It's a sensible thought. I believe that the desktop operating system is (at least currently) a dead-end. It is no longer a place where any kind of interesting platform warring can occur. All desktop operating systems must offer similar resources (access to the capabilities of the underlying hardware, file-system, graphic user-interface including components, media handling etc. etc.)
Until the hardware changes or human requirements change fundamentally, these are more or less static; commoditized.
Windows, Mac and Linux all do a sufficiently OK job of providing for these requirements that the choice of one or the other offers little advantage to the application writer. One chooses to write for Windows because that's where the biggest market is, or Mac because that's where the most exclusive market is, or Linux because of ideological commitment.
Differentiating desktop operating systems boils down to tweaking the inessentials.
This is why Microsoft's huge investment of time, money and goodwill in Vista was a strategic mistake. One which cost the company far more than a desktop OS could ever recover.
Instead, the competitive action has moved to new loci. As I've mentioned on my other blog, there is a dramatic upheaval going on in the software world :
Applications are getting pulverized, fragmented down into smaller, more focussed, single "feature" mini-applications that I tend to call "widgets". This is happening because there are new networks for organizing and plugging the widgets together. Widgets are tethered through web protocols, RSS etc to server based applications in the "cloud" but are only distantly connected to each other.
Increasingly they are running on virtual machines which have successfully wrapped and hidden the operating system. (The browser, the Java Virtual Machine, the Flash virtual machine etc.) One thing that has helped with this is that the web-based applications store user's data in the cloud not on the local machine, and so avoid having to get too familiar with the local file-system.
This software is often self-installing or comes directly from the web. It usually needs no purchasing and relies on no infrastructure of distribution (such as shops selling boxes with DVDs in) It is promoted by word-of-mouse. The extreme example, which I believe is a pointer to the future, is the Facebook app. where knowledge of applications percolates virally through computer-aided-social-networks (YASNS) who's infrastructure helps accelerate their flow. (Google using GMail to push people into it's online wordprocessor and spreadsheet is another example.)
Widgets, then, are small, narrow focussed programs which live natively in both social networks, and data-flow networks, tethered to the cloud.
In general they eschew dependency on the desktop operating system and prefer to run in the browser, JVM, Flash (and maybe Silverlight) virtual machines. By their very existence they contributed to the decline in relevance of the desktop operating system but reignite a platform war among widget containers. There is a lot of room for differentiation in the container : who provides the best video decompression or GUI toolkit, for example.
The other coming platform war is going to be in the cloud. We are seeing the shift from companies providing basic hosting mechanics (servers, database servers) to Amazon providing a suite of cheap, scalable resources for a web-based applications, to Google coming with the beginnings of a complete, integrated development/deployment/hosting platform with Application Engine.
As Folknology puts it in a tweet today :
GAE marks the end of frameworks & the beginning of Platform 2.0. Expect this to be a hum-dimmer of a war between the big players.
Over the next five years there's going to be far more happening on both the integrated cloud development and hosting front, and on the widget-container front than on the desktop. At which point it must start to dawn on Microsoft that they can't afford to waste as much on Windows 7 as they did on Vista. It's not clear that there's any need at all for a new Windows apart from for appearances sake. And to stay in the game.
It's certainly no longer viable to keep the next Windows as the centre of attention or try to fight for world-domination there.
Windows 7 can't be the monolithic, all conquering beast that its ancestors were. Instead, far better to be a swarm of components. Cheap to produce, loosely coupled, a buffet from which users can pick and choose.