Some interesting ferment happening in repository-land, notably this discussion of various types and scales of repositories and how successful they can expect to be given the structural conditions in which they are embedded.
I don't blog repositories per se any more, so I'm not going to address the paper in detail (though I do think it contains serious oversights). What I'm curious about in the Trogool context is the case of institutionally-hosted services aimed not specifically at the institution, but at a particular discipline.
arXiv. ARTFL. PERSEUS. DRYAD. There's any number of these. One can't call them "institutional" repositories. One can't quite call them "disciplinary" repositories, either, because that implies a source of financial support beyond the institution.
Another class of these resources is not open-access, incidentally; it uses subscriptions to support further additions to the corpus. The Brown Women Writers Project is an example. Much of the "sustainability" talk coming out of think-tanks like Ithaka respects and promotes this business model. I do think it important to note that the institution hardly disappears from the support infrastructure when the subscription dollars start rolling in (assuming, of course, that they do).
I heard yesterday that one such corpus, while of impressive quality and very highly regarded in the discipline, was all but invisible on its home campus, according to the corpus's own staff. Basically, these projects are what I have previously called fiefdoms. (If you don't like that word, you may wish to substitute "research lab." Most of what I'll say applies to them as well.)
Sustainability is the crucial flaw in any sort of fiefdom model for data management. Most fiefdoms get the ball rolling with grant money. This may commit the institution to a certain amount of financial or in-kind support (depending on what the grant spells out), or it may not. If it does, that institutional support lasts only as long as the grant does. No one in this cycle—not the researchers in the fiefdom, not the institution, not the grant agency, no one—takes responsibility for the post-grant existence of anything the fiefdom produces.
For some projects, that's fine. Software projects can cast their code upon the open-source waters, or sell it to industry. Projects that are easily print-publishable can be published. Projects that have dollar signs attached to them can go to the tech-transfer office (though I share the general dubiety that tech-transfer offices are a net win for institutions).
For nearly all digital projects, the fiefdom model is a disaster. Fiefdoms live brief lives, die quiet deaths. Many fly under the radar; asking too loudly or too often for institutional support risks the institution looking down its nose and shutting the fiefdom down.
Arguably, institutions should not do this. Institutions, however, can be remarkably myopic about discipline-oriented behaviors—any behavior that doesn't directly and obviously benefit the institution as a whole, really. One of my favorite examples can be found in the Ithaka report on university presses, in which university provosts loudly trumpeted the necessity of (other institutions', presumably?) university presses to their local scholars, but declined to continue supporting those presses locally, as they were perceived as frills, not strictly necessary to institutional continuance.
As usual, I adduce the library as the institutional component whose mission and funding are best-placed to address this gaping hole in the data-management framework. What academic libraries appear to lack, unfortunately, is the will to step forward and accept this responsibility.
I have no answer, then.
This Thursday is Thanksgiving in the States, and I am furloughed on Friday, so I'll be visiting friends. I'll try to queue up a tidbits post for Friday or the weekend.