Quantifying scientific collaboration

Research collaboration is increasingly common in today’s scientific world. The practice grows out of colocation, mentoring, complementarity, and other factors. Much of the “science of science” research has focused on collaboration patterns that emerge from data aggregated over individuals, time, and disciplines. Alexander Petersen of the IMT Institute for Advanced Studies in Lucca, Italy, has now analyzed collaborations, as manifested in publication and citation records, with a focus on the individual researcher. He considered the career data of 473 scientists, 193 from biology and 280 from physics. For a given researcher, Petersen quantified the tie to each coauthor based on duration (the time between first and last coauthored papers), strength (the number of coauthored papers), and impact (the number of citations to coauthored papers). The evolution of each researcher’s network reflects specific career events such as relocations, discoveries, and major prizes. Petersen found that in both biology and physics, collaboration networks are dominated by weak, short-lived connections, especially late in one’s career. But his analysis also revealed the existence of “super ties”: extremely strong, long-duration collaborations that lead to above-average productivity and citation rates. Such super ties—the research analogue of a life partner—are thus a major factor in career development. Biologists tend to have more super ties than do physicists, yet the frequency of super ties for both groups is surprisingly high: typically one for every 25 coauthors, independent of a researcher’s prestige or overall productivity. (A. M. Petersen, Proc. Natl. Acad. Sci. USA 112, E4671, 2015, doi:10.1073/pnas.1501444112 .)