In my view, everything mentioned in there was good but nothing was ground-breaking and revolutionary.

The smart thing they are doing is doing the bulk of their work upstream. This works well for nitty-gritty things like making gtk/cairo faster or making the gcc arm compiler better, etc. However, they do have an end-user face too and that is where there seems to continue to be an expectation mis-match between the Nokia team expecting to throw a finished immutable infrastructure over the wall and the end-users and application developers wanting to be able to rip out/swap out/upgrade/tailor some of that infrastructure easily.

It seems like they don't really mind if devs break out on their own with the 770 and want to rebuild the firmware from scratch with some community-driven project but they are unwilling to give guidance for fear of absorbing some liability. I have a lot od sympathy for them as it is a fine line to walk when you are trying to provide something for both mass consumers that expect a 110% polished product and Linux hackers that don't mind the odd wire hanging out of the box provided they can clip onto it and read the output on a 'scope.