What's wrong with it? If you can't run a controlled trial, then it's hard to finger causality. You can consider parallel natural experiments as occur in differe

What's wrong with it? If you can't run a controlled trial, then it's hard to finger causality. You can consider parallel natural experiments as occur in different polities, but there are so often many confounding factors. You can grow more confident in a strategy as the stats pile up, but it's not nearly as cut and dried as a physics experiment.