Association of Research Libraries (ARL®)

Does the responsibility for scientific and scholarly findings lie at the grass roots with individual scholars or should there be institutionalization and centralization -- or both? Nobel Prize Winner Joshua Lederberg, looking to the practical uses of more and better information that the scientist can use, introduces the idea of institutional rather than discipline-based archives. From the library community, encouragement to recover some control over the economic fate of faculty products; then discussion of the place of the large learned societies in the publishing landscape ensues.

Date: Tue, 28 Jun 94 11:53:03 EDT
From: "Stevan Harnad" harnad@princeton.edu

Date: Tue, 28 Jun 94 00:26:17 -0400
From: Joshua Lederberg jsl@rockvax.rockefeller.edu

Dear Steve:
Now we face some strategy decisions. The new wrinkle that I got from
what (I thought) you said was to bypass waiting for the disciplines
to organize themselves; instead let each institution set up its own
ftp-able archives for all of its scholars. That way, each place can
also set up its own ground rules. Let the disciplines then come into
play ad lib with peer review and those embellishments at their own
pace and microstructure. Physics was a fairly homogeneous ensemble:
just look at the journal structure compared to biomedicine. For the
latter, there is scarcely consensus how to classify its subjects.

Institutions also compete with one another, so there is likely to
be constructive emulation of the pioneers; and there is some tacit
quality labelling just by the name of the institution.

Josh

That is exactly right. The grass-roots initiative should be at the individual scholars' end: everyone should establish a personal ftp archive, starting now...

But there's no harm (and a lot of good) in working from both ends (actually, many ends) at once, and what the two Pauls' respective projects (HEP and CIC) are doing is working on the centralized repository aspect, which will also be very helpful in getting the dominoes to fall.

As soon as a researcher establishes a personal ftp archive, it is on the global map. But what the two Pauls' projects will do is make the map all the more rationally structured and navigable. A veronica (or www) search on "lederberg" would already point to your personal ftp archive, but browsing a HEP or CIC catalogue, especially once hierarchical classification systems -- plus the all-important quality-control (peer review) tagging -- are implemented, will be extremely valuable too.

So let us encourage private archiving AND centralized archiving projects. Eventually, a rational method of automatic link-up and transfer will surely evolve too.

Stevan Harnad

Date: Tue, 28 Jun 94 10:43:31 -0600
From: Paul Ginsparg 505-667-7353 ginsparg@qfwfq.lanl.gov

Date: Tue, 28 Jun 94 00:26:17 -0400
From: Joshua Lederberg jsl@rockvax.rockefeller.edu

Institutions also compete with one another, so there is likely to
be constructive emulation of the pioneers; and there is some tacit
quality labelling just by the name of the institution.

sh> That is exactly right. The grass-roots initiative should be at the
sh> individual scholars' end: everyone should establish a personal ftp
sh> archive, starting now...

sh> But there's no harm (and a lot of good) in working from both ends
sh> (actually, many ends) at once, and what the two Pauls' respective
sh> projects (HEP and CIC) are doing is working on the centralized
sh> repository aspect, which will also be very helpful in getting the
sh> dominoes to fall.
sh> So let us encourage private archiving AND centralized archiving projects.
sh> Eventually, a rational method of automatic link-up and transfer will
sh> surely evolve too.

i have been through this argument many times. when i set up the first physics archives three years ago, it was clear that a distributed database, with only the indexing and pointers stored centrally, would be in principle much more desirable than a centralized one. the problem is that it would have been too restrictive back then to require even that every physicist have access to his/her own anon ftp site (much less gopher/www server). while the community may appear "homogeneous" from the exterior, it is nonetheless a global community with a wide variety of interests and wide variation in computer literacy and network access. (our indian, chinese, and russian colleagues have only recently established full internet access -- but still lagging in their local computational infrastructure, high quality workstations and printers rare i'm told). moreover a significant percentage of papers are written by transient grad students / postdocs / junior faculty. were they expected to establish and maintain control over their personal portable archives? if the anon ftp server was left to the institutional department or library to maintain, how could author maintain control after leaving? how many dept's have the competence/staffing to maintain such servers?

there were just too many unknowns had i not begun with a "lowest common denominator" e-mail interface and centralized archive, this never would have gotten off the ground. (and indeed earlier efforts e.g. in mathematics were stillborn for precisely the above reasons) now with platform for training the userbase, i can implement higher level functionality (e.g. the www interface) while continuing to democratically support the lowest common denominator. i estimate that it will still be years before a full distributed database will be logistically feasible (as opposed to technically feasible, which it already is), but a migration has begun where people insert pointers to additional resources (jpegs/mpegs/software/postscript) at their local ftp/gopher/www sites when available (and any http:... in title/abs is automatically converted to a link for www interface). people who do not have local archiving capability have recourse to the centralized database which automatically guarantees stable storage and high bandwidth network access through a variety of protocols at centralized site (and its mirrors).

Paul Ginsparg

Date: Tue, 28 Jun 1994 12:58:59 -0400 (EDT)
From: Paul Southworth pauls@locust.cic.net

On Tue, 28 Jun 1994, Paul Ginsparg 505-667-7353 wrote:

Date: Tue, 28 Jun 94 11:53:03 EDT
From: "Stevan Harnad" harnad@princeton.edu

Eventually, a rational method of automatic link-up and transfer will
surely evolve too.

i have been through this argument many times. when i set up the first physics
archives three years ago, it was clear that a distributed database,
with only the indexing and pointers stored centrally,
would be in principle much more desirable than a centralized one.

Obviously a centralized point of access for the end-user is desirable. The issue of local archives is really a matter of how to handle distributed maintenance by humans in different locations. We could possibly (a) give out accounts to maintainers on a central server and delegate areas to them completely, or (b) we could also evaluate wide-area distributed file systems (such as AFS) as a foundation for the server. Those are both alternatives (with their own problems) to relying on multiple servers. One other possibility, since I suspect we will be dealing with a number of specific highly-motivated experts in each area, would be to house subject-specific servers on location with their caretakers who would be responsible for coordinating with other archivists working on the same subject area (ie, physics) so as to have a handful of servers that we link to from the central access point, rather than thousands of them, one for each professor (in the worst case). Then we could approach organizations with a proposal for funding specific subject area servers, get them to agree to a standard for what type of server interfaces to support and how they should look, and then link them all up with a central access point at the server-application level (ie, http and gopher). Central access to distributed archives via ftp is not really possible unless we go the AFS route.

Date: Tue, 28 Jun 94 17:13:13 EDT
From: "Stevan Harnad"

From: Ann Okerson ann@cni.org
Date: Mon, 27 Jun 1994 21:55:51 -0400 (EDT)

Stevan,

Thank you for forwarding me the two messages, though I don't really
know what they are about (in context).

It was an abstract of a talk I'm to give in London in November, but it served a double purpose: I also branched it to other interested parties (as usual), including a UNESCO group headed by Joshua Lederberg that is interested in making the scientific literature available worldwide.

However, you may be interested to know that one outcome of the AAU task
force reports (Association of American Universities) is a recommendation
from the Intellectual Property Task Force (the one I worked on) to
involve universities more in publishing the works of their faculty
electronically.

Bravo. But, as you know, "publishing" is an ambiguous term these days. There are two distinct questions here: Publishing of the refereed literature, and "publishing" of the preprint literature. My subversive proposal was intended to make the latter break down the doors for the former.

All help is welcome, but I find the present path to direct electronic refereed publication too slow, even with the good will and help of universities and libraries. My proposal would have all scholars AUTO-"publish" their preprints, in personal ftp archives at their institutions, and then the subversive step comes as they swap the refereed reprint for the preprint. (I may be wrong, and it may not do the trick, but there is a good chance it might; so please pause to note that this is NOT the same as having universities and libraries get directly into the publishing game. They are welcome to help now, but they will more probably be needed waiting in the wings when the dominoes fall, just in case the paper publishers are not ready to retool themselves to take up the slack. [How's that for a hopelessly miscegenated metaphor?])

At this point a couple of the universities are interested in doing
a prototype in which the works of their faculty go up on the Internet
associated with that university. If the works are subsequently
published in print or in a more formal way, the version can either
be replaced or can point to the more formal one -- details to be
worked out. It is not unlike the preprint model.

The devil is in the details. Before refereeing and publication, they are preprints, the author's property. After refereeing, acceptance and publication, they are (in virtue of copyright assignment), the publisher's property, and there immediately arises the critical question of whether a price-tag will be allowed to intercede between the work and its readers. Call it what you like, before, it's preprints, after, it's (usually copyrighted) reprints. And my subversive proposal is intended to shake off for ever the shackles of the pay-to-see model that reigns now. By the way, most ftp archives are ALREADY associated with the author's university, so we're on the same wave-length there. But I think the initiative will have to come from the author community, initially on an individual basis (though backed up by the electronic resources of the author's institution). When a critical mass is reached (as it has with the HEP community), then Universities and Libraries can be poised to jump in if publishers are not willing or able to do the right thing.

I imagine we will have a meeting early in the fall (summer is too
fragmented with vacations and stuff) of the interested parties to
start to hammer out details. Or at least, I hope so. It seems to me
that with key, well-endowed American institutions trying this and
supporting it, we can prove that it can succeed. Or if we fail,
we will have a better understanding of how/why. I'm very excited
about this potential.

I agree that it will be helpful to have concerted efforts -- but concerted at what? If the Universities act preemptively (i.e., before a critical preprint inertial mass is formed) they will make publishers act preemptively, and authors will be intimidated ("If you publicize your preprint electronically, we will not consider it for publication; and if you make it available electronically after publication, you will be in violation of copyright.") That's why I think a natural subversive process rather than a premature formal confrontation might be the best.

Forward to Chapter IV

Backward to Chapter II