Probably because the details of S-expressions differ underneath the hood - a simple (foo ()) can mean a different thing on CL (where the second element is the symbol NIL) than on Scheme (where it is an empty list and not a symbol); also because Clojure introduces additional bracketing patterns, such as [] for arrays and {} for maps, which will confuse non-Clojures. And let's not talk about sexp variants of Lisp-Flavored Erlang.

Also, if you squint hard enough, https://en.wikipedia.org/wiki/Canonical_S-expressions works.

Also I have no idea if anyone ever cared enough to implement cross-language software transactional memory that would work across a process boundary, to close the gap between a Clojure process and a Lisp process. It could work in theory, but someone would need to do it - either get paid to do it, or have enough mojo of their own to write and bulletprove such a thing.

And, since everyone serializes to JSON anyway, serializing to JSON is a decent enough choice for communicating across Lisps as well.