The landscape of science is changing radically. In particular, there is increasing heterogeneity of actors, research sites, knowledge and networks. Science policy-makers have to respond to these changes, but heterogeneity makes it more difficult to impose own goals on the research system. Yet, if the dynamics in and of the system are understood, other policy approaches become possible. Here two important systemic aspects of research systems — ‘steering’ (the extent to which the system is sensitive to attempts of a principal, generally the state, to implement own objectives) and ‘aggregation’ (the organisation of processes of agenda-building within the system) are introduced. It is argued that, because of the changes in science, a post-modern research system is both possible and desirable in which aggregation is favoured over steering.