The growing importance of citation-based bibliometric indicators in shaping the prospects of academic careers incentivizes scientists to boost the numbers of citations they receive. Whereas the exploitation of self-citations has been extensively documented, “higher order” manipulation strategies of bibliometric indicators have not yet been studied.
We study the propensity to exchange citations between pairs of authors in a large citation network (American Physical Society dataset), and assess its statistical significance by means of a suitable null network model ensemble. We represent authors as nodes in a directed weighted network, where a link A → B of weight w signifies that author A has cited w times author B over a certain period of time, and measure the propensity to exchange citations in terms of weighted network reciprocity.
We find evidence of a steady increase, starting from the 1950s, of the excess reciprocity with respect to our null model, which we interpret as an echo of the academic community’s collective response to the increasing importance of citations and bibliometric indicators. Furthermore, we find this phenomenon to be largely fueled by an increase in the exchange of citations between past coauthors. Yet, at the level of individual authors, we find that citation strategies based on reciprocity do not provide any competitive advantage in terms of success. Quite to the contrary, we find long-term career success to be negatively correlated with reciprocity. We elaborate on these findings and discuss their implications to academic decision-making.