There are a variety of kernel issues (think wireless drivers and other hardware support) that have a major impact on the userland experience. I'm not about to say where Canonical should invest their time -- there are more than enough issues to go around, and it isn't shameful for them to concentrate elsewhere as the GP implied -- but what happens with kernel development certainly impacts the Ubuntu userland.
While your premise is true, the implication that Canonical should contribute in a greater way towards direct kernel development is misplaced.
In classic Slashdot fashion, I present a car analogy:
Goodyear has built a company out of making tires. While the quality of the roads those tires run on effect the performance of their tires, contributing directly to the development of improved road surfaces is outside of Goodyear's core competencies. A tire company should focus on making better tires.
In a similar vein, Canonical is doing the right thing by focusing on their core competency - improving the userland experience. They should continue with that focus where they've already established competency.
I'd also suggest that Canonical is contributing greatly to kernel development thru secondary channels - broadening the desktop-centric Linux userbase. With a more broad userbase, everything get tested and improved at a faster pace. Canonical is doing this for the desktop where Linux has normally struggled in comparison to server deployments.