> But what's the alternative to looking at examples of success?
Controlled experiments and/or causal analysis.
Controlled experiments and/or causal analysis.
Does statistical causal analysis work when there are this many variables? Feels like that would just be essentially a case study again.
Yes, that's the whole point. It works when you can "simulate" experiments by looking at a large number of events.
E.g. DNA evolution in genetics, drug testing in medicine, studying failure modes in aeronautics...