This article argues that there is a paradox at the heart of Hall's “Policy Paradigms” framework stemming from the desire to see both state and society as generative of social learning while employing two different logics to explain how such learning takes place: what I term the “Bayesian” and “constructivist” versions of the policy paradigms causal story. This creates a paradox as both logics cannot be simultaneously true. However, it is a generative paradox insofar as the power of the policy paradigms framework emerges, in part, from this attempt to straddle these distinct positions, producing an argument that is greater than the sum of its parts. In the second part of the article, I discuss the recent global financial crisis, an area where we should see third-order change, but we do no not. That we do not strengthens the case for the constructivist causal story.