Given all the intellectual excitement surrounding the new ideas on complexity, it is easy to overlook the fact that the apparent simplicity of the past was often more a function of the constraints put on the framing of the issue or problem at hand, both conceptually and in policy making, than it was a reflection of any inherent properties. Revisiting several case studies helps to illustrate the point that complexity, now or in the past, resides especially in the social relationships within and between institutions and agents. Much current thinking about complexity is moving towards development of ever more sophisticated methodologies with which to probe complex systems, hence to facilitate their management and control. We argue that such methodological elaboration frequently acts as a direct substitute for institutional development and reflexivity, and we urge instead for exploration of new forms of institutional mediation.