Could be better, could be worse. The implementation definitely needs a bit of love, yet I think this project could turn out to be rather useful. I guess I see it not as a low-level Lisp, but as a Lispy high-level assembly. Applying Lisp macros to assembly generation seems like a pretty natural thing, similar to Forth in a way. Instead of trying to make high-level stuff efficient, just make low-level stuff easy.