Regular collective dynamics of research collaboration

Research collaboration, a ubiquitous phenomenon marking the increasing stratification of science, is hailed for its intellectual and societal benefits. Being an essential element of the economics of science, it has become a matter of scientific inquiry attracting efforts from fields as different as network analysis and bibliometrics. The studies have provided insights into the local structure of collaboration networks, relations between the team composition and research impact. However, the laws governing research collaboration at the (macro)scale of countries and institutions are yet to be established. In this work, a simple yet unexpectedly accurate logarithmic relation between measures of the research output and self-reliance is demonstrated. Restriction of co-authorship analysis to high-ranking journals suppresses geographical and cultural differences to expose universalities in the international research collaboration. The emerging law holds across research areas; it benchmarks collaboration patterns of research institutions grouped by type or country. The self-organized regularity of the collaboration dynamics manifests an important role of national research systems in globalized science, may have profound implications for research policy development and implementation.