The three experimental desktops also face another challenge: charting their own course, instead of following the time-honored approach of copying what proprietary equivalents are already doing.
In the case of mobile devices, proprietary norms include viewing mobile devices as what Aaron Seigo refers to as "app buckets" -- "places that you dump a bunch of toy apps into."
Seigo goes on to suggest that this profit-centered approach ignores potentially more useful courses of development, such as viewing all form-factors as part of a larger system, in which devices can be easily connected and synced as useful:
It's a sort of reverse tragedy of the commons, where there is no commons, and you have one group that owns the landscape and stands to benefit from it. What happens is that everyone else who would stand to benefit from the commons just stands out in the cold. So, in the case of both iOS and Android, for instance, there is an immense amount of innovation and product development that is simply not happening, because these platforms are not suited to creating devices for vertical markets.
So far, neither GNOME nor KDE has imitated this tendency. However, Canonical has increased efforts to monetize Ubuntu by providing links to online services as part of Unity. The most obvious of these is the inclusion of results from Amazon in online searches, a move that benefits Canonical but is an irrelevant distraction for users trying to be productive. It also opens up major privacy issues.
Assuming that the current situation continues, the future of innovation on the Linux desktop does not look promising.
By their nature, the traditional desktops constrict the amount of innovation they are likely to provide. By contrast, the experimental desktops share design assumptions that inhibit their abilities to innovate in ways that benefit users. Neither alternative is desirable.
So far, the only bright spot seems to be KDE, which has not abandoned the traditional desktop and has implemented or at least analyzed alternatives that avoid the limitations of other experimental desktops. Unfortunately, though, KDE is not to everyone's taste, and its perceptions have been shoved out of the spotlight by Canonical's aggressive commercialism.
Meanwhile, whether GNOME can produce a useful alternative remains to be seen.
Perhaps a new desktop with new priorities will become popular and change this situation. As events of the last two years have shown, such a development is perfectly possible.
Yet for now, the irony is that, having reached a stage where developers can seriously consider innovation, the Linux desktop has become a victim of its own recent history and is less able to innovate usefully than ever before.