[Moderator's Note: The research community may not yet have grasped the significance of the dramatic causal connection between the number of times a paper is accessed and the number of times it is later cited. Tim Brody's remarkable download/citation correlator http://citebase.eprints.org/analysis/correlation.php is able to predict from download counts what the citation counts will be 6-24 months later. This kind of early-days predictor could add a new dimension to scientometric evaluation of performance. It also strongly corroborates Michael Kurtz's "rule of 17" in astrophysics: 17 "reads" generates 1 "cite." It follows that whatever multiple of 17 reads is lost because of access-denial, that is also the number of citations lost! -- Stevan Harnad] http://www.ecs.soton.ac.uk/~harnad/Temp/self-archiving_files/Slide0025.gif
Dear Stevan, The "rule of 17" is the observation -- from http://cfa-www.harvard.edu/~kurtz/jasist2-abstract.html -- that the Astrophysics Data System (ADS) http://adswww.harvard.edu/ ratio of "reads" to "cites" is about 17/1 for papers of the same age, once a few years have past since publication. Because for these papers ADS is the predominant means of reading them (about 90% I would guess), this [correlation] seems to be a causal one. For papers from the Astrophysical Journal, the reads/cites ratio is really more like 12/1, but 17/1 may be correct for the related and secondary literature which dominates the "missing reads." Because citations scale exactly with reads (eqn 3 and fig 3 in the paper), missing reads translate directly into missing cites. This different from the [phenomenon] that preprinted articles are more frequently cited, which is due in large part to the differing quality of preprinted vs non-preprinted articles [i.e., preprinted ones are of higher quality]. In 1997, for example, 25% of articles from the Physical Review are in the ArXiv (according to Tim Brody's data) but 4 of the top 5 and 7 of the top 10, ranked by number of citations, are in the ArXiv: http://opcit.eprints.org/opcitresearch.shtml For me the bottom line result of my little experiment is that even for astronomy, with its near ideal [i.e., small, universally affordable] toll-based system, fully 20% of all potential article reads are thwarted by the access controls (and lack of electronic versions). (The fact that many of the inaccessable papers are in the ArXiv probably does not change this much, as the additional effort involved [from leaving ADS's unified resource to go to another system] is a great deterrent.) Cheers, Michael ------------- Michael Kurtz Harvard-Smithsonian Center for Astrophysics 60 Garden Street Cambridge, MA 02138