Today, the Linux desktop is dominated by two design philosophies: the traditional and the experimental. Neither seems ideal for future innovation, especially when compared to the previous generalist tradition.
Innovation may not have been the priority for the generalist design philosophy, but sooner or later its followers got around to considering the possibilities. By contrast, both the current design philosophies constrain the possibilities of innovations due to their innate assumptions.
On the one hand, traditional desktops are undeniably popular. Probably, if you added up the users of each, they would comprise a majority of Linux users, or at least the largest single group. Linux Mint in particular owes much of its popularity to giving average users what they want, while KDE and GNOME seem to have seriously under-estimated the strong conservatism of the user base when they started experimenting.
For one thing, a good deal of the development effort on the traditional desktop has gone into salvaging them. At times, the improvisations have been nothing short of ingenious. Yet the development teams involved are small. So far, they have had few resources to spare for anything else.
Although more resources should become available as traditional desktops complete their initial development, by definition, they can hardly be expected to be the source of innovation. After all, the fact that there was little new that was left to be done on the traditional desktop is a prime reason for the current situation.
At the most, the traditional desktops can be expected to add support for new standards and protocols. But new ways to work? Major new features? If users wanted to experiment, they wouldn't be using traditional desktops in the first place.
Nor can anyone expect traditional desktops to adapt to new form factors. An Xfce for tablets or a Mate for smartphones is a contradiction in terms. Traditional desktops are ideal for workstations and laptops, but they are too specialized to shift elsewhere. While they are not going to disappear any time soon, they amount to an evolutionary dead end.
On the other hand, the experimental desktops have their own limitations for future innovation. The only difference is that, unlike the traditional desktops, their limitations are a result of the assumptions that they have chosen to build upon.
KDE, GNOME 3 and Unity were all built with an awareness of the growing importance of mobile devices. Partly, this awareness was a recognition that mobile interfaces were those most familiar to users.
However, both GNOME 3 and Unity went on to assume that what users wanted was what they knew. If anyone on the development team ever thought to consider that what people tolerated because of the small screens on mobile devices was not necessarily what they preferred, their doubts failed to influence the final design.
Nor does there appear to be any consideration of the possibility that what is acceptable when doing consumer tasks, such as reading email or chatting, which is what tablets are primarily for, is unsuitable for productivity tasks, such as word-processing or coding, which is what workstations and laptops are best-suited for.
Instead, GNOME and Unity ported all the limitations of mobile devices to other form factors. Screen changes, apps that open full-screen -- all the most annoying features of mobile devices became the standard for all devices.
The fact that such design elements were a step backward on workstations and laptops never seems to have been raised. The excitement of change and the convenience of having a single code base across all form factors overwhelmed all other considerations. Both GNOME and Unity have flexibility, but at the cost of reduced usability in many common circumstances.
KDE, by contrast, has avoided this limitation by building modularly. In the KDE 4 series, even the interface is a separate module. Consequently, much of the code can be re-used as it stands for different form factors, while the interface can be rewritten as needed for circumstances.
This arrangement is not as convenient as a single code base. It requires somewhat more recoding. But it does mean that, unlike GNOME and Unity, KDE can optimize the interface for each device.
From the end user's perspective, KDE's arrangement is more likely to be satisfactory, if only because, eventually, users will be able to choose the interface they want regardless of the device they are working on. Already, Plasma Desktop and Plasma Netbook are available on any installation, and within a year, Plasma Active, the mobile interface, is likely to join them.