Chris Armbruster, as in the past, and like many others, completely conflates the problem of content and the problem of functionality:
(1) Virtually all OA repositories today -- institutional and central -- are low on content: Only about 15% of annual refereed research is being deposited today. (2) The only two exceptions are the fields of physics and economics, where authors have been spontaneously depositing their papers in, respectively, Arxiv and various collections of working papers in economics, now harvested by RePEc. (3) Even after many years of their positive example, the self-archiving practices of these two fields have failed to generalize to the rest of scholarly and scientific research. (4) This is why self-archiving mandates -- from research institutions and research funders -- are needed. (5) Since all research, in all fields, originates from institutions, institutional repositories (IRs) are the natural, convergent locus of deposit for both institutional and funder mandates. (6) Because IRs are OAI-compliant, hence interoperable, their contents (metadata + links, or metadata + full-texts) can be harvested into central collections (CRs) of various kinds (subject-based, funder-based, nation-based, or global). (7) Functionality can be enhanced at the harvester level in many ways; all that is needed is the content itself. (8) But we won't have the content unless we mandate it. (9) And mandates won't work if funder mandates and institutional mandates are in competition, and diverge. (10) Institutions are the content-providers, in all fields, funded or unfunded. (11) Institutions share with their researchers a joint interest in maximizing the accessibility, uptake, usage, and impact of their joint research output. (12) Institutions can also monitor and ensure compliance with funder mandates (alongside their own institutional mandates). (13) Locus of deposit has absolutely nothing to do with functionality. (14) But locus of deposit has everything to do with ensuring that the content is provided. On 17-Jul-08, at 3:54 AM, Armbruster, Chris wrote: > I would like to publicly applaud the NIH policy makers for strengthening > a central repository. NIH could "strengthen" its central repository (CR) (PubMed Central) irrespective of the locus of deposit. Locus of deposit is relevant to maximizing content provision and unrelated to functionality. > As far as I can see, after several years, > institutional repositories have not made decisive progress in being > useful to either authors or readers by providing services that are > of any value (beyond storage). The purpose of IRs is not to provide services but to provide content. The services are provided at the harvested collection (CR) level. And the usefulness of CR services depends entirely on whether the content -- on which the service is to be based -- is actually provided in the first place. > If I look at the kinds of services > that arxiv, SSRN, CiteSeerX, RePEc and PMC offer, I see no equivalent > emerging from the IRs, no matter how much you synchronize and > harvest. I have great difficulty understanding the point Chris is trying to make here: Both CiteSeerX and RePEc are harvester services. There is no CR there in which authors deposit directly. CiteSeerX and RePEc (like Google Scholar) harvest their contents from IRs and other institutional and personal sites on the web. Arxiv, as noted, is a longstanding CR in which physicists have been depositing directly since 1991, but there is no sign of that spontaneous phenomenon duplicating itself in any other field (even though CRs are available in other fields too, including CogPrints, in cognitive sciences, which I created in 1997). SSRN is a CR, but the way to assess how full it is is to divide its annual contents by the global annual output in all the fields covered. It will be found to hover at the very same spontaneous deposit level (15%) as the IRs. And no matter how many or wondrous the services you provide over it, 15% is still just 15%. No one would search a topic IR by IR, so it makes no sense providing certain services at the IR level. (IRs provide local services pertinent to the institution itself, such as generating CVs, research assessment data, and usage statistics. If you want to search across IRs, go to OAIster, Google Scholar, CiteSeerX, or Citebase. But you will be disappointed, because all you will find is about 15% (except in physics and economics). That's what the mandates are for. And that's why it's important that institutional and funder mandates converge on the providers, the IRs, rather than competing, by requiring direct deposit in institution-external CRs (instead of just having the CRs harvest). > Also, centralized repositories seem to lend themselves > much more easily to the creation of overlay services that extract > further value for the scholarly community. Overlay services can be developed over any OAI-compliant repositories, whether IRs or CRs. The locus of deposit makes no difference whatsoever. That was the whole point of the OAI protocol. > Just consider the following > service: http://www.gopubmed.org/ (developed in Germany, based on > the efforts of the NIH, a splendid example for the kind of > trans-national innovation that has become possible on the basis of > repositories). And if NIH mandated direct deposit in IRs, and harvested PMC content from there, the very same services could be built on it. The difference would be that the NIH mandate would be convergent and synergistic with institutional mandates, generating far more content, beyond just what is funded by NIH, across all fields, institutions and countries. > I hope the NIH holds fast and that more research funders will ensure > deposit in centralized repositories - either discipline-specific > or at least national. For the "success" of national CRs, see France's HAL. Without mandates, it languishes at the usual 15%, no matter how you cut the cake. No, Chris, what's missing is content, not functionality. And the reason for the focus on IRs is because that is the convergent, systematic way to get all the content, not depositing willy-nilly and hoping that that will somehow cover all of OA space. Stevan Harnad