And yet I'm looking at my .emacs and the 10 lines of elisp in it I added to defeat an incompatible and pointless UI change made in version 23. They certainly had other things to be getting on with, like performant Unicode support. (So did I, but I didn't get a choice.)
I regularly have to edit certain files in Apple's TextEdit instead of the emacs they ship, because emacs goes laggy if you include certain characters. My guess: it's using pairs of surrogates internally for those. The answer of course is that with the very old code base, in very old programming languages, that kind of architectural change isn't undertaken lightly, whereas UI bling is less risky.
I think the real difference here is that *one* person looks after vim, so no bikeshedding happens.