The Book · Chapter 15 of 28

Chapter 15Self-Modification, Constitutional Revision and Their Limits

When does an intelligent system’s revision change its constitutional authority, and what must remain contestable as the system changes?

Summary

How can an intelligent constitutional arrangement learn without acquiring the power to redefine the limits of its own authority? This chapter classifies self-modification by its effects on legal relations, distinguishing operational self-management, delegated rule revision, constitutional amendment and constituent change. A hypothetical public allocation system shows how apparently minor changes to forecasting, eligibility proxies, review thresholds and update permissions can cross different legal boundaries. Comparisons with constitutional entrenchment, normative multiagent systems and formal assurance then expose the limits of immutable files, compliance proofs and nominal vetoes. The argument extends to rollback, institutional identity and the constitutional consequences of replication. Across human-exclusive, participatory, autonomous, hybrid and coexisting settlements, it proposes that lawful adaptation must preserve identifiable authority and effective contestation. Technical improvement supplies evidence of capacity; whether a participant may revise its constitutional position remains a separate question requiring institutional and, where relevant, constituent justification.

• constitutional revision
• normative effect
• entrenchment
• formal assurance
• effective veto
• legal reversal
• institutional identity
• contestable limits