Re: [EuroscienceOA] A personal view about institutional archives

From: Stevan Harnad <>
Date: Sun, 18 Dec 2005 00:55:06 +0000 (GMT)

On Sat, 17 Dec 2005, Pierre.Baruch wrote:

> This choice may call for some explanations, linked to French idiosyncrasies.
> Why go through these huge institutions, better than through
> local entities, universities or institutes?
> As a first explanation, I quote the extreme dispersion of the French
> research system.
> We have a symbiotic co-existence of agencies and universities, meaning that
> a single laboratory is, in many cases, simultaneously affiliated to more
> than one of these bodies

The only question of any relevance at all is whether the proposed French
centralised self-archiving system manages to capture French research output.
The global average in self-archiving is now 15%. That is the figure to beat.
But the target is 100%. And the only entities that are well on the road toward
self-archiving 100% of their annual research output are the four universities
(Minho, Southampton ECS, and soon Zurich) and one research institution (CERN) that
have adopted a policy of *requiring* self-archiving. The other institutions
that have merely provided archives and merely *recommended* self-archiving are
doing far less well:

Arxiv is a notable exception: It is not really an institution at all. It
is merely a substantial subset of the world physics community. And the
reason for its high self-archiving rate is that its contributors had
already been doing something very much like self-archiving even before
the online era, and merely carried over and optimised their existing
practice on the Web. But after nearly 15 years of providing their
practice of spontaneous self-archiving as an example to the world,
there is no sign that other disciplines are following their example.
It is hence quite clear by now that the *spontaneous* self-archiving
model, whether central or local, does not generalise or propagate

That leaves mandated self-archiving. And a nebulous entity like an archive
or a "discipline" cannot mandate self-archiving: Only researchers'
institutions or funders can mandate it. And in most countries,
researchers' institutions are local entities (largely universities,
plus research institutes). Research funders can (and do mandate) too,
but as they can only mandate the archiving of the research they fund,
their best chance of making the effects of their self-archiving mandates
to propagate across researchers, institutions and disciplines is to
mandate that their fundees self-archive in their own local institutional
archives ("institutional repositories", IRs). Institutions share with
their own researchers (employees) the benefits of maximising the usage
and impact of their joint research output. They are also in the position
to monitor and reward compliance with institutional self-archiving
policy. These were the conclusions of a British (JISC) study of the
relative merits of central/national/disciplinary self-archiving vs
distributed/local/institutional self-archiving:

    Swan, A., Needham, P., Probets, S., Muir, A., Oppenheim, C., O?Brien,
    A., Hardy, R., Rowland, F. and Brown, S. (2005) Developing a model
    for e-prints and open access journal content in UK further and higher
    education. Learned Publishing 18(1) pp. 25-40.

But the decentralised case of the UK may well differ from that of much
more centralised France; and the critical factor (in the OAI-interoperable
age) is almost certainly not central vs. local self-archiving but
required vs. merely recommended self-archiving.

> universities... are not presently equipped to create and maintain
> their own archives... Most universities are weak, underfunded, understaffed,
> subject to local or national political pressures. I am not sure they can
> provide the continuity and permanence required.

I cannot judge how accurate this a-priorism is, given that the cost of creating
and maintaining an OA IR is only about 10,000 euros, but let me repeat: The
critical element is not central/local but required/recommended -- and
the measure of success is the rate at which self-archiving approaches
100% of annual research output. The target for an IR is its own institutional
research output. The target for a central or national archive is necessarily
the total output of all the researchers involved, if the comparison is to be
valid and informative.

> and one solution is to have
> archives at a national level, for each of the main research councils (CNRS,
> INSERM for medical research, INRA for agronomy, etc), through agreements
> between all the actors. This is precisely what is being set up. The usual
> linkages, common protocols, should insure that the system look transparent
> to users, such as building virtual local archives. but it will take time to
> arrive to full operation.

If self-archiving is mandated, it will succeed; if it is not, success is
less likely. Success is measured (and compared with other approaches) by
calculating the percentage of the total annual research output that is
being self-archived, and the annual rate of growth of this percentage.

> Now some personal remarks : I am perfectly indifferent to organisational
> details, but I feel more at ease with disciplinary archives, such as ArXiv
> for physics (I am a physicist), being closer to colleagues working on the
> same problem on the other side of the planet than with the chap on next
> floor, busy on a totally different subject - which does not prevent to have
> friendly relations. This may sound parochial, and the system may allow each
> user to build his own configuration, but when ? while ArXiv is already in
> operation.

Arxiv has been in operation for nearly a decade and a half, and for those who
use it, it works. and they have become accustomed to it. But the spontaneous
self-archiving practices that created Arxiv do not have counterparts in other
disciplines, and their example has not generalised. Moreover, another
discipline with a high rate of spontaneous self-archiving -- computer science,
which has been doing it even longer than physics and has a larger total number
of papers self-archived -- has been doing it on a local distributed basis,
harvested since the late 1990's into a *virtual* central archive, Citeseer.

But spontaneous local self-archiving practices have not propagated either.
So mandates seem to be necessary, if we are not to go on waiting for the
obvious, optimal and inevitable outcome for many more years to come.

> Another point : an institutional (university) archive gives to the
> administration a power for watching the activity of its members, evaluating
> it and using the device for management of careers - in the extreme, using
> automated methods for promotions. Another avatar of Big Brother?

I think this is altogether beside the point: Institutions will continue
to evaluate the performance of their researchers irrespective of whether
there is or is not 100% self-archiving and 100% OA. The only connection
is that the (unchanged) institutional evaluation ("publish or perish")
incentive and reward system -- which is *already* based not just on
the quantity of publication, but also its quality and importance, as
measured in part by its usage and (citation) impact -- can be used
to incentivise self-archiving, by mandating it, to the benefit of
researchers, of research usage and impact, and of research itself. If
this is Big Brother, He is a very benign One (or no more malign than the
"publish or perish" mandate that keeps researchers from putting their
results in desk-drawers -- or not bothering to seek results at all).

> One may object that this may happen with any type of archives, distributed,
> centralized, or disciplinary, as long as they are correctly linked. Who
> will build safeguards ?

Of course all OA archives will be OAI-compliant and interoperable, and usage
and impact will be measured and rewarded globally. What "safeguards" are at

> Shall we discuss sometime the psychology and ethics of the new modes
> of dissemination of science ?

Yes, but first let's get 100% of research output OA, whether centrally
or locally!

Bets wishes,

Stevan Harnad

EuroscienceOA mailing list
Received on Sun Dec 18 2005 - 02:19:59 GMT

This archive was generated by hypermail 2.3.0 : Fri Dec 10 2010 - 19:48:09 GMT