Traditionally innovation goes like this: Make something new, convince people of its merits, gain converts.
GNOME did this: Make something new, confiscate the old, ignore people who ask for their things back.
I'd probably hate Linux, too, if in 1997 someone had wiped my hard disk and installed it without consulting me and then hid the install discs for my previous OS. You can't *make* people want to change just by changing things.
Want to stay relevant for a long time? Don't make a DE, make a platform. Build the base tools other developers use to make awesome things for users... and then *maintain API and ABI compatibility for as long as possible*. Note that I didn't say "Until you want to do something new and the old cruft gets in the way." Add whatever you like, but maintain it forever. UX fads will come and go; which fad to follow should be the user's choice.