It is fashionable to label Linux a disruptive technology, with Windows 2000 playing the part of the soon-to-be-evicted market leader. A Google search for "disruptive technology" together with "Linux" returns over a thousand hits. The reason is obvious: In the disruptive technology script, the market leader cannot adapt, fails miserably, and its Chief Software Architect winds up on the street selling apples -- an ideal scenario for Linux advocates.
Unfortunately, not every exciting new technology is automatically disruptive. According to Christensen, the key attributes of a disruptive technology are that it is in some way slower, smaller, or cheaper, with lower profit margins. The existing companies ignore the new technology because the metrics their customers use to evaluate technologies make the new one seem less appealing, and because the companies are seeking higher profits, not lower ones. But the disruptive technology creates new markets, with new kinds of customers, who have new evaluation metrics -- a combination that Christensen calls a "value network." For example, when Novell was the leading network operating system vendor, speed was one of the main evaluation metrics. Windows NT was not faster, but it was a much better environment for developing server applications. Eventually this became a much more significant metric than speed, creating a new value network that drove Windows NT to overtake Novell.
The "dilemma" of the book's title is the fact that the time-honored management practice of listening and responding to existing customers will inexorably lead the market leaders towards faster, larger, more expensive products. Eventually the new technology becomes "good enough" for most customers (as Windows NT was deemed "fast enough" compared to Novell), and the faster/larger/more expensive products, while increasingly profitable, become a high-end niche market. Thus, best management practices not only do not help the large companies harness disruptive technologies, they actually exacerbate their damaging effect.
The first point about Linux is that it is not that different from Windows 2000, its main competitor. On a basic level, it is still an operating system that manages hardware and provides a platform for applications. This is in contract to most disruptive technologies in
The Innovator's Dilemma, which are physically different products (an example is hydraulically- vs. cable-actuated excavators).
In some cases, however, the final product is the same, and only the methodology is different, as in the case of steel minimills vs. integrated mills. But again, the methodology under which Linux is produced is not that different from Windows 2000. Programmers write code, compile it, test it, debug it, package it together with other code, and so on. Of course, after all this work is done, Linux is given away for free while Microsoft charges hundreds or thousands of dollars for Windows 2000. This certainly qualifies it as cheaper, one of the key attributes of a disruptive technology. Part of this cheapness is indeed due to the open source development process, which allows a smoother linking between code that needs to be written and people who want to write it, and also allows easier bug-fixing. You could argue that Linux is disproportionately cheap -- that is, while the open source process may mean that producing Linux only takes (as an example) 80% of the effort that Windows 2000 does, it sells for 0% of the price. This may imply that at some point in the future Linux won't be as cheap, but for the moment we can unquestionably state the obvious, that Linux is much cheaper than Windows 2000.
Furthermore, it certainly may appeal to different customers in a different value network. That is, while Windows 2000 would appeal to customers who value the Microsoft name, Linux appeals to those who care about price and access to source code.
Nonetheless I would argue that Linux does not fit as a disruptive technology relative to Windows 2000.
First of all, it is not a "lower end" product. In the spectrum of operating systems, Linux is more situated towards the reliable, power-user end of things than Windows 2000. Using the language from the book, it is "upmarket" from Windows 2000. But disruptive technology comes from below, with downmarket technology pushing aside upmarket technology. The only "downmarket" aspect of Linux is its price.
Secondly, Microsoft's customers are not leading it
from Linux's features, but
them. Microsoft may be too dense to notice right now, but what its customers want are what Linux has and Windows 2000 lacks -- reliability, remote management, quicker turnaround on bug fixes, trust in its security, a real shell, and so on. Therefore Christensen's central thesis, that customers will force a company to migrate away from disruptive technology, doesn't apply. Microsoft is ignoring Linux not because of good management, but because of classic bad management: not listening to customers who do web hosting and database management and are defecting from Windows NT/2000 to Linux.
In fact, one could argue that Linux is more likely to be a Microsoft disruptee than disrupter. According to Christensen, value networks severely limit downmarket mobility. Linux has an almost perfect feedback cycle from its "customers," who in many cases are also its developers. But Christensen claims this can be a negative, since those customers will move Linux towards cool features like clustering and 64-bit support rather than focusing on the desktop market. In fact this is true of classical open source software in general; the "developer scratching an itch" method of software design is inevitably going to lead products upmarket, because developers are upmarket customers. This implies that it would be very very difficult for Linux to capture the desktop market, since the natural instinct of customers will push in the opposite direction.
The disruptive cycle would more naturally play out with Windows 2000 gaining the features that Linux has, pushing Linux into a smaller upmarket niche and displacing it in its current markets, as Windows 2000 became "secure enough" and "reliable enough" and so on, and people started worrying about new metrics where Windows 2000 beat Linux (such as better in-the-box driver support from hardware manufacturers).
To take this to an extreme example, at some future date Windows CE might displace both Windows 2000 and Linux, and the Personal Web Server shipped with Windows might displace both Internet Information Server and Apache. This is highly unlikely, but it illustrates the direction in which disruption happens.
This doesn't mean that Linux can't be a disruptive technology, it just points towards which direction it is going to disrupt. The first obvious candidate is Linux on a PC replacing Solaris on proprietary Sun hardware. Further down the road, Linux will be part of a larger trend, already in progress, in which personal computers (including those running Windows) displace minicomputers and mainframes. This is one big reason why Intel and Dell are interested in Linux.
Christensen offers a prescription for preventing disruption, but it might be hard in the case of Linux. The key is to spin off a separate "company" that cares about the downmarket customers, and can get excited about capturing them. So imagine a Linux fork aimed at the desktop. But since Linux does not have financial incentives to dangle in front of its developers, some other reward system would have to be put in place that would generate the proper inducements and peer respect for developers working on that fork. Maybe the legions of secretaries that Linux was going to "save" from Windows could swing by the development lab with a tray of donuts? Hey, whatever works.