In software development you can’t truly say you’ve fixed a bug unless you can reliably recreate it before the fix and can’t recreate it afterward. Can this be applied to social problems? Is there a good way to identify the “bugs” in the system, reliably recreate them, and then solve them?