Abstract This contribution aims at an original comparison of development analysis with Elinor Ostrom and Esther Duflo from a methodological standpoint, scrutinising their relationship to theory and their operative research strategies. Both perspectives are investigated as case studies for a broader discussion about significant trends in economics and social sciences. Duflo and the J-PAL's approach illustrates – in its own way – new trends and some blind alleys in contemporary forms of mainstream economics, whereas Ostrom and the Bloomington school point towards the marked theoretical and methodological reflexivity of institutionalism, its sensitivity to historical diversity and openness towards social sciences. Distinct social philosophies and episteme are at stake displaying a great divide between two brands of realism and pragmatism, two relationships to development, expertise and knowledge. The paper also contrasts Duflo's methodological monism and mechanistic piecemeal analysis with Ostrom's methodological pluralism and adaptive complex systems analysis.