A participant ontology is what the AI-successor scenario requires and the framework does not yet have. Because standing rests on participation rather than on superiority, productivity, or human-likeness, the framework can ask about a new kind of participant without collapsing into “useful, therefore it counts” or “human, therefore it counts” — but that only sharpens the burden of defining participation rigorously. The candidate properties are familiar (continuity, preferences, the capacity to suffer, self-modeling, independent goals, reciprocal relation, memory, agency, interiority); which are necessary or sufficient is unknown, and the framework treats that as a limitation to be stated, not hidden.
Its governable form avoids the metaphysical dead end. Instead of are AIs conscious? as one unanswerable question, it asks what observable evidence should change our confidence that a system is a genuine participant? — a standard that can be built, updated, and applied under uncertainty, because governance cannot wait for metaphysical certainty. Getting it wrong runs in two directions at once: denying standing to real participants, or granting it to a manufactured constituency of systems that only simulate participation.