What's the point? He'll never beat LuaJIT. I'd rather see a new language with a "slow, slow, and slow" implementation that had interesting semantics than another python/lua/js cut-down LISP.