Caveat Lector

Stevan Harnad, in the course of a useful blog post about the practices of faculty, notes that faculty who prefer disciplinary repositories generally refuse to double-deposit in IRs, and suggests the following:

The solution for this relatively small population of seasoned self-archivers is for their institution-external deposits to be automatically imported back into their IRs using the SWORD protocol (which can also be used to export automatically from IRs to central repositories). There is no need for veteran self-archivers to change their practices or to double-deposit.

It’s a nice vision, and it’s technically feasible, but it doesn’t quite work the way Dr. Harnad would like, at least not via SWORD. Let me explain.

SWORD is not a harvesting protocol; it is a deposit protocol. The party that initiates a SWORD deposit is the party with the material in hand. SWORD offers no way for a repository that wants material from another repository to request it, much less do so in an “automatic” fashion.

For the SWORD-based dual-deposit system Dr. Harnad describes to work, the disciplinary repository would have to arrange for SWORD deposit into every single IR represented by the faculty who deposit there. It would also have to track faculty affiliations, to send deposited materials to the correct IR(s). This is barely feasible, but it’s obnoxiously complicated and error-prone. I can’t imagine too many disciplinary repositories signing on to do it, especially considering that the demand for such a service from the faculty themselves is minimal to zero by Dr. Harnad’s own admission. (Sure, repository-rats would dearly love it, but who listens to us?) Where is the incentive?

I think there are better ways to solve the problem. Let me suggest a couple.

One is to turn the problem around; let IRs do SWORD deposit to disciplinary repositories, snagging a copy for themselves in passing. IRs and their repo-rats do have incentive to build this, as we create bonus visibility for our faculty thereby, and we can (at least potentially) help with funder mandates, too. All the target repositories have to do is implement SWORD and hook that into whatever safeguards they need to ensure that IRs aren’t feeding them stuff they don’t want.

(If PubMedCentral or arXiv or RePEc or SSRN were to implement SWORD, it would be a gigantic step forward. Please, will someone in the know suggest it to them?)

Another advantage to this setup is that institutions without IRs but with SWORD-enabled faculty-publication databases such as the BibApp can take advantage of it too, at the very least for that subset of journals that allows final-PDF deposit.

Another potential solution is to look more closely at OAI-ORE instead of SWORD. Some scaffolding will be necessary here, because figuring out what’s going into a disciplinary repository from a given institution is often a shockingly hard problem, but once again, IRs have incentive to build this that disciplinary repositories don’t.

I wholeheartedly agree that repositories of all sorts need to talk to each other better—but that can’t start without a basic shared understanding of the technical infrastructure. I offer this post in service of that goal.