> domain experts using models to validate hypotheses trump p-value fishers without that domain knowledge every time This basically says that people who have on

> domain experts using models to validate hypotheses trump p-value fishers without that domain knowledge every time This basically says that people who have one subset of the required skills are better than people who have none of them. That's obviously true but I'm not sure why it's relevant to use the worst part of a group ("p-value fishers") in that comparison. The more interesting question is whether skills in statistics, applied math and computer science carry over across different domains - and I think the answer is that they clearly do.