Recent years have seen policies of ‘scientific development’ develop in various countries. These policies aim mainly at differentiating the means allocated to universities (or other institutions) based on ‘diagnoses’ and assessments rooted in beliefs concerning the spatial dimension of higher education activities and research. These representations may be regarded as ‘commonly held beliefs’ governed by the idea of an inevitable increase in hierarchical differentiations between cities, the existence of ‘critical mass’ effects imposed by a strengthening globalization, and ‘competitive’ scientific activity. Based on bibliometric research, our results show that those beliefs are often wrong. Though scientific activity is indeed highly centralized, the current trend is towards diversification and de-concentration rather than towards a reinforcement of the most important centers. The spatial concentration of researchers has no specific effect on their individual productivity. National contexts are not fading; they are merely being combined with the growth of international collaborations in a global context characterized by the decline of publications signed by a single person or a single team.