So the analogy with the Krein–Milman theorem would be this: Any proof is (approximately, at least) a combination of the extreme proofs of the same result?

Now I have to go read the paper referred to.