Learning, Legitimacy, and Regeneration: Completing the Architecture of Institutional Adaptation

Institutional architecture has been developed across several distinct objects — capital, authority, legitimacy, memory, commitment. Each has been analysed as a separate failure domain with separate remedies. This paper argues that the resulting picture is incomplete in a specific way: without learning, none of these architectures can correct itself. We advance a structural claim. Among the operators available to institutions, learning is the only one that actively reduces misalignment over time. Regeneration renews capacity without correcting models. Alignment preserves intent without reinterpreting it. Legitimacy manages acceptance without generating truth. Only learning revises the internal representations on which every other correction depends — which makes it not one architecture among several but the one that governs whether the others can be repaired. We then trace the failure cascade that follows its absence, from update suppression through model drift to the substitution of crisis for correction, and close with the framework's limits: it is architecture rather than policy, it does not eliminate conflict, its measurement remains unresolved, and its transition costs are political rather than technical.