Just shy of three years ago, on November 10th, 2003, Sam Docknevich of IBM’s Global Services delivered a presentation with the bold, even shocking title “The Time is Now for Linux on the Desktop.” Browsing the deck before it was pulled from public availability, I remember thinking that the content was actually fairly straightforward; very unemotional, and conservative in its projects and assumptions. As is perhaps unsurprising, however, the content was not what got written about – the title was.
The message was not new – far from it. Linux advocates had been making that same argument for years, to audiences with varying degrees of interest. What got everyone’s attention were the three simple letters that were emblazoned on his business card. For months rumors had been circulating that IBM was migrating some, all or none of its employees to Linux, and rewriting tens or thousands of huge or small internal Windows based applicatons to make this possible. Internal memos “leaked,” PR swung into action, and eventually the whole thing died down to the point that many if not most of you probably didn’t remember it.
Here we are three years later, and what’s changed? Very little. There’s still a lot of Windows, there are a few more Macs, and while there are some huge Linux desktop implementations across the globe, it’s still a distant third marketshare-wise. The first two points are unsurprising to me, but had you told me three years ago what the adoption would look like today, I would have been at least mildly surprised.
I’d written a paper, you see, following Docknevich’s presentation that made the case that the time for Linux was now – and was in fact now or never given the pending arrival of Longhorn. This, my friends, is why I try to stay away from predictions. The paper, which you can download here [1], said many things, but here’s the important bit from the conclusion:
Irrespective of the validity of predictions of breakout success for Linux on the desktop in 2004, it is impossible not to mark the coming year as a critical juncture for alternative desktops in general. Calling 2004 a make-or-break year would be an overstatement, but not by much. Conditions can hardly be expected to improve; from economic factors to available Linux skills to attitudes towards “good enough” solutions, the opportunity for alternative desktops is very ripe. In fact, Linux desktops in our view have a somewhat limited window to establish themselves as a mainstream enterprise desktop offering. Windows has arguably advanced relatively little since the release of Windows 95, with the primary changes being in reliability and ease of use, and as a result “good enough” Linux clients can fare decently in comparisons. But the pending release of Longhorn has the potential to impact the desktop market substantially, as it represents a nearly total overhaul of the operating system.
As you’ll probably note, there two major predictions I made that the official scorer has put in the books as errors:
- First, that Linux would see significant interest from major enterprises and limited adoption for certain low-need user constituencies
- Second, that Longhorn would be delivered in the medium term and reset expectations for operating system features and performance as a result
The latter prediction was off because I failed to heed my instincts that said that it was a Big Dig type endeavor; similarly ambitious, and similarly undeliverable. I simply didn’t believe that an organization of Microsoft’s size could not deliver an operating system in the timeframe promised – whether it was for business or technical reasons. That’s easy to dismiss, because I’m hardly the only one in the industry who expected the folks from Redmond to chuck something out the door before 2006.
But the lack of Linux adoption was not something I would or could have predicted, simply because from a business sense the case is sound. Linux was at the time more secure, cheaper, good enough, and fairly interoperable in Windows environments. There are many reasons that Linux on the desktop has achieved only slow and incremental growth – lack of commitment from major vendors with size and reach, continued inattention from hardware manufacturers, and the sheer inertia of Windows within the enterprise. Ultimately, however, I think the lack of adoption can be summed up by one simple phrase: “it didn’t just work.”
Walk around OSCON, as Paul Cooper did, and you’ll see a ton of Macs. Why? Because they just work. Yes they’re pretty, yes they have all the latest eye candy, but the reason cited over and over by the Mac folks I speak with is that they Just. Work. Contrast that with the experience that Ian – a founder of the Debian project – recently had on his Thinkpad x31 and Linux doesn’t fare all that well. Alex, a Mac user, thought I was nuts for jumping through the hoops I did to get Linux up and running on my x60 (though it’s running like a champ at the moment). As Ian puts it in a comment,
For the most part, everything just works. Unfortunately, it’s the “for the most part” that’s the problem here, and it’s the few remaining little things that don’t always work—wireless, power management, 3D, etc.—that make for a supremely frustrating experience.
The list of things Ubuntu did not support on my x60 out of the box included suspend/hibernate (see here), wireless (thank you madwifi.org) and the microphone (sudo killall esd).
This “just works” factor is all the more critical given a trend I’ve come to understand in much more depth over the last several years, which is that while I had expected adoption of desktop Linux to be driven by large enterprise rollouts, I’m no longer sure that’s realistic. To a degree I wouldn’t have believed at the time, technological innovation is being driven by consumer rather than enterprise facing applications. Cell phones, instant messaging, web applications, and VOIP are all examples of technologies that took off amongst individual users well before they did businesses. I can do things with Google Calendar, which I get for free, that I couldn’t do with Microsoft Exchange, which I paid for. Windows and Office are in fact proof of this; businesses enjoy these operating systems because they transcend the office boundary. People use them at home and at work, meaning the role of the consumer – or individual user, if you prefer – is an important one, even in operating system adoption.
So that’s the bad news. What’s the good news? The desktop is getting very, very close to “Just. Works.” The software is light years ahead of where it was two years ago when I first transitioned over to desktop Linux, and even the hardware manufacturers are paying attention; Intel had a team of people down at DebConf, and we’re having ongoing conversations with hardware manufacturers about their Linux opportunities. I am, it would seem, more optimistic than Ian is on this topic; possibly because my hardware’s newer and better supported.
Consider for just a moment what worked when I did launch Ubuntu for the first time: wired network, browsing, office productivity, the volume controls on the Thinkpad, Windows networking, multiple IM networks, bluetooth, sound, video, USB camera/stick/external drives automounting, and so on. More impressive, there are aspects to Ubuntu that are superior to Windows. My iPod, for example, will no longer mount as an iPod in Windows 2000 or Windows XP – it simply appears as a drive. I’ve gone through Apple’s suggested help, and nothing. Linux? Mounts it just fine, and will write to it. While running Windows on my x40, I couldn’t initially figure out why it ran so much slower than Linux did – and then I remembered that I had a sizable chunk of my resources chewed up by Symantec’s anti-virus application, which was unnecessary in Linux.
The point here is not that Linux is perfect; it is not, and won’t be any time soon. But it is improving, and improving at an accelerated rate. The pace of innovation has been ratcheted up several notches, and I’m interested to see whether or not more high profile Doctorow/Pilgrim style switches have any impact on that.
The real question in my mind is not whether or not Linux can Just. Work. I believe that it can and will. What I’m more curious to see is what direction the desktops take in delivering a next generation desktop experience. Will they try to out-Windows Windows, or take an entirely different tack? I’m hoping, obviously, for the latter. More to the point, I think desktop Linux has an advantage in satisfying next generation users that it’s community has not fully perceived yet.
If one accepts that a great deal of application development and delivery will shift to the network – and there is ample evidence to support this from Google to Salesforce – it’s logical to assume that the operating system that best leverages these network applications will be well positioned for success over the longer term.
The best returns for a desktop provider, then, may be for someone who concedes that the desktop is no longer the end all and be all – the network is what’s important now. When I’m somewhere without network connectivity, I barely even think of my machine unless I’m writing. I think that the ultimate opportunity for Linux on the desktop is to forget about the desktop as it exists today. Microsoft can’t because that’s how it makes money, Apple can’t because that’s how they sell hardware. But Linux? Linux has no such shackles, no such Innovator’s Dilemma to cope with. Either way, it’ll be interesting to see how all of this plays out.
[1] Just ignore the CONFIDENTIAL STAMP across the top – I don’t have acccess to Word at the moment so I can’t regenerate the document without it.
