unique paths taken by a Linux vendor so far. The clustering software essentially combines Linux servers for high-performance, "mission-critical" tasks. While the kernel patch that provides for clustering was issued as GPL, TurboLinux prefers to make the rest of the software available only as proprietary binary-only extensions. How much of this represents the bogeyman known as code-forking remains to be seen. But the effect on the marketplace is clear: by tweaking the kernel on its own, TurboLinux has helped bring Linux closer to UNIX performance levels (the better to win the business of larger ISPs and corporations), as well as given investors an additional reason to support TurboLinux: a defensible asset.

It might be suggested that Red Hat, in its own way, is pursuing a similar strategy. Clearly, there is much to be made in the company's efforts to distinguish Red Hat Linux from its open source brethren. For some, this began with Red Hat's portal strategy, announced in the days leading up to its IPO this past summer. For others, it was Red Hat's certification program that seemed geared toward raising a cadre of programmers and system administrators virtually wet-nursed on Red Hat Linux. Again, the reaction from the Linux community has been varied, with Red Hat bashing slipping in and out of vogue with every new announcement or initiative. But, if the imperatives mentioned by the Forrester Brief are to be taken seriously, it is important to see Red Hat less as a shining dollar sign on a hill and more as a thick ledger book recording inevitably declining returns.

Given the dire predictions for individual Linux companies ("dire" in the sense of lacking significantly defensible assets, living up to Wall Street expectations, and the fact that Microsoft, Sun, and other competitors remain healthy and aggressive), one company that is not mentioned in the Brief is Linuxcare. In fact, considering three of the major tiger traps discussed in the Forrester Brief: the anemic growth outlook for "shrink-wrappers" like Red Hat, the long-term advantage of established systems sellers like IBM over relative newcomers like VA Linux, and the "defensible assets" thesis, Linuxcare would be among the few Linux companies to pass through this gauntlet unsnared. The company, recently in the news with its embedded systems service and support agreement with Motorola, is privately-held and newly flush with venture capital. Its business model is as close to old-fashioned as you can get in the world of Linux: supporting all distributions all of the time. In contrast to pure Linux systems companies, Linuxcare has no long-established competitor in the technical service and support arena. If anything, Linuxcare's greatest competition is and will be distribution companies trying to offset sales revenues with service and support contracts (and training programs). But here again, Linuxcare reveals an advantage: having focused on providing pan-distribution service and support, Linuxcare has developed and continues to attract some of the best talent in the Linux community.

Linuxcare is one of the trees missed in the Forrester Brief--along with the rainforest that is open source. In essence, a 21st century revision of the attitudes of collegiality and peer review, the open source phenomenon draws upon the strength of intellectual productivity breakthroughs of the past: the think tank and the colloquium, the salon and the symposium (derived from the Latin term for "drinking party"--appropriately enough).

And what will be interesting in 2000 will be watching how companies--Linux-devoted, Linux-friendly, and Linux-be-damned--use open source methods to advance or shore up their corporate fortunes. Sun Microsystem's Community Source License, for example, is one model based on open source postures. In some ways an "open front door, closed back door" model, the SCSL provides developers with source code and the right to make and use changes. However, Sun owns all modifications made to the source code under the "community" license-- which is a locus for derision by some who prefer an open back end as well as an open front. Still, these licenses show that, if nothing else, companies themselves recognize how bringing in (potentially) legions of at-large programmers is an effective and cost-effective way to jump-start (or jump-restart) lagging development projects.

Another spin on the open source approach is to release source code for earlier generation software. This has been the strategy of many smaller companies who have open sourced device drivers, games, and other programs and applications in an attempt to both soothe computer users who can now hack their favorite software into better performance and compatibility with new technologies, as well as convince the public that the company is riding the curl of the wave, rather than paddling with all four feet trying to catch up. An additional benefit of this sort of open sourcing is how it enables smaller companies to leverage the sizeable talent of the open source developer community without, well, having to pay a dime.

It should be mentioned somewhere in this aria to open source that just deciding to go open source is, in and of itself, no guarantee of success. The Mozilla project to develop an open source browser is perhaps the most storied, incomplete open source project. And Apple's struggles in attempting to open source Darwin are veering on the pathetic, with much of the released code being referred to as "mostly useless" by many hackers.

Open source methods pave the way for a "quiet hysteria". This apparent oxymoron is meant to suggest the sense of boundless potential that many feel open source represents (also known as "hysteria") and the notion that, while mention of the word "Linux" may still draw "huzzahs" from fans in the box seats, the bleacher bums among us will know that the game remains much bigger and much better than any given group of players.

email: david@ssc.com