Quantifying the growth of preprint services hosted by the Center for Open Science

A wide range of disciplines are building preprint services — cyberinfrastructure that enables publishing non peer-reviewed scholarly manuscripts before publication in a peer-reviewed journal. We have quantitatively surveyed nine of the largest English language preprint services offered by the Center for Open Science (COS) and available through the COS Application Programming Interface. All of the services we investigate also permit the submission of postprints, non-typeset versions of peer-reviewed manuscripts. Data indicates that all services are growing, but with submission rates below more mature services (e.g., bioRxiv). The time evolution of the preprint-to-postprint ratio for each service indicates that recent growth is a result of more preprint submissions. The nine COS services we investigate posted papers that appear in a range of peer-reviewed journals, and many of these publication venues are not listed in the Directory of Open Access Journals. As a result, it is likely that the COS services function as open repositories for peer-reviewed papers that would otherwise be behind a paywall. We further analyze the coauthorship network for each COS service, which indicates that the services have many small connected components, and the largest connected component encompasses only a small percentage of total authors on each service. This indicates all services can continue to grow. When comparing the papers submitted to each service, we observe topic overlap measured by keywords self-assigned to each manuscripts, indicating that search functionalities would benefit from cutting across the boundaries of a single service. Finally, though annotation capabilities are integrated into all COS services, it is rarely used by readers. Our analysis of these services can be a benchmark for future studies of preprint service growth.