SSRN and Download Statistics

Brian Leiter reports on some new SSRN download statistics ranking schools and authors (SSRN login required). One of the best aspects of SSRN is that we, as authors, can get some real-time statistics on readership. Historically, we had no real idea if anyone was reading our articles; we might get occasional oral or email comments, or we could check citation counts months or years after the fact. Now, a simple check of SSRN tells us how many times our papers have been downloaded, giving an instant ego stroke. The statistics also permit a new type of benchmarking among professors and schools.

However, there’s a big problem with SSRN’s download counts—as far as I can tell, SSRN does absolutely zero to validate their accuracy. Specifically, I believe that SSRN counts each time a paper is downloaded, even if the same person downloads the article multiple times. More problematically, an author could boost their own download counts simply by repeatedly downloading their own papers. (While it would take a lot of fuss to break into the top ranks, self-downloading could easily pull an author out of the big pack of authors with relatively low download counts.) I’ve heard whispers of other gaming strategies, but I don’t know if they are true.

Producing accurate download counts is not a new issue. For example, at Epinions, we used to pay authors based on download counts. Therefore, to maintain system integrity, our software excluded author visits from download counts and counted repeat visits from other person only after a certain period of time (like 1 month). Couldn’t SSRN do the same?

Whatever the case, until SSRN fixes how it counts downloads, its statistics lack reliability. I call on SSRN to take some steps to shore up its download counts—and to publicly announce those steps so we know how downloads are counted and can rely upon them accordingly.