The not-so-unchanging desktop

Off the Beat: Bruce Byfield's Blog

Imagine someone who last used a free desktop environment a decade ago. If you sat them down in front of a modern desktop, how long would they take to feel comfortable using it? Probably under ten minutes -- which might lead you to the erroneous conclusion that the desktop hasn't changed much recently.

Admittedly, Unity might take them a little longer. KDE might, too, until they realized that while the organization had changed, the basic features hadn't. But with the other five or six major desktops, they would notice more eye-candy and more consistency in design. They might miss the classic menu whose sub-menus spill out over the desktop, but they'd have little trouble recognizing or using its various replacements. They might notice, too, a few cosmetic changes, such as a preference for toggle switches over check boxes, but little more.

The reason they could become so quickly comfortable is that, in essentials, the free desktop hasn't strayed very far from the desktop metaphor in the last decade. Even modest attempts to extend the metaphor, like KDE Activities, haven't caught on, while slightly more ambitious efforts, like GNOME 3 or Unity, have met strong user resistance. Most of the popular desktop choices -- Cinnamon, LXDE, Mate, Xfce, even KDE -- are based as firmly on the desktop metaphor as their ancestors from 2003 were.

The Design Era
Some would say that the reason that the desktop metaphor remains central to all computing is that it is accurate and powerful. You can't improve on perfection, I'm often told, so the metaphor doesn't need replacement. And maybe there's some truth in that, although I can't help imagining someone 150 years ago explaining on how the horse and cart is the ultimate in transportation, and offering perfectly sensible-sounding rationales, such as it is geared to a human pace and puts the driver in touch with animals.

Besides, outlooks like that become self-fulfilling prophecies. Maybe most efforts to replace the desktop metaphor are going to fail, but, if you don't at least consider them, you guarantee that only minor tinkering will happen.

Anyway, the idea that the desktop is unchanging is only superficially true. To start with, it has moved away from classic UNIX principles in order to provide the same convenience as proprietary desktops. For example, in the first years of the millennium, the idea of automounting external drives was a minor controversy. Many users argued -- for solid security reasons -- that external drives should only be mounted manually by root.

More importantly, the last decade has been the era of design. Prior to 2001, free desktops were too busy adding functionality to worry much about eye-cantrue on a visual leveldy, or even consistency in dialogues. Then Sun Microsystems did the first usability testing on GNOME, prompting the creation of GNOME's Human Interface Guidelines and their occasionally over-zealous application. Ubuntu, with its slogan of Linux for Human Beings encouraged other distributions to start paying attention to finishing touches. On a larger scale, KDE, GNOME 3, and Unity offered reinterpretations of the desktop, all influenced to a degree by the displays on mobile devices.

In fact, rather than being unchanging, the desktop in the last 10-12 years sometimes seems to be about nothing except design innovation, to the extent that -- small projects like elementary OS notwithstanding -- it sometimes seems that the last thing that the average free desktop users wants today is more fiddling with design.

The Coming Thing
The free desktop didn't manage many major innovations while its developers were focused on design. Still, that focus may have been a necessary step to pass through after the rush to add features were over. A lot of routine work remains to be done, but if you dig up old manuals or screen shots and compare them to what you're using today, the improvement is observable at a glance. Despite the failures and the angst, the free desktop of today is cleaner and more usable than its ancestors, even if its improvements are minor.

Which leaves the question: what next?

One possible answer is security. By security, I don't mean the reactive security typified by anti-virus scans and regular updates (which most desktops handle well already). Rather, I mean structural security, which turns away intrusions and malware rather than dealing with them after they are on the system. This is one area in which the modern free desktop has advanced hardly at all in the last decade.

Oh, we have SE Linux and AppArmor. But understanding the mechanics of either is still difficult for many users -- let alone why a few inconveniences are better than having your machine turned into a spambot.  You are far more likely to see people on mailing lists asking how to turn off these features than to learn them.

Similarly, in most distributions, configuring a firewall remains as hard now as ten years ago. Relatively few distributions even bother to include an option to encrypt your home directory.

And what about anonymous browsing and searching?

I realize, of course, that some concepts are hard to simplify. However, the effort has barely been made, so saying that such simplification is impossible seems premature. I remember the same skepticism about design a decade ago.

Security isn't the only practical innovation for the desktop. Developers today are still debating in their designs how apps in the cloud should be integrated in the desktop -- or whether they should replace the desktop altogether. Convergence across form factors, Ubuntu's current dream, might also prove increasingly important, although I don't hear many users that are concerned about it yet. App stores and extensions may also become important over the next few years.

Which of these concerns will become important is impossible to predict. But one thing is clear: the desktop will evolve in the next decade, just as it has in the last. Few developers want to risk radical changes these days and risk another backlash, but the changes will keep coming slow tweak by slow tweak.