Sometimes change is good. Just because we do something one particular way (in the world of desktop environments this is often down to a decision made in the early 90s) doesn't mean that it's the most productive method of doing it. Often initially unpopular changes in user interface prove to be more effective.
The problem is that change for changes sake is different to change for UI improvements sake, but the two are often hard to distinguish between. How do we know that a new UI implementation is better or worse than the old one? You don't really until it's being used in the real world. There's not an easy answer, but I don't think that grinding innovation to a halt is the right one. I guess the beauty of the open source world is that there's room for a minimal non-progressive UI as well as a number of innovative ones.