Writing

  1. What was required, what was judged, and who says so
  2. A receipt that names who signed it
  3. A delivery receipt a skeptic can check
  4. A baseline two parties can name
  5. Self-Certification is not verification
  6. Asking the record
  7. When the benchmark was broken
  8. A verifiable receipt for your best-of-N run
  9. How do you keep an AI agent's decisions honest after the evidence changes?
  10. How do you verify what an AI agent did without trusting it?