I've heard him say this in multiple interviews, so here it is for posterity. He shares these principles:
1. Seek out negative feedback. Ask your friends NOT to tell you what they like about your product, but rather what they don't like. Underweight positive feedback, and overweight the negative.
2. Reason from first principles rather than by analogy. It's easier to compare a situation or a problem to something similar to it and be satisfied by the analogy. It's much harder to actually think and reason about it logically as you would a physics problem.