Jeremy Howard (of FastAI) is doing a major metastudy on masks at the moment. Pretty sure they work.
This is a vacuous statement in this context and adds nothing to the discussion, because you can never prove causation via a study/experiment, only reject the null hypothesis.