Scientific collaboration, in most cases, is seen a joint action on a global scale that involves researchers from not just one region or one country but instead forming an international network of researchers. This type of epistemic communities build up especially in the case of analytical modes of knowledge production. Rationales for a global science system are needs for complementary ressources in an increasingly specialized world. Further, information and communication technologies contributed to flatten the world for intense, yet spatially distant collaboration. Based on the large scale analysis of the production of scientific publications in six distinct technologies/scientific fields from 2004 to 2008 these notions may be challenged. The probability of a collaboration is a power-law function of the distance between the participating authors: The higher the spatial proximity, the higher the chance of a jointly written paper, i.e. collaboration. Two main universal effects can be isolated: an intra-country effect following a power law with a negative exponent, and an inter-country effect that shows features of a random distributed distance function. The former effect outcompetes the latter by a factor of around 50-100. This behaviour has inevitable consequences for the design of research collaboration programs by governments world-wide.
