Beneath their skins, the operating systems are remarkably similar. Given the same hardware platforms, the same jobs to do and many of the same applications, it is no wonder they have converged. There is no job any of them can do that the others cannot, beyond the limitations set by marketing, rather than technical, considerations.
Such stability and commonality are a boon for users, who can switch between platforms and be productive with a minimum of fuss. But it is a poor environment for genuine innovation: when marketing is the primary differentiator, we should expect it to be the area that gets most attention. True technological advances are harder to justify.
Which wouldn't be so bad, were there no real problems left to fix on the desktop. Yet there are, and they need to be fixed. They are best described as the four 'tees' — usability, reliability, mobility and security.
Usability is the easiest, and hardest, thing to reinvent. It's easy, because it is the most human aspect of computing and, thus, most capable of absorbing good ideas from many places. It's hard, because it requires huge imagination and creativity to shift it away from the windowed desktop metaphor that is now so far past its prime.
Reliability means rethinking the cloud. It's becoming clear that the cloud will never be entirely reliable, for the simple reason that connectivity inside and out with the cloud is never entirely reliable. At the same time, the cloud is a great place to keep data: intelligent, invisible, inclusive mirroring of data and functions on client and in cloud gets the best of both worlds. That's an OS function.
Mobility continues to ask more questions than it answers. Data and applications do not want to be constrained to the desktop, but efforts to create seamless experiences for the whole range of displays — from phone to projector — have largely failed. That it is technically possible for the same code to run anywhere does not make it desirable: the next OS will have to morph from platform to platform.
That leaves security. The desktop OS is the place where most of us encounter and control our valuable data, yet it remains locked in a perplexing, complex and fallible set of conflicting requirements. The default is to get security wrong by making it hard. It, too, should be invisible, intelligent and inclusive.
All of the above need radical thought, which boils down to a simple idea: where things are not invisible, they need to be vastly more usable. The winner of the next OS war will be the design that gets this right. It remains an entirely open race: Windows, OS X and Ubuntu aren't even at the starting line.