I meant to mention another thought that lodged itself in my head during the ACRL institutional-repository preconference, but it slipped my mind until just now. Executive summary: IRs are springing up like mushrooms on an old log, and I’m not at all sure the market is going to bear that many.
I believe the University of Wisconsin is ahead of its time in this; as I understand things, its just-launched IR is intended to serve the entire UW system, not just the Madison campus. I don’t think they’re actually having much luck with that at present—awareness of the new service is low enough in Madison, never mind River Falls or Stout. I do think, though, that the folks in charge have thought this through and concluded that campus-specific IRs involve too much redundant effort.
As always, I could be wrong. It’s possible that IRs will become a standard academic-library service, as ingrained as interlibrary loan. Truly, though, I don’t think so.
The central reason I don’t think so actually came up during the preconference: someone mentioned that faculty tend to identify more with their disciplines than their institutions. They may switch jobs, but they’ll always be anthropologists or film scholars or mathematicians or whatever. As they search for outlets for their materials, they’ll look first to discipline-specific outlets. What is a journal, after all, if not a discipline-specific publication outlet? What faculty member seriously publishes in a journal specific to her institution?
So institution-based IRs have an uphill climb to persuade faculty that they’re useful. Compare that needed effort to something like arXiv, which every physicist worth the name knows about. Now, that’s immediately, obviously useful, both to faculty-as-researchers and faculty-as-publishers. Even the UW’s system-wide IR can’t possibly be as attractive to faculty as something arising within their discipline and publicized in discipline-specific arenas.
Campus IRs are fighting the tides in several ways; this D-Lib article especially impressed me with its faculty-centric approach to attracting content. Such an approach, however, creates a dilemma for libraries contemplating IRs: most IRs (as I understand matters) want only the finished products of faculty research, whereas faculty most need librarians’ help coping with half-finished work and grey literature. So do we uphold our standards and refuse everything but the final outcome, or do we spend effort becoming (in essence) publishing and production partners with faculty?
I lean toward the latter approach, myself. Where do we get off claiming the end result if we take no responsibility for helping produce it? Why shouldn’t research production be a research-library service? But then, we all know I’m a text artisan; production is baked into my bones. Every IR will have to make its own decision.
Whichever decision it makes, however, that still doesn’t help the problem of institutional versus disciplinary resource aggregation. To my mind, the solution there is brilliantly simple: libraries should ally with scholarly societies to build discipline-specific IRs. Obviously each library should pick disciplines its host institution is strong in; equally obviously, there’ll have to be partnerships and consortia and general shakedowns, as IRs try to claim the same disciplinary spaces. In the long run, though, I think that gets us the best bang for the library buck—faculty will be naturally attracted, the research gets saved and preserved, and only those IRs that need to exist, will.
From a technical standpoint, this analysis argues strongly for both open-source IR solutions and a well-considered data-migration strategy for existing IRs. An institutional IR that can’t move its content into appropriate discipline-specific repositories without data loss risks making an awful mess on the large scale. An IR should even be able to dissolve completely, moving all its materials elsewhere, without loss.
Vendor lock-in is just not on. Any library considering a proprietary IR system (and I see now that ProQuest is not the only player in this field) had better beat its vendor over the head with data-migration needs, and refuse outright any so-called solution that doesn’t migrate. Use DSpace or eprints or Fedora instead; proprietary polish is simply not worth migration agony.
