A team focused on researching support for new types of hardware devices should be started.
I find that linux developers in general always tend to wait for announced hardware, or until hardware becomes popular before adding general support for it. The issue is, that because most companies wont announce the existance of a new type of product until close to release date, we are always behind in support, because we never even have the infrastructure for it.
Some examples of technologies we should design an infrastructure for now are: Android mobiles, Multitouch, 3D printing, acellerometers and 3D projection. And there are plenty more which we should at least start coding proper libs for, so it doesn't break down into a bunch of adhoc drivers, and utilities that aren't universal.
Many people say "we should work on more important things". Like what? Making gnome look prettier? Thats all great, until manufacturers realise we don't have the infrastructure for the hardware they were thinking of designing, so they wont bother with us, because it will take months before programs could use it even with drivers (so they may as well just focus on the most popular platform).
Lets jump the gun, and start being the best at supporting NEWER hardware, and not just older hardware. If we don't do this, hardware in linux will never be truly "plug and play", instead, it will remain "plug and pray".
If we have an awesome infrastructure for cutting edge technologies, we will have something to brag about, and software developers will see linux as the best at supporting technologies in a generic manner.