> You just fundamentally cannot put everything on the web. It works for incredibly simple things. But that's it.
The web isn't everything - no perfect solution exists for every problem. But you go way too far in the other direction. Modern web browsers can run very complex code at very high speeds. Saying the web can only do incredibly simple things is like saying Java can only do incredibly simply things (which people did say, several years ago).
> What would interest me more in B2G is whether the Linux base underneath it is fully functional. And not just a slightly advanced "bootloader" for B2G. A fully functional, mostly free GNU/Linux system that gets widespread adoption, even just as a smartphone OS, would be a great blessing to the efforts of the Free Software community.
The problem is that such an OS cannot easily get widespread adoption, because it would have no apps. That is the problem of every new phone OS today, people develop for iOS and Android, and that's it.
That's what makes it so hard for Bada and Windows Phone and WebOS to get any traction - developer attention is all on the top two mobile OSes, and users buy those OSes to get those apps.
So inventing a new development platform is incredibly hard, and that's what using Linux as a development platform would get us. The only real way to get around that is to leverage an existing development platform. The best existing such platform is the web - it's open, has multiple compatible implementations, is royalty-free, and a lot of effort is going into it (look at recent advances in web browsers).
That's why B2G, WebOS and Tizen are all using the web as their development platform. It's the only approach that has a chance.
They are all also using Linux underneath - it's the best OS for that. But it's not the best app development platform.