Android fragmentation and update lag is a very real problem. It undermines consumer as well as developer confidence in the long-term sustainability of the platform, this despite the fact that Android now occupies the lion’s share of the smartphone market.
The actual solution to the Android update problem has been materializing in the background for the last two years and will almost certainly show its face sometime in productized form in 2012. And that's Android Virtualization.
There's no doubt that fragmentation and non-existent updates for handsets is a serious problem for the Android platform. And there's no doubt that virtualization would help make deploying the Android platform to handsets a lot easier. A virtualized OS would be far offer a far more platform-independent platform than the current offering allows. But there are still serious issues that virtualization doesn't solve.
Virtualization doesn't guarantee forward-compatibility
While virtualization solves a lot of problems related to hardware compatibility, there's no guarantee that a hypervisor installed on a handset today will be compatible with future Android code. It should work for updates within a major version, but for version jumps, problems would still remain.
There's a performance hit
Virtualization comes at a price ... performance. Not much of a performance hit, but it's not a zero-cost solution.
OEMs will still want to add their own branding and crapware
One of the delays in getting Android updates out onto user handsets is that handset OEMs want to add their own branding and crapware to handsets.
Carriers will also want to add branding and crapware
The carriers will also want to have their oar in the update process.
To achieve this, Google may well have to throw all the OEMs under the bus and go the Apple route, leveraging Motorola (assuming that deal doesn't go belly up at some point). If |Google got serious about creating a range of quality Android handsets, other OEMs would be forced to follow suit.