That would be harder. I shuffled data into a new form; they wrote papers. All I had to understand was what they wanted, and how to get there; they had to understand enough of the world to argue that a given natural experiment showed drug X was an effective treatment for condition Y.