A preprint would be terrific too.

A tough(?) question and a tougher(?) question: When self-modifying AI's are citizens of Terry Tao's Island of the Blue-Eyed People/AIs, can the AIs trust one another to keep the customs of the Island? On this same AI-island, when the AI's play the Newcomb's Paradox Game, according to the rules of balanced advantage, can the PredictorAIs outwit the ChooserAIs, and still satisfy the island's ProctorAIs?

Questions in this class are tough (as they seem to me), and it is good to see that they are being creatively formalized.