Google outlines plans for mainline Linux kernel support in Android
arstechnica.com
arstechnica.com
2. Is it time to discard backward compatibility and transition to a mainline kernel for future versions of Android? Apple does this every few years, with great technical and user success without much of an uproar
3. Is it time for kernel devs to acknowledge the reality on closed source drivers, and standardize an interface in the mainline kernel for the same. Those who want to open source can, but those who don't find a way to retain source even today, possibly at the cost of performance (forcing code into user space). Let's just allow an LGPL interface with looking exception for well known interfaces.
4. If Linux is unwilling to acknowledge industry trends, perhaps switching to a BSD kernel should also be evaluated. Oh the things BSD could do if it had the kind of funding that Linux does
People like the fact that they can install Linux on any old computer and it'll "just work". Linux supporting a stable in-kernel ABI means less drivers will get upstreamed plus they'll have to bear the burden of maintaining this stable interface across many years. Which is unrealistic, so your older devices will be stuck using an old kernel (which unsurprisingly is the same problem Android has).
If Google wants make a stable in-kernel ABI then that can be part of their Android Common Kernel and they can maintain it themselves. But don't ask Linux devs to do all the work for vendors who refuse to open-source their drivers.
What is Fuchsia's advantage over BSD? If anything, they have opinions on first-class language support for the user space (Dart). Do you dart?