Research policy is a complex matter. Copying best practices in research policy, as identified by benchmarking studies, is popular amongst policy makers but fails because of ‘knowledge asymmetries’. Research fields exhibit distinct knowledge dynamics that respond differently to governance interventions. Extending the idea of search regimes, this paper aims at providing a policy model for different knowledge dynamics by elaborating the notion of knowledge production as a complex adaptive system. Complex regimes emerge from three interacting sources of variance. In our conceptualisation, researchers are the nodes that carry the science system. Research can be considered as geographically situated practices with site specific skills, equipments and tools. The emergent science level refers to the formal communication activities of the knowledge published in journals and books, and announced in conferences. The contextual dynamics refer to the ways in which knowledge production provides resources for social and economic development. This conceptualization allows us to disaggregate knowledge dynamics both in horizontal (field related) and vertical (level related) dimensions by articulating the three different dynamics and their path dependencies (in research, science and society) in co-evolution with each other to produce distinct search regimes in each field. The implication for research governance is that generic measures can sometimes be helpful but there is clear need for disaggregated measures targeting field specific search regimes. Governing knowledge production through disaggregated measures means targeting in a distinct way not only different fields, but also, and more importantly, the interactions between local research practices, emergent scientific landscapes, and the field’s relationship to its societal context. If all three “levels” are aligned, there is a stable regime.
