it's best to try and find by science the optimal level of each practice (while holding all other aspects in the best-known-so-far range).

Yes. This is the underlying guiding principle. But it begs the question,

What is the best-known-so-far? Clearly the standard American diet is far from it. Clearly less modern diets have benefits over modern ones (and vis a versa!). There are many traditional diets known to have very low levels of diabetes, stroke, and heart disease (the predominant diseases of western civilizations), and which fly in the face of modern nutrition. Why aren't these diets considered the best-known-so-far?

No one is arguing that a paleo mantra should replace science. People are arguing that the currently accepted theories are grossly wrong, and are offering a new hypothesis.