Eh, the replication crisis shows that the "methodological discipline" of RCTs is really not all that strong. But they're still better than nothing. I think you've set up a false dichotomy.

If the effect size for a given treatment was small, I would expect it to be outcompeted by other treatments with larger effect sizes. Practitioners who are disciplined in their thinking will make better discoveries, can treat patients more effectively, and will have more business, thereby training more apprentices in their methods. Over hundreds of years, it's easy to see how this process finds treatments with relatively large effect sizes.