And, in this age of desktop environments, how valuable is the WM paradigm, anyways? I realize that Gnome uses several as options, but even they have trimmed down (consistent with their philosophy). A decade ago, when even KDE was still a desktop environment that used window managers, compiz made sense, but it's increasingly a bolt-on in a built-in world. Making it a library might see its demise come sooner, but it also stands a chance of bringing the benefit compiz actually has to all sorts of desktop environments through compiled-in support for 3d effects in the DE itself.
Except for the fact that KDE has gone the built-in-effects-library route for themselves, in a big and usable KDE-internal way, and lobbying freedesktop.org to make it a standard library will just make it a Gnome compile necessity, something Gnome is just *begging* for more of.
(Cue flames from the remaining FVWM users out there... :) )
The free software ecosystem is like any other evolutionary environment, in that it pursues all branches into all niches, and lets the environmental and competitive constraints do the pruning. We've seen it before, and we'll see it again. A project's value proposition is a description of the niche value, not the value of the specific code that occupies it. Competition is for niche occupancy, and maybe the niche is taken, in which case the survival option is to evolve to a niche that is more available.