I remember GGI. As I recall, one of the big problems they had was getting their module into the mainline kernel tree: without users like XFree86, there was no impetus to merge it. And another was getting XFree86 to actually consider switching to using GGI instead of doing their own userspace mode-setting: without mainline kernel support, there was no impetus to merge it.
Thus we had a situation in which neither train can go until the other has passed.
If the GGI project had begun after the big Xodus (sorry!) to X.org, it might have had better success. As it is, it's hardly the first example of a great idea that was just unable to flourish in the environment at the time.