This paper provides a review of the contradictions and conflicts in the literature on economic governance and sketches an approach to use some of the conceptual and empirical findings from that literature for development policy. The literature offers conflicting conclusions on big questions: whether history and geography preordain a country's economic fate, whether democracy or authoritarianism promotes growth; whether informal or formal mechanisms are best; whether "big bang" or gradual transitions promote growth; and whether disasters and demographics are stumbling blocks or stepping stones. The author finds recipes for success that are infeasible, contradictory and shifting, and that ignore the role of luck in development policy. While the researcher may ask, "What creates success on average across countries?" the policymaker needs to know, "What is going wrong in this country and how can we put it right?" The author suggests a preliminary approach to combine the practitioner's detailed knowledge of country conditions with the broader patterns uncovered by scholars, building on "growth diagnostics" that identify binding constraints to development. But he shifts from the sequential "decision tree" framework to a more directly "diagnostic" approach that recognizes that policymakers must deal with many factors simultaneously. The framework he suggests combines empirical information on potential causes, estimates of their probabilities, and observed effects. He proposes this framework as the foundation, not for another recipe, but for a broader mode of thought to tackle the complexity and variance in development processes and patterns across countries and time-one country at a time.