COVID-19 has made evident that we are ill-prepared to respond to an international health emergency, the complex interdependence of social and ecological systems, and that to reduce the risk of future zoonotic pandemics we must safeguard nature. Approaches based on complexity science taking into account that interdependence and its associated systemic risks must be mainstreamed in current policy making, in general. However, at present, that could result in failure for three main reasons: (1) those approaches might be too sophisticated for current policy making pursuing sustainable development; (2) the reductionist views from conventional economics still deeply influence economic and environmental policy making; (3) it is unlikely that far-reaching policies aimed at stimulating post-pandemic economic development can be steered through radically innovative approaches that remain untested. Here, using COVID-19 as an example, I suggest that the use of innovative complexity-based approaches could be enabled through intermediary approaches equipped to resonate with the mindset pervading current policy making. In particular, I propose to understand the response to unexpected systemic threats as instances of reactive policy making driven by radical uncertainty, and advance three notions that could enhance that understanding: modulating contingency, adaptive inference and blue uncertainty.