[24 May 2009]
At the Summer Institute on Data Curation in the Humanities (SIDCH) this past week in Urbana (see previous post), Dorothea Salo surveyed a variety of threats to the longevity of humanities data, including lack or loss of institutional commitment, and/or death (failure) of the institution housing the data. People serious about maintaining data accessible for long periods need to make succession plans: what happens to the extensive collection of digital data held by the XYZ State University’s Institute for the History of Pataphysical Research when the state legislature finally notices its existence and writes into next year’s budget a rule forbidding university administrators to fund it in any year which in the Gregorian calendar is either (a) a leap year or (b) not a leap year, and (c) requiring the adminstrators to wash their mouths out with soap for having ever funded the Intitute in the first place?
Enough centers for computing in the humanities have been born in the last fifty years, flourished some years, and later died, that I can assure the reader that the prospect of going out of existence should concern not only institutes for the history of pataphysics, but all of us.
It’s good if valuable data held by an organization can survive its end; from the point of view of URI persistence it would be even better if the URL used to refer to the data didn’t have to change either.
I have found myself thinking, the last couple of days, about a possible method of mitigating this threat, that runs something like this.
- A collection of reasonably like-minded organizations (or individuals) forms a mutual assistance pact for the preservation of data and URIs.
- The group sets up and runs a PURL server, to provide persistent URLs for the data held by members of the group. [Alternate approach: they all agree to use OCLC’s PURL server.]
- Using whatever mechanism they choose, the members of the group arrange to mirror each other’s data in some convenient way. Some people will use rsync or similar tools; Dorothea Salo observed that LOCKSS software can also do this kind of job with very low cost in time and effort.
- If any of the partners in the mutual assistance pact lose their funding or go out of existence for other reasons, the survivors agree on who will serve the decedent’s data. The PURL resolution tables are updated to point to the new location.
- Some time before the count of partners is down to one, remaining partners recruit new members. (Once the count hits zero, the system has failed.)
[Wendell Piez observed, when we got to this point of our discussion of this idea, “There’s a Borges story in that, just waiting for someone to write it.” I won’t be surprised if Enrique is working on one even as I write.]
In some cases, people will not want to use PURLs, because when they make available the kind of resources whose longevity is most obviously desirable, the domain name in the URLs performs a sort of marketing or public awareness function for their organization. I suppose one could allow the use of non-PURL domains, if the members of the group can arrange to ensure that upon the demise of an individual member the ownership of their domains passes seamlessly to some other member of the group, or to to the group as a whole. But this works only for domain owners, and only if you can find a way to ensure the orderly transfer of domain ownership. Steve Newcomb, my colleague in the organizing committee for the Balisage conference on the theory and practice of markup, points out a difficulty here: in cases of bankruptcy, the domain name may be regarded as an asset and it may therefore be impossible to transfer it to the other members of the mutual assistance society.
It’s a little bit like the burial societies formed by immigrants in a strange land for mutual assurance, to ensure that when one dies, there will be someone to give them a decent burial according to the customs of their common ancestral homeland, so that their souls will not wander the earth restlessly in perpetuity.
It would be nice to think that the data collections we create will have a home in the future, lest their ghosts wander the earth restlessly, bewailing their untimely demise, accusing us of carelessness and negligence in letting them die, and haunting us in perpetuity.