Re: Central vs. Distributed Archives

From: Stevan Harnad <>
Date: Tue, 29 Jun 1999 16:32:55 +0100

On Tue, 29 Jun 1999, J.W.T.Smith wrote:

> A monopoly in the sense that it could become 'the place' where readers
> look for items relevant to their subject. The non-presence of an article
> in a recognised subject specific archive could imply it is not relevant to
> the subject. More on this later.

Papyrocentric thinking. We live in the era of metadata tagging and
search engines that trawl it all.

> I am not concerned with its availability, I am
> concerned with the implied validation of the presence of an item in a
> given archive.

Don't be. The validator is the journal, as it always was. The Archive is
only the free cosmic bookshelf in the Sky...

> Even if the archive is mirrored it is a mirror of somewhere
> and the address of that somewhere has value. If this has no value why to
> we need an archive at all? Why don't we all mount our papers on our
> University servers?

We should! That was the gist of my 1994 Subversive Proposal:

But there are currently still interoperability problems with
institutional servers, so the colossal success of Los Alamos has shown
that we will reach the optimal and inevitable faster by taking both
routes, the centralised and the distributed one:

> There are two advantages that I can see of a subject
> specific archive:
> - It can be properly maintained (it is a true archive)
> - It can be a 'one stop shop' of where to look for items on a specific
> subject.
> I have no problem with the first role. It is the second that carries the
> possibility of monopoly. As long as the archive is maintained by a neutral
> organisation (like a large University) this is OK but what if it should
> become privatised?

EVERYTHING runs the risk of being "privatized": Universities, Los
Alamos, NIH. Fighting against the privatization-frenzy in whose grip the
entire planet seems to be at the moment is a worthy enough mission, but
it is completely irrelevant to the centralization/monopoly red herring
that I believe you are preoccupied with -- for the simple reason that
the menace of privatization is completely nonspecific, and afflicts ALL
options, in principle.

In practise, I would not worry too much about a hostile take-over of NIH
by the private sector in the near future, nor about NSF tossing the
Los Alamos Archive to the Trade Winds. Besides, one of the STRENGTHS of
"centralization" is that the authors that have put their precious eggs
in the collective basket and the users who forage them tend to monitor
them zealously day and night, and are likely to squack vociferously if
they sense any threat:

    Taubes, Gary. E-mail withdrawal prompts spasm. (temporary
    shut-down of Los Alamos Laboratory e-print archives succeeds in
    raising funds) Science v262, n5131 (Oct 8, 1993):173 (2 pages).

    ABSTRACT: Paul Ginsparg shut down the e-print archives of Los
    Alamos National Laboratory, the physicists' pre-publication
    bulletin board for a few days. The closure incited users to
    petition the Department of Energy and National Science Foundation
    for funds and secured official funding from Los Alamos.

> Once an archive (or its mirrors) is seen as 'the place'
> to search for items of interest and access to that archive can be
> controlled it might be temping to place some restriction on access like
> payment of a fee (for purely reasonable reasons like getting enough money
> to maintain the archive).

A lot of other networked services are likely to get a price tag before
the tiny refereed literature archive is likely to: It is the flea on the
tail of the dog, and we will all be best served if it is given a free
ride. Again, this worry is papyrocentric and misplaced.

> Now I know the actual quality control/validation
> is provided elsewhere (maybe by the 'old' journals, maybe by other
> players) but from the point of view of the author they may also need to be
> in the archive as well as have the validation/stamp of approval of an
> external organisation.

This sentence was a bit difficult to decode, but from what I can make of
it, one entity (the established journals -- why on earth not?) can
continue to do the quality controlling and certification-tagging, and
another (new, virtual) one, the Archive, can provide free access to the

What is the problem?

> > As I have noted before, this central/distributed issue is a red
> > herring, based in part on papyrocentric thinking (we are in reality
> > talking about a distributed virtual library where locus has little
> > meaning)
> You seem to contradict yourself here. If 'locus' (I don't mean physical
> position) has no meaning why do we need a Physics archive, or a Biomed
> archive, or any other subject archive? Why can't we either have one
> universal archive which simply stores and serves on request (at no cost
> and forever) any item sent to it, or no archive at all with items being
> stored on a user site or a University site or a commercial site (or all
> three or some other option/permutation)?

We can have all of it, and nothing substantive is at issue here; this
is mere pseudo-territoriality where there is no call for it and no
sense in it. Vide supra, regarding subverting the candle from both

> > > Summary: It is possible to escape the problems of the 'trade model' of
> > > current academic publishing without running headlong into the possibly
> > > equally constraining model of a monopolistic central archive.
> > Yes. Change the vocabulary.
> Why don't you drop the word 'journal' then? Why not use 'validator' or
> some other word that indicates the role and doesn't carry over
> connotations from the old "papyrocentric" model?

Suit yourself. But I think "Physical Review Letters" will continue to
prefer to call itself by its current familiar and trusted brand name --
and why on earth shouldn't it?

Stevan Harnad
Professor of Cognitive Science
Department of Electronics and phone: +44 2380 592-582
Computer Science fax: +44 2380 592-865
University of Southampton
Highfield, Southampton
Received on Wed Feb 10 1999 - 19:17:43 GMT

This archive was generated by hypermail 2.3.0 : Fri Dec 10 2010 - 19:45:33 GMT