The draft approach paper is a very rich sum of experience, lessons learned, and new proposals, and the comments on this blog page already present a wealth of points and counterpoints. So I will just focus on three points. The emphasis on diagnostics is well taken but should not be misinterpreted as “whatever works” http://en.wikipedia.org/wiki/Whatever_Works. A doctor that is diagnosing a patient has in the back of her mind a range of theories that are connecting symptoms to medicines. It is therefore important that the research agenda of the PSM approach paper includes further elaboration of these “theories”, perhaps more correctly named “contextual guesses”, and what each of them means in terms of interventions. It is fair to worry that such categorization could bring us back to “best practices”, but this is unavoidable. This way of looking at diagnostics is also important so that we can learn over time. If each diagnostics is based on a range of ultra-specific contextual factors, the sum of intervention case studies will never add up to any “lesson learned”. Second point, on political economy. I think Richard, Willy and Sanjay have thrown a very healthy debate in the discussion, which Alice takes up nicely. It seems to me a fair critic to challenge that we can operationalize (or that we ever have) findings from political economy analysis (although failing by ignorance does not seem more attractive than failing by incapacity to influence). I think that where political economy could help would-be reformers, and their advisors, is in helping formulating a realistic theory of change (cf. also point Nicola makes). Most PSM interventions are predicated on some implicit change process – forcing this process to be explicit and tested against basic political economy (at least ex ante) would seem to be progress. A basic feature of such theory of change would be to set some realistic expectations on timing (cf. nice post by Michael Woolcock on the difference between an oak tree and a sunflower http://blogs.worldbank.org/impactevaluations/guest-post-michael-woolcock-on-the-importance-of-time-and-trajectories-in-understanding-project-effe). Another area where political economy could be directed is the issue of demand for PSM reforms. This links back to the issue of connecting PSM to sectors – service delivery or regulatory functions. Who is supporting change? What kind of change is in demand and how far are stakeholders willing to push for it? As Nick notes, PSM is difficult because it requires many thousands of public agents to alter their behavior: so if demand is not strong, it will fail. Third point, on results. Again, a welcome addition. This should really push the envelope in terms of measurement. As the “former sector manager turned academic” noted, “public pay and employment” seems basic. But they are a range of other functions where comparative data would help diagnostics, research, and focus on results.