Even the small payload becomes a big logistical challenge when you're looking at doing it globally, for large numbers of devices and want to make it fast (means having data centers in all regions), and make it reliable (means having redundancy, at multiple levels). Oh, and the "all the data is encrypted" bit may expose regulatory problems, too.
I really want an alternative to Android, but it's an even bigger challenge than I thought.
What specifically are you looking for? As an alternative, are there some ways that Android could be improved to alleviate whatever concerns you have? If your concerns are non-technical and primarily about insufficient ecosystem diversity (i.e. insufficient fragmentation), then there's probably not much to do. If your concerns are related to technical problems with Android, or privacy concerns about its relationship with Google, there may well be.
I'm an engineer on Google's Android Security team, and I'm actively looking for things that we can do to address security and privacy concerns. One of the ideas I've been kicking around is a "pre-encryption network tap"... basically, what if you could turn on a mode that logs a copy of everything your device transmits and receives? Most of that data is (and should be!) encrypted, but since most apps and all system services use the framework implementations (yes, plural... sigh) of SSL/TLS it should be possible to hook in and grab the plaintext. My goal here is to enable users to examine what their device is sending, and to whom, because I think right now it's too hard to tell, and because specifically I think there are a lot of erroneous assumptions that Google is receiving a lot of data from Android devices without user permission.
The downside, of course, is that adding such a hook into the system makes it a prime target for various sorts of attacks. So I don't think we would want to do that, not as stated, anyway. Though there may be some variant of the idea that isn't too risky.
Anyway, given a system like that, it should be possible to build an alternative ecosystem of apps and services that run on Android and don't use Google's infrastructure, and that would be much easier than building an entirely new platform. You'd still need to address the problems I mentioned at the top, but at least those could be the bulk of the challenge rather than just another piece of it.
As another alternative, I think if Google became more transparent about how it manages user data and what it does with it, many peoples' concerns would be addressed. But although I make that argument regularly, I don't have the same degree of influence there as I do over platform technology.
Or if you have other concerns, what are they, and do you have any ideas about how the platform could address them?