Code decides facts, the judge decides reading

Two kinds of question, one fixed order

Code runs first ยท the judge answers only what code could not prove

Code checksfacts it can read directly

  • Files
  • Commits
  • Deploys
  • The live page
  • Test runs
  • Who wrote what

The decisionthe same order, every time

  1. A block beats a note
  2. A code fact beats a judgement
  3. One voice per turn end

Judge (Jev)questions that need reading

  • The owner's words
  • Does the proof show the work?
  • Can this question be answered on one read?
  • Does this reply promise something with nothing behind it?

When the two disagree on a fact, code wins. The judge is asked only what needs reading.