---------- Forwarded message ----------
Date: Thu, 9 Mar 2006 15:35:52 +0000
From: Stevan Harnad <[log in to unmask]>
To: [log in to unmask]
Subject: Re: Poynder Again on Point on Institutional Repositories

> On Thu, 9 Mar 2006 Fytton Rowland wrote:
>
> Three different questions are getting mixed up here.
>
> 1. How do we encourage/mandate/force academics to post their published
> papers so that they are available to the world free of charge?

    (a) A mandate does not mean "force"! (The universal "Publish or
    Perish" mandate is not based on force either.)

    (b) A mandate does not need sanctions or penalties. Impact is
    already rewarded in performance evaluation, and OA self-archiving
    enhances impact.

    (c) Like seat-belt mandates (Alma's Swan's apt analogy, which
    I recently misattributed to Arthur Sale, another admirer of the
    analogy who uses it too), self-archiving mandates are not mandating a
    charitable act on the part of the mandatee; they are mandating an act
    that is in the mandatee's self-interest, just as publishing itself is.

    (d) So institutional (or funder) self-archiving mandates are
    not adopted in order to free the rest of the world of a charge
    (although they do indirectly help in that regard too) but in order
    to enhance the visibility, usage and citation -- i.e. the impact --
    of the institution's (or funder's) own research output (thereby also
    enhancing its rewards to the the author/fundee).

In other words, self-archiving mandates are based on institution/funder
self-interest, not on philanthropy (or on solving libraries' journal
affordability problems -- though they indirectly ease those too, with
the back-up provided by the authors' self-archived supplements for
the articles in journals the library cannot afford).

We mandate self-archiving by mandating self-archiving. JISC surveys
found that 95% of authors report that they will comply, and when tested
by the four institutions that have implemented a self-archiving
mandate so far, this is confirmed in practice:

  http://www.ecs.soton.ac.uk/~harnad/Temp/Sotpolfiles/16surv-mand.jpg
  http://www.ecs.soton.ac.uk/~harnad/Temp/Sotpolfiles/22insts-w-mand.jpg

> 2. What sort of facilities should the place they post them at have?

The bottom line is a website, but BOAI rightly recommends OAI
interoperability, hence an OAI-compliant repository or archive.

> 3. Should that place have only published papers on it, or can they be mixed
> up with other things?

This is not the question at issue. The question at issue is whether
that place should be focussed specifically and urgently on reaching
100% OA content -- for which a specific mandate is the most important
prerequisite -- or whether it should proceed at a leisurely pace,
preserving and curating any digital contents that make their way into
the archive.

And the answer (quite apparent from both (i) the testimony in Richard
Poynder's article and (ii) the actual practice of institutional
repositories with OA self-archiving mandates versus those with diffuse
digital omnibus deposited ad lib) is that the diffuse digital omnibus does
not generate OA content above the 15% spontaneous self-archiving baseline.

In other words, it is not the content-mixing that is the problem, but the
lack of specific focus on acquiring the OA target content in the digital
omnibus archive. Librarians of course cannot mandate OA self-archiving:
Administrations need to do that. But if librarians are to collaborate
and advocate with OA, they need to be focussed on the target. (See the
excerpts from Richard Poynder's interview with Ann Okerson reproduced
below.)

In other words, the problem is not mixed content, but a mixed (or
non-existent) agenda for acquiring content, or even for what the content
is, and what needs to be archive for.

> The answer to the first question is, at least in the early years, that you
> need a dedicated staff member somewhere in the university - probably in the
> library - helping/encouraging academics to do it.

I regret to have to point out, Fytton, that that is most definitely not
the "answer" to the question (of how to generate institutional
OA self-archiving above its 15% spontaneous baseline level). Library
help/encouragement is (perhaps) a *necessary* condition for generating
OA self-archiving, but it is certainly not a *sufficient* condition (except
possible at CERN, where librarians can and do go after every single
full-text produced by their researchers, and deposit them all on behalf
of the authors).

    http://www.ecs.soton.ac.uk/~harnad/Temp/Sotpolfiles/18mand-no-ut.jpg
    http://www.ecs.soton.ac.uk/~harnad/Temp/Sotpolfiles/19mand-no-qu.jpg
    http://www.ecs.soton.ac.uk/~harnad/Temp/Sotpolfiles/20mand-yes-qut.jpg
    http://www.ecs.soton.ac.uk/~harnad/Temp/Sotpolfiles/23cern.jpg

The *sufficient* condition (including at CERN!) is the institutional
self-archiving mandate.

Please read Arthur Sale's excellent analyses of the necessary and
sufficient conditions for successful institutional OA self-archiving
policy:

     http://eprints.comp.utas.edu.au:81/archive/00000264/
     http://eprints.comp.utas.edu.au:81/archive/00000222/

And if you want to see how far library help alone will get you, see
St Andrews' well-meaning but alas insufficient library proxy
self-archiving service (<300 records):

    "St. Andrews University Eprint Archive" (Aug 2003)
    http://www.ecs.soton.ac.uk/~harnad/Hypermail/Amsci/2977.html
    http://eprints.st-andrews.ac.uk/proxy_archive.html
http://archives.eprints.org/?url=http%3A%2F%2Feprints.st-andrews.ac.uk%2F

Though, in fairness, St. Andrews may only have been offering the proxy
self-archiving services, not the advocacy.

> When we investigated
> costs of setting up IRs a couple of years ago, we found that, in addition to
> the relatively small costs recently listed here by Chris Gutteridge, the
> major cost in most institutions was the salary cost of this 'advocate'
> person.

That's for advocating without mandating. Here at Southampton ECS, we
mandated without advocating, without that added expense.

I invite you to compare our OA growth rate (for a mere department, with
a mandate but no paid library help or advocacy: >10,000 records)

http://archives.eprints.org/eprints.php?url=http%3A%2F%2Feprints.ecs.soton.ac.uk%2F

with that of our university as a whole (minus ECS), with library help
and advocacy, but no mandate: <10,000records.

http://archives.eprints.org/eprints.php?url=http%3A%2F%2Feprints.soton.ac.uk%2F

[Note that not all records are full-texts.]

There are signs, though, that Southampton may soon adopt a university-wide
mandate (in fact, the Heads of Departments are meeting to discuss it
this very day, as we speak!):

http://www.ecs.soton.ac.uk/~harnad/Temp/sotonSApolicy-linked.htm

I note that Loughborough's self-archiving growth rate could use a bit of
acceleration too (<500 records!)

http://archives.eprints.org/?url=http%3A%2F%2Fmagpie.lboro.ac.uk%2Fdspace

but I hear that Loughborough is contemplating a mandate too!

> To the second question, the answer is 'it depends how important you think
> proper metadata is'.  One can of course just use Google to look for an
> author's name and the article's title, and find papers available free of
> charge on the web that way.  But an OAI-PMH-compliant server will provide
> metadata that enables search services to locate papers reliably.  This is
> the old, old 'natural language versus controlled language' argument in new
> clothing.

Not quite! I am all for OAI-PMH, but that does not imply reversion to
paleolithic hand-classifying of each deposited item in a prefabricated
subject taxonomy! That is obsolete. What will emerge is automatic
AI-generated subject classification as an extra option, but that will
be derived from computational processing of the full-texts, not from
a priori hand-classification at deposit. (And I'm willing to bet that
boolean full-text search of a 100% OA/OAI target corpus -- as opposed
to the web as a whole -- will beat subject-based search hands-down any
day of the week! (Especially with the help of citation-based search:
CiteRank, co-citations, hubs/authorities, co-text, etc. etc.)

> To the third question, I guess the answer is that it is a matter of opinion.
> Stevan and others think it is very important that an OA IR providing
> published and to-be-published papers should provide nothing else.  My view
> is that this issue is unconnected with question 1.  If effective ways exist
> to persuade or coerce authors to post their papers on the IR, and the
> published or to-be-published papers are clearly and indelibly identified as
> such, it is irrelevant whether or not there is other stuff on the same
> server.

But I agree completely! It is not mixed IR content I am worried about,
it is mixed messages about what should be deposited and why, and what
the priorities are. If you tell people IRs are for preserving and
curating digital content, research article authors say (rightly)
"Ho-hum -- I won't bother, then. I'll leave that to the journals and
libraries, for my published articles."

And most of all, I worry about mixed -- or nixed -- library agendas
with respect to OA in particular. I append, with no further comment,
an excerpt from RP's article, noting only that with friends like AO,
OA hardly needs enemies! And it is for this reason that RP rightly
suggested that it might be time for OA to stop mixing company with the
digital omnibus curation/preservation agenda and make its own way:

    AO (Yale): "the library... is responsible for the university
    archive, and since the official documents of the university -- memos
    from departments and the president's office, videos, and recordings of
    university events -- are increasingly being developed electronically
    it is our responsibility to preserve all this... there are [also]
    a whole lot of things  being created by faculty for classroom use --
    syllabi, notes, outlines, lectures, images, illustrations etc. --
    that has to be archived somewhere... [there is also the need to
    deal with materials that are created digitally in the library]...
    texts or images, for instance..."

    [RP: But where in this picture does e-print archiving fit?]

    AO: "I presume that many faculty on our campus put their papers
    up on their web sites, and perhaps deposit them in a [central]
    preprint service," replies Okerson vaguely. "Certainly we haven't
    aggregated them in any way."

    Okerson estimated three-year start-up costs for hardware and software
    alone for the kind of institutional repository Yale has in mind at
    over $300,000.

Stevan Harnad
American Scientist Open Access Forum
http://amsci-forum.amsci.org/archives/American-Scientist-Open-Access-Forum.html