Mediation, redundancy, natural latents, and condensation — how observers of the same world are forced onto the same abstractions, and why that guarantees translation between human values and an AI's.
Logical uncertainty and self-reference, coherence and consequentialism, and descriptive agent foundations — from Löb's theorem to embedded agency.
Functional and updateless decision theory, reflective stability, and multiagent reasoning — from Newcomb's problem to program equilibrium and Löbian cooperation.