bepress and SSRN: Part II

Yesterday I posted some thoughts about the usage of SSRN and Berkeley Electronic Press (aka bepress) in the legal academy.  Specifically I wondered why my bepress stats seemed higher than my SSRN ones, and whether use of bepress' Expresso service had caused any of the difference, when it seemed (to me at least) that SSRN was the market leader.  That post drew a helpful comment from Bernie Black, who, in addition to being a guru at U-Texas, is the Managing Director of SSRN.  Bernie wrote:

(On behalf of SSRN): SSRN takes great care to ensure that paper downloads are an accurate measure of reader interest in an author's work. First, we ensure that only informed decisions to view a full text of a particular paper, rather than uninformed explorations triggered by a catchy or vague title, count as a download. Every download starts with a reader visiting the paper's "abstract page". Only readers who still want the paper, after seeing the abstract, can download the paper. In general, only one out of three abstract views result in a download. This ratio would be lower if we include views of the title or abstract through our email abstracting journals. Second, we do our best not to count multiple downloads of the same paper by the same person nor machine or "robot" downloads of a large number of papers. If SSRN permitted a single click to download using a link from another source, such as a search engine or a blog), and mechanically counted all downloads, this would likely inflate our download counts by a factor of five or more and, degrade their quality as a signal of paper quality.

BE Press, as best as I can tell, takes the opposite approach. If you click on a Google link, for example, you have just downloaded the paper. I would imagine that when Google or Yahoo crawls their site, those downloads count too; at least they have never publicly said otherwise.

Well, at PrawfsBlawg, we provide nothing but fair and balanced coverage of these issues, so I'm happy to report that our hard-hitting investigative research has generated a response from bepress.  After the jump, see the response to Bernie from Jean-Gabriel Bankier, who's a VP at bepress:

The discussion about bepress download figures surpassing SSRN is most interesting. I am grateful to Dan for inviting me to participate.I will do my best to shed some light on what is and what is not driving bepress downloads. Dan asked me a superb question by e-mail. He wanted to know if our download count includes law review editors reviewing manuscripts submitted via ExpressO. The answer is no. Downloads by law reviews for the purpose of review are NOT included in author download figures. SSRN, as far as I can tell, does count downloads from law review editors using e-submissions (ExpressO a-la SSRN). I hope Bernie will let us know if those downloads are substantial or not, and if my understanding is correct. Search engine robots and crawlers cannot explain our download numbers either. 

Contrary to what Bernie says, we do NOT count downloads by Yahoo, Google and all the other well-known (and many less well known) search engines.  In order to ensure that paper downloads are accurate our computers run nightly through the days download logs to exclude numerous downloads of a paper from single user/IP and hits by robots. Paper download totals are then updated the next day. Bernie is correct that if a person clicks to request the full-text we count that as a full-text download.  It is a full-text download. The implication that anything is wrong with that approach is unfathomable to us. Bernie suggests one reason that bepress is getting more downloads than SSRN is because we are providing the fastest path to the content.  I think he is correct.  If you were to conduct a web search for a paper you might very well find two bepress links among the top search results.  One link would take you to the abstract page and a second one, identified as the full-text, would open the full-text of the paper.  If there happened to also be an SSRN link it would surely require the reader to first go to an abstract page and only from there grant access to the full-text. Our view is that readers don’t care if they find the paper they are looking for on bepress or SSRN.  They do care about limiting the time and effort it takes to get a hold of a paper.   I am not dismissing SSRN’s approach. The benefits to a company like SSRN that come from requiring readers go to an abstract page before letting them download the full-text are clear.  At the abstract page you have an opportunity to sell adverting and otherwise engage readers in ways that are impossible if you just serve the full-text. We, however, felt strongly (and still do) there were even greater long-term advantages to providing the fastest and easiest access to papers.