I don't see how this may be relevant to the usefulness of microdosing.
- Assume that a microdose either improves an outcome, or does nothing. Then, averaged over a large group, such random improvements would lift the group's outcome a little. It's a microdose worth taking.
- Assume that a microdose can both improve and worsen the outcome, or have no effect. If the experimental group's averaged results are indistinguishable from the control group's averaged results, it means that the microdose worsens the outcome about as often and / or as much as it improves it. This, to me. means that a microdose is a gamble not worth taking.
There is, of course, a difference between a microdose having no effect at all, or having an effect which can be either positive or negative. This difference is important for further research. For usage here and now, it's sadly irrelevant.
If a microdose only works when the person taking it knows it works, then it's basically the placebo effect. A good placebo can be useful, at least for commercial purposes.