On Wed, 11 Aug 2004, Heather Morrison wrote: > as we move towards global sharing of information, there probably is no > one model that will fit either all disciplines, or all countries. Within > the next few years, I fully expect that universities around the world > will have created their institutional repositories or archives. For now, > however, many of these projects are still in the planning. However, > with PubMedCentral, we can have OA right away.
Agreed about the mix of both. (OAI moots the difference.) But the number of institutional archives is growing much faster worldwide than the number of central archives: http://archives.eprints.org/eprints.php?action=browse And there is a good reason for this: Institutional archives, being local, and simpler, are incomparably easier and cheaper to create and maintain. Moreover, institutions share in the motivation for and the benefits of OA with their own institutional researchers. Central archives do not. If we have learned anything from network ecology, it is that online information growth and development are distributed and anarchic, not central and uniform. Meanwhile, the OAI protocol provides all the uniformity needed to make all the distributed OAI archives -- institutional and central -- interoperable, as if they were all one global virtual-archive. > Researchers in France (and soon, the UK) who have good institutional > repositories to work with, in the event they are recipients of NIH > funding, should submit identical copies of their papers to both > PubMedCentral and the local IR. But if there is one thing we have learned from the sluggish history of OA so far, it is that if you wait for authors to self-archive spontaneously (i.e., to do what they "should" -- even when the "should" is strongly in their own interests), then you have a long wait ahead of you! That is why it has become evident that a self-archiving mandate from authors' institutions and research-funders -- a natural extension of their existing publish-or-perish mandate -- is needed in order to get authors to do the right thing for themselves (and their institutions and funders, and for research itself). So that's why it's important to get the mandate right: The more powerful and general mandate, the one that will propagate across all institutions and disciplines, is the UK Mandate: Institutional self-archiving. http://www.ecs.soton.ac.uk/~harnad/Temp/UKSTC.htm Yes, central archives can (and should!) harvest the metadata, the full-texts, or both, form those institutional archives. But that would all follow quite naturally from an Institutional self-archiving mandate. The reverse, however, is not true: Mandating central self-archiving for particular funded research in a particular discipline and archive will *not* propagate across disciplines and institutions. It will cover all and only its own mandate. Nor is the process of distributed institutions harvesting their own content back from central archives as natural or straightforward as central archives harvesting from distributed ones (though with OAI, either is possible). But the main point is that mandated institutional self-archiving will *drive* the spread of self-archiving across disciplines and institutions, whereas mandated central discipline-specific self-archiving will not. If the US keeps its mandate central, it will be a much bigger opportunity lost for OA (but of course still far better than no self-archiving mandate at all!) > I see no reason why harvesting of documents themselves, not just > metadata, could not be automated in the very near future, to facilitate > this process. It can, will, and is being! But the natural direction to harvest is from the distributed local institutional archives to global, central ones, not vice versa. Even more important, the way to propagate the practise across disciplines is to do it within universities (institutions), because universities have all the disciplines, and having done it in one, they will naturally do it in the others too. Central discipline-based archives are monads, and their individual practises do not propagate across disciplines. > Plus, of course, centralized search tools, whether OAIster or PubMed, > can search documents that are archived in a distributed fashion. That too. All that's wanting is that 100% OA corpus. So let's not put the central cart before the distributed horse! Stevan Harnad
