Proclamations of Linux's commercial success are frequently punctuated with statements about how some analyst firm has found that sales have grown faster than has any operating system since 1999, or that Linux server shipments make it the second most popular operating system for servers.
To me, this kind of "Linux as product" mentality misses the entire point of Linux and the open-source development model that created it.
Linux is not a product. Rather, Linux is a collection of software components, individually crafted by thousands of independent hands around the world, with each component changing and evolving on its own independent timetable.
To think of Linux as a product is to freeze an inherently dynamic thing in time and to close something that is inherently open. It cannot be done without losing something--and something significant at that.
No, Linux is not a product. It is a process.
Sure, Linux distributions--integrated collections of software combined with the Linux kernel to form complete operating systems--have been the primary drivers of the commercialization of Linux over the past several years. And the distributions do play a vital role in the Linux ecosystem. They collect "best of class" open-source technologies and integrate them into complete offerings. This allows Linux to be distributed to end users as an integrated solution as opposed to just a collection of disparate technologies.
Does this mean that the only way to commercialize Linux is to sell the integrated solution as a traditional, "one size fits all" product? To do so, I reiterate, is to miss the entire point of Linux, because Linux is fundamentally different from traditional operating system products--both technologically and, for lack of a better word, culturally.
Yet the business models that are built around today's leading commercial Linux distributions are remarkably similar to those built around the proprietary operating systems they seek to replace: Position the OS as a standard, then control the standard by locking customers in through proprietary technology, subscription agreement legerdemain or an "embrace and extend" approach to existing standards and best practices.
Let's step back a bit and look at why people are flocking to Linux. It's an open platform that is not owned or controlled by any single company. It comes with unmatched customization, optimization and integration possibilities. It is the ideal "invisible engine" for driving the next generation of applications and services. And it gives its users greater control over the evolution of the underlying platform, putting the user firmly in control of product release timelines and rollout schedules. In short, with Linux, the balance of power has finally shifted back from company to user.
The Linux distribution industry needs to start looking at Linux in a different way.
"Around the edges" incompatibilities introduced by the company lead, once again, to a lock-in and reliance on that company for services and support. The seller, rather than the customer, once again controls the evolution of the platform and dictates the timeline on which the customer must release or implement new products and services. And the one-size-fits-all, Linux-as-product approach complicates or eliminates the ability to customize, optimize and integrate. Sure, the product can be modified--but only at the cost of losing service and support.
The Linux distribution industry needs to start looking at Linux in a new and different way--as a platform to be shared rather than as a product to be owned. Linux distributors need business models that better match the fundamental differences that Linux brings to the market in technology, culture and process. They need business models that preserve the magic that has made Linux what it is today.
At stake here is not just the commercial viability of Linux distributors but the Linux ecosystem itself. In Nicholas Carr's recent Harvard Business Review piece, "IT Doesn't Matter," he draws a distinction between proprietary technologies and infrastructure technologies and points out that infrastructure technologies "offer far more value when shared than when used in isolation."
Linux is clearly an infrastructure technology. After all, it grew out of the Internet, the most recent infrastructure technology to shake the foundations of the information technology industry. Imagine if the Internet had at some point changed from an infrastructure technology to a proprietary technology. Where would that have left us?
By pursuing traditional platform lock-in strategies, the Linux distribution industry is in effect attempting to turn an infrastructure technology into a proprietary technology. The source code may be available, but what good is that if I'm stuck with the company who gives it to me?
Biography
Ian Murdock is a long-time Linux user, developer and advocate. He founded Debian, a Linux distribution project, in 1993. He is also a co-founder and chairman of Progeny, a Linux distribution company that sells Linux as a platform.
