I definitely think it will be hard, and I'm also pretty skeptical of the idea that all languages can compile down to Guile. I'd love to see it be true! But thus far, it seems that it's been hard to have that happen.

But I think emacs lisp -> guile (or Parrot, for that matter) may be different: there's only one source that uses emacs lisp, which is emacs, and it seems pretty likely that collaborating upstream with emacs would make this possible.

Given that and how much works, and given the things that don't or aren't ideal, it seems like the hurdles remaining aren't impassable.