This is an area where the compiler and programming language communities often seem to be talking past each other. In the post you linked, Weeks is talking about CPS and SSA as the datastructures manipulated by optimization algorithms. CPS is interesting primarily as a language in the formal sense of assigning semantics to programs. Whether you choose to represent those programs naively as a tree structure or with an efficient graph representation as in SSA, you're still using CPS.

This paper is really good because, unlike a lot of compiler literature, it treats both semantics and term representation.