If a substance can reliably produce a strong placebo effect that helps a lot of people with minimal serious side effects, probably just start offering/administering it anyway!
Doing science that answers just that question would be valuable on its own. Knowing why/how it helps would be nice, but if it definitely helps…
If you can reliably solve the problem by doing that then why not.
I guess saying “this is just a sugar pill but it might help” and “this is a psychoactive substance that might make you trip a bit but also seems to measurably help some people though we’re not sure if it’s just a placebo” might have different placebo effects. Or maybe it actually works and it’s just hard to know.
The point is if it measurably works better than other substances, including definite placebos, why not use it?
We should want the outcome with sufficiently high probability and within whatever risk tolerance we define. Not the outcome in a way that satisfies a specific group of scientists. That’s not to criticise the scientists or their science, it’s important to do it, and knowing is better than not, but not knowing everything shouldn’t necessarily stop us if the risk/reward is there.
Doing science that answers just that question would be valuable on its own. Knowing why/how it helps would be nice, but if it definitely helps…