let the FSF create a GPLv4 that alienates enough developers and I'll bet that in about 6 months there will be a new libc available.
remember, even the libc development is nor primarily done by the FSF, and hasn't been for years.
we came close to finding this out with the GPLv3, but the FSF backed down and revised the drafts to make it more acceptable to people
there isn't any piece of the system that couldn't be replaced relativly quickly by something else if the developers of that piece were to go nuts (and this includes the kernel, remember that debian is shipping a BSD kernel based version)
if you look at any component of your system, someone can name a alternate piece of software to use. it may not be quite as good right now (remember, different people have different definitions of 'good'), but if there was pressure to fix it quickly the rate of improvement would be staggering.
and that's ignoring the possibility of a fork taking place (like happened with xfree86/x.org)
probably the hardest component to replace in the short term would be X, but with wayland in the wings, even that is being challenged.