Methodology

Keep enough evidence to be wrong properly.

The method is designed around traceability, repeatability, and a healthy distrust of convenient results.

  1. Frame the question.
    Define the observation, the proposed relationship, and what would count against it.
  2. Preserve the input.
    Retain raw responses, collection times, source details, and the transformation path.
  3. Separate development from evaluation.
    Avoid letting the same evidence invent and validate the idea.
  4. Test the unflattering cases.
    Include delay, missing data, changing conditions, and assumptions that turn out to be false.
  5. Record the result.
    Useful failures deserve the same provenance as attractive charts.