Begin with whether mistakes can still be corrected, connecting foundational logic, future intelligence, real cases and evidential status. Readers need not consume all the theory at once; continue by question.
Begin with whether mistakes can still be corrected: why correction after the fact may come too late in an era of high consequences, and which four questions ordinary people can ask today.
Protocols still matter. What must endure through time, however, is not one eternal protocol but the capacity to regenerate correction and cooperation after old actors, equilibria and rules fail.
Separate subject qualification, understanding of time, reopening purposes, irreversible closure, meta-rules and permission discipline into questions that can be precisely challenged.
Place protocols, time, changing power and the regeneration of correction within a real high-consequence window. Strictly distinguish political commitments, operational rules, actual use and effects.
Use Canonical deduplication and F/I/S/U discipline to manage evidence of AI-agent boundary behavior: distinguish real external events, controlled capabilities, environmental properties, inference and unknowns, allowing new evidence to upgrade or downgrade claims.
This is a translation of the existing public web page. A description of a work is not its full manuscript. Source publication and version labels describe the Chinese work; they do not claim an English full-text edition is available. Unavailable English texts and downloads are marked explicitly.