Over the past several weeks, particularly at OSCON, I’ve spent a fair amount of time canvassing a variety of people – both in technology and not – concerning the concept of an Online Desktop. More specifically, the GNOME Online Desktop. Though visions of what that, specifically, might look like vary, the basic concept is simple: a tighter alignment of the desktop experience with the online applications that define the computing experience for many of us.
The transition is as far as I’m concerned, inevitable at some point. Indeed, considering developments like the extension allowing your del.icio.us bookmarks to replace those maintained by Firefox it could be argued that it’s already underway.
But from the conversations I’m having, there is both a lack of consensus in how – precisely – this should take place and a subtantial amount of concern for the implications.
I had the pleasure of speaking with OpenedHand’s Matthew Allum on Tuesday, as an example, and we got to debating some of the pros and cons of the online desktop approach. While he was certainly open minded on the subject, he raised some interesting obstacles that I hadn’t heard before, which I’ve lumped in with the more common objections. So while making no attempt to be comprehensive, then, a brief look at some of the most common problems that Matthew and many others see on the road ahead for the online desktop.
Problem 1: APIs
Description: Probably the most commonly cited objection: the reliance – even over-reliance – on APIs that may be brittle, overly dynamic, unstable, or all of the above. One example cited was the cat and mouse game IM clients used to gaining access to rival networks, getting cut off by protocol changes, then patching to reestablish connectivity.
Response: A legitimate concern, without question. Web APIs to tend to evolve quickly and as for compatibility, well, let’s just say it’s probably not their highest priority.
That said, I see a couple of factors mitigating against this weakness:
- The assumption most or even many of the APIs in question will be undocumented or one-off in nature is probably incorrect. A potentially sizable portion of the APIs will be non-dynamic standards (e.g. Atom, iCal, POP/IMAP, CSV, etc) or not terribly dynamic microformats (hCard, etc)
- If current usage patterns are any guide, most APIs will be either REST based or REST-like. While they may be brittle, then, they should be simple (relaitvely speaking) to adapt to given inevitable changes.
- Attitudes towards web APIs have evolved significantly. While the walled garden model stubbornly persists, it’s becoming increasingly obvious that open, documented APIs closely correlate to service usage and adoption (e.g. Facebook). Ergo, providers are unlikely to be quite as cavalier with breaking API compatibility as they would have been a few years ago.
Will be API breakage be a problem? Undoubtedly. But it need not be a showstopper, IMO.
Problem 2: Audience
Description: Much of the discussion around the online desktop has centered around questions of audience. Much of the world is diconnected, goes one argument, and therefore an online desktop will discriminate against those with limited connectivity options. Another says that elderly audiences won’t be sophisticated enough to use it – preferring as many of them something purely online (e.g. Web TV) – and that younger audiences may not see the benefit of offline at all given their heavy reliance on a.) browsers and b.) mobile phones. Meebo uptake somewhat validates this.
Response: To some degree, I believe that the vision of a desktop is necessarily exclusionary. It would be pointless, for example, to design for an online desktop that will never be online. Nor, likewise, would I aim the Online Desktop at older audiences, simply because they’re on balance less avid consumers of the variety of services discussed and less likely to exploit intermittent connectivity capabilities.
The Online Desktop should, in my opinion, be fairly narrowly focused in terms of its intended audience: heavy users of online services for whom the current desktop options are lacking.
Problem 3: History
Description: It is no secret that the vision articulated by the Online Desktop is neither unique nor particularly original. One could contend that in some fashion Sun’s original “The Network is the Computer” mission statement foreshadowed today’s efforts by decades. The real cautionary tale here, however, is Netscape. Their vision of the browser replacing the operating system not only didn’t come to fruition, it galvanized the sleeping giant of Microsoft who subsequently wiped them off the map. History, in other words, has not been terribly kind to visionaries in this space.
Response: History also told us that Software-as-a-Service was a doomed model, as just about everyone that tried it save Microsoft and a handful of others dried up and blew away in the face of massive customer indifference. Fast forward today, and it’s one of the more important trends affecting technology. Not to mention the likely backbone of any potential Online Desktop experience. Just because we weren’t ready for an Online Desktop in Netscape’s heyday doesn’t mean we aren’t now: the world has changed significantly since then.
Problem 4: Convention
Description: Conventional wisdom holds that non-Windows clients will have a tough time gaining a foothold on any sort of basis if for no other reason than they’re not Windows. Generations of business users have been raised on the Office/Windows combination, and many chafe at the idea of anything even marginally different. At the very least, non-Windows offerings start with a two-strike count, and have little margin for error.
Response: It would be foolish indeed to question Office/Windows’ dominance in the marketplace. Or most users’ acquaintance with the platform. But this argument breaks down for me on several levels:
- This essentially cedes the desktop to Windows – or Windows clones – indefinitely. That not only stifles opportunity for innovation (would OS X have innovated if it had to mirror Windows closely?), it condemns users to a future with no or very little choice.
- As important as the Office/Windows experience has been historically, the absolute dependency on these tools is lessening. Led by consumer technologies, enterprise applications (e.g. Salesforce.com) are increasingly migrating towards brower based delivery models: models which make the operating system underneath somewhat less important.
- Just as one generation of users grew up on Microsoft tools, so too is another growing up on web based tools. Not that they are not using Windows or Office extensively – they obviously are – but they’re spending far more time online. And businesses – even large ones, interestingly – are taking their cue, marketing on Facebook, MySpace, and so on.
- Although it’s still a distinct minority, marketshare-wise, OS X is living proof that the Windows paradigm is not the only one that works or that users can adapt to.
Problem 5: Opportunity
Description: The foundation of the online desktop vision is the belief that much of the computing experience as we know it today has transitioned to the browser. Witness experiments like the Alex’s Pyro desktop, which replaces the Window Manager on Linux with Firefox componentry. This sounds good on paper, but may lead a great many software developers to ask: why build anything for the desktop, then? If the world beyond the browser is destined to be a ghetto, what profit building there?
Response: This argument presupposes several things which I’m not terribly confident in. First, that the browser is or will be in the foreseeable future adequate for all or nearly all computing tasks. I do not agree – media players being exhibit A (Songbird notwithstanding – it’s a hybrid of internet and client side technology). It can’t be turtles all the way down, as they say. Second, that the browser technology will be up to the standards of client side demands: forgiving Pyro’s flakiness b/c it’s very alpha, consider Firefox’s tendency to leak memory. Window managing technologies may not be perfect, but they’re more stable – IMO – than browser components. Last, that the browser will be as eye candyish as the latest UIs from Apple, Microsoft, and yes, Linux.
Problem 6: Proprietary
Description: There are (at least) two facets to this argument: first, the very fact that the online applications the Online Desktop would integrate with are largely proprietary. Second, the fact that these proprietary applications are for the most part non-cross platform. While the first is mostly a problem for idealists and staunch free software advocates, the latter is relevant to anyone – think iTunes.
Response: The “open service” component to this argument is important, but it’s my hope that free and open source software advocates do not stall on this point. All things being equal, would I choose free and open source alternatives to, say, Flickr or Gmail? Probably. But all things are rarely equal – I’m not convinced that the F/OSS have a complete understanding of the costs involved in delivering SaaS – and I can guarantee you that the majority of everyday users will value brand over openness.
As for the second, it’s admittedly a significant issue. The fact that iTunes does not, and is not likely to at any point in the near term, operate on platforms other than OS X and Windows will certainly throttle adoption. That said, there are alternatives to iTunes, and over time it’s possible that this others will emerge. Given that Apple and Microsoft will have a vested interest in protecting their own services to preserve software and hardware revenues, this could end up being an advantage for other platforms: they simply don’t have as much to protect.
Is the Online Desktop a preordained success, then? Hardly. There are too many moving parts to be comfortable, and Apple and Microsoft have both demonstrated in the past the ability to adjust – and quickly – to emerging business opportunities. But the opportunity has, as I argued above, the air of inevitability to it. We’re going to see a convergence one way or another, it’s just a matter of who gets it right.
Update: It was pointed out to me in #redmonk that I hadn’t provided any context for the comments, so I added the clarification that this is somewhat GNOME specific.