From Learning to Execution: Completing the Architecture of Institutional Action

Institutions fail to act on what they know. This is usually attributed to political constraint, resource scarcity or weak leadership — explanations that identify obstacles without explaining why the obstacles are encountered at the same points across otherwise unrelated institutions. This paper argues that institutional action is a path, not a handoff, and that it traverses four architectures: learning, which determines whether an institution can know; commitment, which determines whether what it promises constrains it; enforcement, which determines whether the constraint has teeth; and regeneration, which determines whether the resulting capacity survives. Failure enters at the seams between these, not inside them — and we identify four such decouplings, each with a distinct signature and each producing a characteristic cascade. The practical implication is a shift from outcome evaluation to structural diagnosis. An institution's capacity to act is determinable from the integrity of its seams, in advance of the outcome, and this explains a persistent puzzle: why more data so rarely fixes action failure.