You are right that the "largely ineffable" feeling needs to be cashed out. But I use that phrase largely because the psychology of UI interactions and usability that would justify my like of Unity is a new, inexact, and developing science. So, I'm starting with the basic observation that after giving Unity a try, it grew on me, as it did with others, to the point where we actually prefer something new. Part of it is increased simplicity. Another part is more screen space. Another part is that the OS doesn't get in your way (like KDE does). Another part of it is that is the tendency to place things of importance in the upper left which is, in our Western culture, where we tend to start looking for things (like when reading a book).

As far as evolution is concerned, my point is that I often notice, say, when Facebook changes their interface that users complain because they have to learn something new -- they are lazy, but if you force them into something, they may end up preferring it later. It is true that we shouldn't redesign things if they work well. But the way things are designed the first time may not be the best possible way.

The new features in Unity might not be molded by some process like natural selection. But there is some argument to made that it is. Creating one platform for all devices and levels of users would have enormous advantages, and is a response triggered by changed in the computing environment -- mainly, the widespread use of touch phones and tablets.

Aside from that, all I can say is that Ubuntu does do some user testing, and I hope they are listening to their users. They also have to take some chances and try out new things and see what users respond to. But I agree that their interface changes should respond to real needs, not imagined needs.