Mako gave a talk a few years back about what makes open source/free software projects successful. One aspect is that it's VASTLY easier to recruit volunteers if what you are building is something that lots of people have already used, i.e. a clone of an existing project. Another aspect he doesn't touch on, but David Eaves does, is collaboration vs. cooperation - basically, it's VASTLY easier to get useful volunteer effort for a project if the abstraction boundaries are crystal clear and person A can build subproject A, person B can build subproject B, and nothing that person A does will block person B from being productive.

GNU as a UNIX clone has the above powerful properties.

GNU as a LISP Machine clone does not.

Therefore IMHO GNU as a LISP Machine clone would have failed to be built.