Re: Repositories: Institutional or Central?

From: Ept <ept_at_BIOSTRAT.DEMON.CO.UK>
Date: Fri, 13 Feb 2009 18:45:57 -0000

Since deposits may now be tagged according to who funded the work,
since deposits in institutional repositories and central repositories
can now be readily harvested in either direction to meet any needs of
institutes or funders, I don't understand why this debate continues.
As one of a group working with information deprived developing
country researchers, I share Bernard Rentiers mystification as to
why people spend all this time on these repetative discussions
instead of just focussing on helping the academic communities make
their publications OA. There are major problems waiting that science
can solve faster once  research discoveries are shared. Hats off to
CSIR India's recent announcement, see EPT blog.
Barbara Kirsop
EPT Blog
Electronic Publishing Trust for Development
      ----- Original Message -----
From: Stevan Harnad
Sent: Friday, February 13, 2009 3:44 PM
Subject: Re: Repositories: Institutional or Central?

On Fri, Feb 13, 2009 at 5:46 AM, Peter Millington
<> wrote:

      Following the LOCKSS principle, I personally
      support both institutional and central
      repositories. Depositing in both may also help to
      disseminate research more widely.

Let 1000 flowers bloom, but not by asking their creators --
currently loath to plant even one unless mandated -- to deposit
their seeds multiply. One deposit is enough. The rest is what
(to mix metaphors) automatized harvesting is for.

And the natural and optimal locus for that one-time
implantation is the creator's own Institutional Repository
(IR), since institutions are the source of all the planet's
research, across all disciplines, across all nations, and
whether funded or unfunded. 

The rest is all a matter of automatized harvesting, whether for
the purposes of global search, subject-based collections,
grant-fulfillment, preservation, data-mining or what have you.
Being Devil's advocate however, I can understand why
research funding agencies would want to stipulate
specific central repositories [CRs] in their OA mandates.

Being Logic's and Practicality's advocate (and also universal
OA's, rather than just the funder's fraction), I have to point
out, again, that once mandators stipulate convergent IR
deposit, the rest is just a matter of software-based
harvesting. IRs can provide funder metadata fields (as EPrints
does). The fulfillment conditions for the grant can stipulate
that the deposit must be automatically exported by the IR to
the funder's designated CR, whatever that may be.

The point that is not being appreciated in all this effluence
of enthusiasm for multiple deposits willy-nilly is that
currently most research is not being deposited anywhere at all.
That's what the mandates -- both funder and institutional --
are for. And the universal providers of all research are the
institutions. The funders just fund a portion. But their
extremely welcome, timely and valuable deposit mandates can do
a lot more than just ensure that their own funded portion is
deposited -- if the designated locus of deposit is each fundee
institution's own IR rather than some institution-external CR,
for that will motivate each institution to deposit -- and
mandate the deposit of -- all the rest of its output in its own
IR too. 

Not so if deposit is mandated institution-externally. 

(And please don't cite the possibility of automatised
back-harvesting from the CRs to the IRs as the symmetric
remedy, because the fundamental fact this fallacy overlooks is
that most IRs are lying fallow today, and are as unmotivated to
back-harvest as they are to deposit or mandate deposit: That's
what funder mandates stipulating IR deposit are in a position
to jump-start, at long last -- if only we first grasp the
logic, as well as the practicality, of the strategem.)
      If you are serious about policing compliance, it is
      much simpler for a research funder to monitor a
      central repository than a myriad of institutional

Its also easier for a funder to communicate with fundees at a
single email address rather than a myriad of institutional
addresses. But in fact the solution is much the same in both
cases. Each fundee has on record with the funder his
institution and his institutional email address. By the same
token, he can add the URL (or DOI) of the IR locus of deposit
of each of the published articles he lists as having been
outputs and outcomes of the grant funding. (And, as noted,
automatized export, import, or harvesting can be done on the
basis of those data, and IR metadata tags.)

None of this calls for direct central deposit of articles by
the fundee, and all good sense militates against it.

 Of course for institutional mandates it is the other way

This sounds like the "symmetry" fallacy again: No, it is not
true that there is any functional, practical or logical
symmetry between (1) the hypothetical back-harvesting to
their own IRs -- by unmotivated, nonmandating institutions
-- of the funded fraction of their own institutional research
output, deposited institution-externally in mandated funder CRs
(or other CRs scattered across the planet) and (2)
funder-mandated direct deposit in the institutions' own IRs to
motivate institutional deposit mandates for all the rest of the
institutions' research output.

The underlying rationale for this fundamentally important but
subtle strategic point about locus-of-direct-deposit (for
funder mandates especially) deserves to be carefully thought
through, rather than dismissed with vague assumptions about
harvesting symmetry and letting 1000 flowers bloom....
Stevan Harnad
Received on Fri Feb 13 2009 - 19:43:08 GMT

This archive was generated by hypermail 2.3.0 : Fri Dec 10 2010 - 19:49:42 GMT