I’ve never read Milton as a naive positivist, although some of the quotations from his article sound that way. Milton was more of a methodological pluralist and a Marshallian. Do what works.
Bring a variety of kinds of evidence to bear on a problem (it is sad
how neglected this principle is in modern economics), be pragmatic, and
don’t reject a model just because it doesn’t meet one of your
prejudices. I think that is what he meant by not judging a model by
its assumptions. Falsifiability is a chimera, but did Friedman really
seek or advocate that as a standard? Most of all he wanted propositions with empirical content. He understood that no single fact can refute a theory, and that many tests are of frameworks, not single propositions. Data should and indeed must be viewed through the lens of theory, and our goal should be to solve problems.
Was Milton closer to Quine than to Carnap? Here is his classic essay.