1. I have no idea what to do about bepress. One of my tax colleagues recently reported over 1400 bepress downloads. Her bepress downloads would probably make her one of the top 20 most downloaded tax professors in the U.S. But for reasons that escape me, bepress is willing to tell her her download counts, but will not make the same information available to anyone else. Her solution has been to vow never to post in full text on bepress again. Not a great solution, but I can't really blame her.
2. I've responded by full post to the issue you raise regarding per capita computations. I would add that in the course of generating my monthly tax faculty rankings, I've explored the SSRN posting of several hundred scholars in some depth. Many of the most productive are not tenured or tenure-track. Some are emeritus, some clinical. If the numerator includes their downloads, the denominator must count them as well. The reverse is also true. My wife, not a tax geek, is a clinical contract professor here at Loyola. Although the ABA counts her as only half a faculty member because of her administrative duties, her list of publications is longer than mine. Her most recent placements were at Utah, Minnesota, and U Pa. I have a hard time taking the position that her publications should not be taken into account in ranking the scholarly accomplishments of our school. But any reasonable per capita protocol would require that her productivity be omitted.
3. The fact that some authors decline to post on SSRN is, as you note, a real problem. As rankings based on SSRN downloads have proliferated, however, there is evidence that deans are placing pressure on authors to post. In addition to enhancing SSRN ranking, posting moves legal scholarship much closer to the open source model advocated by Larry Solum. I myself find that because of SSRN, folks who do not have free access to Lexis and Westlaw read and cite my work. (I am surprised to find myself cited in papers posted on European university websites.) My guess is that in the long run, the non-posting problem will largely disappear. Indeed, my recent contract with the U Pa Law Review REQUIRED that I post the article on SSRN.
4. I agree that citation counts are valuable. Two points: (1) They are incredibly expensive to do. I recently spent 3 RA-semesters trying to develop a citation count system that would avoid some of the most obvious problems with such systems and could be routinely updated. I gave up after concluding that the project was simply too expensive. (2) Citations measure something different, and not necessarily more valuable, than downloads. I read lots of work that I will never cite. I would not characterize such work as less useful. A lot of the articles I cite in my own work are there (frankly) to satisfy student editors; many are articles I do not consider particularly good or interesting. But I only download articles if I think they are likely to be worth the time it will take to read them. I am not, of course, advocating SSRN download counts in lieu of citation counts. Citation counts have their place; I simply do not have the time to do them properly.
Hope the foregoing thoughts are helpful.