Participatory processes for water management are expected to induce learning and could impact on the capabilities' (ala Sen or Nussbaum) distribution among participants, driving sustainability and equity. However it cannot be measured reliably and repeatedly, for diverse conditions. This research aims at proposing, testing on real cases and improving a new impact assessment method, based on a social experiment apparatus, complementary to classical surveys and observations. Such controlled participatory simulations should elicit changes (learning and redistribution) in knowledge, relationships and water-related practices. The goal is to get a lightweight micro-experiment useful both for the users' group reflexive capacity, and for external stakeholders to qualify the process and impacts. Comparing with classical evaluation methods could feed the research on social experiment for water management.