I would prefer random trials compared to "guessing" the action. Our understanding of human body is still very primitive at best and we just don't know the mecha

I would prefer random trials compared to "guessing" the action. Our understanding of human body is still very primitive at best and we just don't know the mechanism of action of many frequent things like headache or hangover, where we could get data from millions of people. Even in pharmacy, the mechanism of action runs backward. We get many candidates for drugs, do animal testing, and see if something changed in them, and if it works, we later try to come up with some mechanism, which a lot of time in history was completely wrong.