Departmental Repositories, Institutional Repositories, and Research Record-Keeping

---

SUMMARY: Individual departments should not sit and wait for their university to get its OA act together: Unless a consensus on adopting a top-down university-wide OA self-archiving mandate can be reached promptly, departments should go ahead and adopt their own bottom-up mandates (the "patchwork mandates" recommended by Prof. Arthur Sale), as Southampton's ECS did in 2003. Apart from maximising the visibility, accessibility, usage and impact of departmental/institutional research output, an OA Repository (with a self-archiving mandate) can serve as an institution's or department's internal record, ending its need to rely on external proprietary databases in order to monitor and assess its own research output. (Southampton now has at least 13 EPrints Repositories to date; Cal Tech has a whopping 25.)

---

On Mon, 1 Oct 2007, N. Miradon wrote in the American Scientist Open Access Forum:

NM: I know that Dr Leslie Carr wrote: "...we genuinely can't answer questions about the percentage of our research output that gets put into our repository, because we have [no] independent way of knowing what the size of our research output is!"

But a quick search in the staff publications site www.civil.soton.ac.uk/staff/allstaff/staffpubs.asp?NameID=nnnn (where nnnn is a random integer between 1 and 1890) gives many publications that do not seem to be available in http://eprints.soton.ac.uk

It would be interesting to know the % deposit rate in ECS from all faculties (Divisions, Research Centres) in Southampton University. Could someone do a quick spidering of allstaff/staffpubs.asp?

The ever-alert Napoleon Miradon, has raised two very important and valid questions in connection with my posting about Les Carr's estimates of the current deposit rate in the repository of the Department of Electronics and Computer Science at Southampton University. One of the questions is an explicit empirical one, and the other is an implicit methodological one. I shall answer the methodological question first. Reformulated explicitly, the question is:

Since, apart from their Repositories themselves, there is today no record of their total research output, Southampton -- and most other universities and departments worldwide -- have no way of knowing what their total research output is:

So how can they determine what percentage of that total research output is being self-archived?

The answer is that that can only be estimated today by consulting external databases, such as ISI's Web of Science, or ACM's Digital Library (or Google Scholar), to sample what has been published, and then to check back to see whether they are in the Repository. That is what Les Carr did, and that is where his percentages come from.

The point to note here is that one of the added benefits of having an OA Repository and a self-archiving mandate is that once the deposit rate has been confirmed (by such external sampling) to be at or near 100%, the Repository itself can be used as the internal record of the institution's or department's research output. Count that -- alongside the fact that it maximises the visibility, accessibility, usage and impact of departmental/institutional research output -- as yet another reason for having an OA Repository, and for mandating deposit: It is a very powerful and effective form of internal record-keeping, so an institution (or department) can track its own research productivity as well as submit it for external performance assessment. Having and filling its own OA Repository also releases the institution or department from the need to consult and rely on external proprietary databases in order to monitor its own research output.

But I think M. Miradon in fact understood the fact that Repository deposit rates currently have to be estimated through sampling; I have only made the methodological point explicit for readers who might have needed the clarification. We now move on to M. Miradon's empirical point: He has done a bit of random sampling himself, and indeed he has managed to do this using a Southampton-internal record of Southampton publications: He has sampled the staff publications list in Southampton's Department of Civil Engineering, and he has found many publications to be absent from Southampton's Department of Electronics and Computer Science's IR: http://eprints.ecs.soton.ac.uk/

But the explanation for this is very simple: I was reporting the results for the 4-year-old Repository and mandate of the Department of Electronics and Computer Science at Southampton (ECS), not the Department of Civil Engineering! I stressed -- in connection also with Arthur Sale's recommendation that universities should promptly proceed with adopting bottom-up departmental mandates ("patchwork mandates") until/unless they have rapid consensus on adopting a top-down university-wide mandate -- that ECS's mandate, the world's first OA mandate as far as I know, was a departmental mandate, not a university-wide mandate.

Which prompts me to describe a few more historical details about self-archiving policy at the University of Southampton: As anyone can see by consulting ROARMAP, Southampton does have another, bigger Repository, and another Repository policy: It has a university-wide Institutional Repository (IR) -- ePrints Soton -- and has had it for nearly as long as it has had the ECS Departmental Repository ("DR"). Southampton also now has a university-wide mandate proposal (not yet announced), one that has even been officially approved; but it has not yet been officially adopted.

(Don't ask me why it is taking Southampton so long to adopt its approved self-archiving mandate! I have no idea, except that I note that the delay is depressingly commensurate with similar bureaucratic delays at many other institutions. All the more reason for individual departments like ECS to push ahead with Arthur Sale's "Patchwork Mandate" rather than sitting around waiting for their university to get its act together: Southampton has at least 13 EPrints Repositories; Cal Tech has a whopping 25.)

Last point: I confidently count a self-archiving mandate a success if it generates a deposit rate of 100%. That means the keystrokes are getting done; and it is (and always was) keystrokes alone that have been standing between the research community and the 100% Open Access to its own research output that has been within its reach ever since the dawn of the online era.

Les Carr points out that some of the ECS Repository deposits are Closed Access (CA) rather than Open Access (OA). That is not a problem, because the Repository's semi-automatic "Email Eprint Request" Button (also known as the "Fair Use" Button) can provide almost-immediate, almost-OA during a Closed Access embargo period, providing for all user needs until either embargoes die their natural and well-deserved deaths under mounting webwide pressure from the increasingly palpable benefits of OA, or authors simply tire of performing the extra keystrokes involved in fulfilling individual eprint requests one by one, and hit the master key that transforms their deposit from "CA" to "OA" once and for all.

A slightly more problematic case is the one where the authors have only done the keystrokes to deposit their metadata, but have failed to do the last keystroke, the one that deposits their full-text (whether as OA or CA). There we have a visible but orphaned reference, with no text to request or send. The EPrints IR software has not implemented a second button, with which would-be users can prod the author to deposit the missing text (and then send it), because we are confident that this dysfunctional practice is becoming increasingly rare and will remedy itself with time and experience of its own accord -- inasmuch as it needs to be remedied at all: For there are cases where an author may legitimately wish to deposit only a paper's metadata, for record-keeping purposes, but not the text itself. Examples would be seminars and conference papers that are written but not published, being merely precursors of later published papers. And of course there are books, of which the author may not wish to deposit the full text! ECS's self-archiving mandate applies only to published, peer-reviewed articles (in journals or refereed conference proceedings). Authors are not obliged to deposit every text they have ever keyed, let alone make them all OA!

NM: I have received some results from a random spidering of staff publication lists at www.civil.soton.ac.uk/staff/allstaff/staffpubs.asp?NameID=nnnn
Here are the first three entries:

Prof. Mike McDonald: 307 publications (17 in ePrints Soton)
Prof. Chris Clayton: 221 publications (14 in ePrints Soton)
Prof. AbuBakr Bahaj: 155 publications (25 in ePrints Soton)

I am afraid I did not explain sufficiently explicitly: U. Southampton has a number of other Repositories (at least 13) besides that of the ECS Department. The biggest of them is the university-wide IR, ePrints Soton.

However, as explained in my reply above, whereas the departmental mandate of the School of Electronics and Computer Science (ECS) was officially adopted in January 2003, the university-wide mandate, though already officially approved some time ago, is still (for reasons unknown to me!) awaiting official adoption. In the meanwhile, the university as a whole has no self-archiving mandate yet, so whatever deposits you find in ePrints Soton will be the usual spontaneous (unmandated) ones (for which the worldwide baseline deposit rate is about 15%), probably increased in this case beyond that baseline also by library mediation and encouragement (which Arthur Sale's analyses show to produce a better deposit rate, but nothing anywhere near Alma Swan's predicted 80-90% for mandated deposit). The deposit of only a subset of a researcher's total publications reflects the fact that depositing current and future papers requires far fewer keystrokes and effort than depositing one's full prior opus (which may even require scanning and OCR) -- though eventually that should of course all be deposited too!.

I hope this dispels any further ambiguity!

Stevan Harnad
American Scientist Open Access Forum