The Hidden Costs of Gatekeeping in Open Science: Quasi-Experimental Evidence from arXiv's Endorsement Policy Tightening

Abstract

Preprint servers and open science platforms have become core infrastructure through which scientific knowledge circulates on the Web, greatly facilitating the dissemination and diffusion of online scientific information. Yet to safeguard content quality, many of these platforms rely on existing members to vouch for newcomers; how such network-mediated gatekeeping reshapes entry barriers across different scientific communities remains systematically underexplored. To address this, I examine the potential unequal impacts of this mechanism through arXiv's tightening of its endorsement policy in early 2026. Using the metadata of three million arXiv papers and the co-authorship networks of each research community, I compare the influx and dynamics of first-time submitters before and after the policy across eight disciplines. The findings show that the policy reduced the entry of new authors, and this reduction was concentrated in fields where co-authorship ties are sparse, while fields with dense collaboration networks were largely spared. A comparison with a parallel preprint server in biology suggests that this likely reflects a shift in where researchers choose to post rather than a decline in research activity. These findings demonstrate that the network structure of a scientific community shapes who can enter online scientific production, advancing our understanding of how the Web's open knowledge infrastructure shapes inequality in the diffusion of science.