Or Maybe Just the Extraordinary Impact of Search Results?!?

The result is interesting , but could be simply driven by the way scientific papers are ranked in search results.

If citations to scientific papers are used to rank search results, then older papers are more likely to rank higher on any given page (by design, they had more time to be cited). This would lead them to accumulate even more citations over time (a positive feedback loop).

The phenomena is much older than the Internet, and one of its best characterizations can be found in the work of Robert K. Merton’s on the Matthew effect (1968).

If one were able to reconstruct where a paper ranked in the search results over time, then a regression discontinuity analysis would be a better approach, as it would allow us to isolate the impact of rankings from the quality and usefulness of the underlying science.