However it's insane if you consider how many device drivers there are for Windows and Linux. The large majority of Linux kernel code is drivers. When you consider that Linux still has problems to run on certain hardware, then the problem becomes more obvious.

My bet is, in case this ever becomes a success, then only for a subset of vendors that are willing to cooperate closely with Google. (Oh yeah, and everybody needs to throw away the old hardware.)

I wished efforts would instead go into improving Linux Kernel. It reminds me of Google saying that JavaScript is a complete dead end and must be replaced. So they created Dart which surely has a great design but nobody uses it - and JS is now better than ever.

Moreover it's a bit of a lame excuse that vendors don't update their Android modifications. It already start that Google doesn't manage to get their Linux kernel modifications into the mainline Kernel - almost violating OpenSource principles...