Method
- A thesis before the data. If you do not know what you are trying to disprove, you collect for ever.
- Few interviews, but long ones. Twelve one-hour conversations are worth more than forty of fifteen minutes.
- Report back early, at the risk of being wrong. A hypothesis corrected in week three costs less than a false certainty in week ten.
- Leave behind an owner, a ritual and a vocabulary. The rest is documentation.