International organizations utilize a variety of numerical technologies and practices – among them statistics, indicators, monitoring, evaluating and simulating – to produce social policy knowledge. Based on their claim as global organizations, their knowledge products are similarly supposed to be applicable across countries and thus conceived to be of a universal character. This is an ambitious task, given the huge disparities across the globe in social policy institutions, traditions, as well as problem definitions. The paper analyzes how numerical technologies achieve universality, by deconstructing the 'black box' of their concrete operation. To this end, the paper draws on a conception of the translation of knowledge that scrutinizes how social policy knowledge is transformed by being observed and represented by international organizations’ numerical technologies. ‘Translating’ is not simply a process of knowledge replication, but newly imagines knowledge and is fundamentally dependent on the operating procedure of a technology of observation, as well as the underlying concepts and classificatory frameworks. The theoretical basis of this paper stems from science & technology studies and actor-network theory, as well as newer research on quantification and comparison. Empirically, the paper is based on an ongoing research project on global social policy knowledge in international organizations.
