Changing the Desktop Metaphor

Let’s talk about windows.

No, not Windows. Windows–the kind made out of glass.

It’s been raining a lot here in the Midwest these past couple of weeks, so I have been spending some contemplative moments away from the computer staring out my windows watching the garden grow, or the baby robins figure out how to master flight before the neighbors’ cat finds them.

Windows are a key feature of most buildings in modern times. They allow us glimpses of places near and far, keep us from feeling boxed in, and let sunlight or the night air wash over us. If you don’t think windows are such a big deal, try staying in a closed room for a while–and then remember the feeling of relief you get when you can first look outside again.

I’m waxing a little philosophically here for a reason. Lately, the debate has been stirred up on just what it means for Linux to be successful on the desktop. Many have put forth the notion that the introduction of netbooks, smartphones, and other mobile Internet devices (MIDs) into the marketplace puts Linux on a much better stance than its competitors, who seem shackled to the moribund desktop and laptop platforms. Others counter that MIDs aren’t really desktops, so counting them in the pro-Linux arguments is just a weak justification.

Let me start with the shocker. I will gladly concede the point: MIDs are not the desktop. So, to say Linux is making strides on the desktop because it is making strides on netbooks, phones, and the like is inaccurate in my opinion.

Before you get out the pitchforks, let me further assert that I no longer believe in the desktop model as metaphor/workflow for how we approach data any more.

Remember, desktops came about primarily because business users (and then home users) needed a way to duplicate the power of the thin-client/server model with independent, number-crunching machines that would take the processing workload off of the servers and on to the clients. Thus, the clients had to become more than just dumb terminals, and they had to have their own operating system and application framework to handle user data locally rather than remotely.

As a 40-something, I watched this happen in the corporate workplace, as we moved from dumb terminals to 386 machines running Win95. Data was crunched faster, and servers became useful more for storage and network services than data management (except for database servers, which still filled the role of massive data manipulation). Work was done on our desktops in the cubicles we inhabited like so many gophers.

But now, the metaphor is shifting yet again. Servers have grown powerful enough with clustering and processing advances that they have started to take back the heavy workloads from the desktop/laptop clients. Sure, some work is done locally, but more and more we are seeing really big workloads going back to the server environment, which thanks to Internet connectivity, we euphemistically refer to as “the cloud.” For me, I am watching the cycle of the early 90s reverse itself, as the data lives and works out in the cloud, and we access it through more sophisticated browsers and add-ons.

The desktop metaphor, then, is dying. The new metaphor is the window.

It is through these windows we can see the world. We can use specialized filters to see specialized data (“I’d like to see this KML file, please”) and because of the sophistication of the window tools, it doesn’t matter what form the window takes. It can be physically big, like a TV connected to an IP-driven cable feed, or small, like a hand-held MID. If the window is well-built, then it doesn’t matter what it looks like, nor what software it runs. The cloud is handling most of the work… all our machines need is a clear view of the data-world around us.

This, I believe, is the future for Linux. The concentration should be on providing the best views of the cloud, combined with just enough local tools to handle important local jobs.

Let’s not worry about Windows and its old way of doing things. Let’s look out our Linux-driven windows to see and manipulate the data we want.