"The most important thing in empirical work is to be clear about the question you are trying to answer."

It is remarkable how much long-term advantage people like us have gotten by trying to be consistently not stupid, instead of trying to be very intelligent.