The story of KDE's user revolt is well-known. What is less well known is that, in the six years since then, KDE has been steadily regaining its user-share.
In fact, for the last few years, KDE has registered as the most popular desktop among experienced users. Despite the obvious limitations of reader polls, they show KDE's popularity consistently well ahead of Xfce, the usual runner up, in2011, 2012, and 2013 in LinuxQuestions' Members Choice Awards and in Linux Journal's 2013 Readers' Choice Awards. Even assuming a wide margin of error in the polls, the consistency makes KDE's recovery hard to question.
What are the secrets behind KDE's comeback? I count at least nine design guidelines, some of which may be unofficial, and others that are at least working guidelines, if not official policy, for the development team.
9. Reform, Not Revolution:
The current KDE release series implements the standard features of a classical desktop in different ways. For example, adding icons is not a matter of using the desktop's context menu, but of adding a Folder View widget to the desktop.
However, despite the change in implementation, the feature of a classical desktop remain, and are not discarded in KDE the way that they are in Unity or the default installation of GNOME. Other major changes, such as Activities, are added on top of the classical features, rather than replacing them. This arrangement allows users to choose whether to ignore or accept changes rather than being forced to accept them.
8. Old Features Remain as Options
KDE does sometimes replace old features with new implementations. For example, Activities might be said to make virtual desktops obsolete.
However, instead of removing virtual desktops, KDE continues to have them co-exist beside Activities. Users have a choice of allowing each Activity to have its own virtual desktop, or of setting virtual desktops so that each can have its own collection of icons and widget and ignoring Activities altogether.
7. Not Designing for the High End
KDE includes Desktop Effects that require 3-D hardware acceleration. However, many effects do not, nor does KDE itself.
At a time when 3-D support is still spotty, especially if you only use free-licensed software, this practice gives KDE a major advantage over other desktops, allowing it to be used without workarounds and/or a major reduction in responsiveness. Which raises the still unanswered question: Why should two-dimensional desktops require three-dimensional support in the first place?
6. Desktop Modularity
Average users may rarely notice, but the KDE 4 release series began as modular, and has become more so with every release.
This modularity has not always been successful. In particular, Akonadi, the personal information manager, has been buggy over many releases, and offers few hints of how to repair it.
However, in other modules such as Plasma, modularity has allowed KDE to plan for different form factors, redesigning interfaces for notebooks and tablets without doing major redesigns of the rest of the desktop.
As a bonus, even when an interface is no longer needed, such as the Plasma Netbook, it can be retired as a template for Activities. At its best, modularity makes KDE easier to adapt to evolving demands and user cases, and to waste as little as possible.
5. Constant Revision
Once the basic feature set was added to the KDE 4 release series, the project began to regularly update the code -- so much so that some of the recent releases have been more about revision than new features. In fact, the KDE 4.12 release was plainly announced as being for applications rather than for the desktop itself.
Such an emphasis on updating can be disappointing to users. However, it promises to make KDE more adaptable. It should also avoid crises like the one approaching for MATE, which will soon have to upgrade from the GTK2 toolkit if it is to continue supporting GNOME applications.
4. Encouraging Clutter
Both GNOME and Unity are designed to limit clutter -- that is, icons on the desktop and applets on the panel.
The only trouble is, this so-called clutter is a major source of customization. Instead of eliminating or reducing this clutter, KDE has encouraged it. Folder View widgets give users the option of maintaining multiple icon collections on the same desktop, or of swapping them in and out as needed. Similarly, widgets place even more on the desktop. Far from reducing clutter, KDE has increased it -- a move that users seem to rank highly.
3. Functional Widgets and Desktop Effects
Widgets, the small utilities added to panels and desktops, are severely limited today. GNOME and Unity have banished all but the most basic system indicators, and even those on MATE and Cinnamon are mostly severely limited. The same is true for desktop effects, which are often no more than eye-candy.
KDE has its share of utilitarian widgets and eye-candy. However, its over 90 widgets and 48 desktop effects also include extremely practical tools.
In particular, exploring the widgets and effects reveals tools like virtual keyboards, mouse-trackers, desktop zooms, and magnifiers that make KDE much more accessible than it is usually given credit for. Other widgets include menu and task manager substitutes, other desktop effects, a screenshot utility, taskbar thumbnails and a tool for use with the window manager. Far from being extras or minor features, KDE's widgets and effects can alter a default installation in significant ways.
2. An Emphasis on Configuration
When KDE 4.0 was released, many users feared that it would have fewer configuration options than earlier versions. However, in subsequent releases, the missing configuration options were slowly added.
Today, KDE retains its reputation for configurability, in marked contrast to GNOME's minimalism. Even with tools like GNOME-tweak to add the configurability tools that GNOME itself fails to provide, KDE still has far more options.
When I questioned experienced users about why they used KDE, the ability to customize was mentioned by almost everyone. In the aftermath of the user revolts, customization has become proof that a desktop is being designed for users, instead of to fit developer's design theories.
1. A Policy to Manage Change
After users reacted strongly to new versions of KDE and GNOME and the introduction of Unity to Ubuntu, innovation has been limited in most desktops to incremental tweaks. The last six years have made KDE cautious, but, instead of avoiding major innovations altogether, the project has opted for slower introductions.
Features have been first implemented in Plasma Active, KDE's mobile interface, and only introduced into the main Plasma desktop after being tested in several releases. Similarly, the new FrameWork 5, which uses Qt 5, has been introduced a few components at a time over several releases.
As Aaron Seigo, Plasma's project lead, told me a couple of years ago, this approach "helps us introduce changes only when they're ready. We find that with the way we've been doing things in the last couple of years, even when we release big things, the response has been a lot more positive. As a community, we’re learning how to introduce changes better.”
A Different Direction
In many ways, KDE's direction in the last six years has been the direct opposite of that most desktops are headed in. While KDE has never made a show of listening to users or of trying to include them in the development process, the project has consistently encouraged further options and customization.
As a result, users may not find that a default installation suits their needs, but adapting KDE to their work-habits is usually far easier than it would be in more desktops. KDE remains popular because it favors a generalist design that can be tweaked almost indefinitely.
At the same time, KDE has clearly learned from the initial reception of KDE 4.0, implementing policies and practices to help prevent a repetition. In doing so, it has managed the difficult task of turning the initial reception around, and achieving success where messy failure once seemed inevitable.
There is no better indication of this success than comparing the state of GNOME and KDE today. Where alternatives to GNOME have proliferated, discontent with KDE has generated only one alternative that registers less than a couple of percent on the user polls.
Similarly, GNOME's changes in approach, such as encouraging extensions, has reversed many of the initial objections to the current release series, but has apparently done little to encourage users to return to it. In comparison, the rising percentages of KDE users seem to indicate that supporters are returning to it. If so, then KDE offers lessons from which other major desktop projects could learn.