Re: Plan B for NIH Public Access Mandate: A Deposit Mandate

["Stevan Harnad" <amsciforum@gmail.com>]

On Fri, Sep 19, 2008 at 10:17 PM, Joseph J. Esposito <espositoj@gmail.com>
wrote:

> I feel obliged to state the obvious:  Stevan Harnad's comment 
> in this thread about "that rare, lucky author" is an admission 
> that OA has little impact. That author is rare and lucky 
> because he or she has so many requests for copies of articles 
> that are otherwise not available to other researchers. Most 
> authors, of course, will not be troubled much with requests 
> because the articles are indeed available to most researchers 
> through institutional subscriptions.

I'm afraid that is not the explanation at all! The reason the 
author with 1000 eprint requests is rare and lucky is because 
most authors get far fewer eprint (or reprint) requests than 
that, whether or not their articles are OA. Not only does the 
Seglen 80%/20% rule (the "skewness of science") apply to 
citations (the top 20% of articles get 80% of the citations) but 
it applies to downloads and eprint-request effects as well.

(There is one interesting yet-to-be-answered empirical question 
there, however, which concerns the degree to which the 80/20 
filter is based on the metadata -- author/title/abstract -- 
alone, versus the extent to which it is -- or will be -- based on 
a browsing of the full-text. Probably there is an 80/20 effect at 
each level -- citations, downloads, eprint requests -- but with 
different scales, and possibly browsing will have a somewhat 
flatter ratio (say, 70/30, who knows?) than citing, because it is 
ergonomically "cheaper" to browse a paper whose title looks 
promising than it is to read it through to make sure it is NOT 
promising after all.)

Seglen, EO (1999) The Skeweness of Science. JASIST 43: 628-638 
http://www3.interscience.wiley.com/journal/10049716/abstract?CRETRY=1&SRETRY=0

There is also a positive correlation between earlier downloads 
and later citations, and I don't doubt that a similar correlation 
will turn out to be operating with eprint requests.

(As you know, even Phil Davis's premature APS journal study with 
randomized OA detected a significant download advantage in the 
first year, when it was still too early to detect any citation 
advantage. There is your evidence, if you still needed it, that 
access is NOT "available to most researchers through 
institutional subscriptions..." The OA citation advantage is the 
further evidence.)

Brody, T., Harnad, S. and Carr, L. (2006) Earlier Web Usage 
Statistics as Predictors of Later Citation Impact. Journal of the 
American Association for Information Science and Technology 
(JASIST) 57(8) pp. 1060-1072. 
http://eprints.ecs.soton.ac.uk/10713/

Last point: The absolute scale of the citation 80/20 differential 
is of course quite a bit lower than the download or 
eprint-request 80/20 differential, so whereas the author who gets 
1000 eprint requests is rare and lucky, the author who gets 1000 
citations is even rarer and luckier! OA will raise that absolute 
number, but probably not the ratio. And what that means is that 
OA benefits the better articles more. (The "Quality Advantage.")

> Whatever one feels about the legality of the NIH policy, the 
> conclusion is inescapable (citing Harnad as above) that OA is a 
> small idea.  How it has come to dominate discourse concerning 
> scholarly communications is a marvel, comparable in its way to 
> the sudden interest of the popular media in hunting moose.

(As a vegetarian, I can say that I certainly hope there is no 
affinity between the two!) But, to repeat the same point as 
above, the reason OA benefits a small portion of research more is 
not that OA is a small idea -- doubled downloads is a big idea! 
-- but that scholarly and scientific quality (and hence usage and 
citation) is skewed.

Stevan Harnad