It's precisely the features-level thinking that gets in the way of a usable Linux desktop.
How many more reimplementations of basic functionality on the major desktop environments do we need? It seems like whatever the current window manager is, and however well it works, it'll be thrown out and a new one will take it's place, and take a few years to do what the old one did. We'll throw out the audio stack again at some point. X isn't good enough, so we'll throw that out. Since Wayland isn't network transparent, there will be twelve different ways to add it back in and five incompatible ways to run old X applications on top of it. When those aren't sufficient, we'll throw it out again and start over, ad nauseum. And this applies to most things on the system that are "good enough".
Linux on the desktop is stuck in the 90% category, where 90% is implemented (and as we know, this represents 10% of the work). It's always tempting to avoid the unsexy 90% of the work and write from scratch. The end result is throwing increasing amounts of the baby out with the bathwater.
This is why I fully support what's going on with the ubuntu desktop. While I love the open source model, I think we need more cohesive top down decisions if we want to be able to use a desktop with a cohesive feel.
I really feel natty is heading in the right direction.
I have no issues with reliability or usability on my current Linux desktop.