Re: NIH's Public Archive for the Refereed Literature: PUBMED CENTRAL

From: Stevan Harnad <>
Date: Thu, 23 Sep 1999 09:57:09 +0100

On Thu, 23 Sep 1999, Frank Norman wrote:

> Mark Doyle said:
> > There are two big hurdles: 1) reducing the cost of handling electronic
> > manuscripts and 2) author/institution/funding agency acceptance of
> > paying submission fees up front.
> Regarding 2), I have heard some concerns that this would unfairly
> penalise the most productive institutions/departments. Institution
> A producing 500 research papers p.a. will be liable for twice the
> level of charges of less effective Institution B producing only 250
> papers p.a. This may make the management of Institution A less
> than enthusiastic about such a move towards submssion charges.
> I'd welcome any further insights on this.

See the archives below. There are threads on this issue of "net
provider" vs. "net consumer" libraries and research institutions.

Someone will have to do the arithmetic, but for the moment it has a big
fudge factor in it: What will the page charges actually turn out to be,
per paper, once journal publication has down-sized to providing the
service of Quality-Control/Certification (QC/C) alone?

Andrew Odlyzko estimates about $10-$20 per page.:

    Odlyzko, A.M. (1998) The economics of electronic journals. In:
    Ekman R. and Quandt, R. (Eds) Technology and Scholarly
    Communication Univ. Calif. Press, 1998.

The Journal of High Energy Physics -- which is way ahead of us all in
this, and is in the best position to estimate, being in a field already
softened up for 9 years by LANL self-archiving, and being a refereed,
online-only journal that has already attained what has been informally
estimated as an impact factor of ELEVEN (sic) in what is now only its
thrid year -- estimates that, factoring out one-time start-up costs
that they have already borne, and scaling up to the many journals that
they could handle at much the same cost, rather than just the one they
do currently, the cost per article is about $300.

These figures agree. Now do the arithmetic, adding up the total current
serials budgets of "net provider" institutional libraries like Harvard
(which generate more papers than they consume) and "net consumer"
libraries that consume more than they generate. Call that grand amount

First of all, it will turn out that the total QC/C cost (averaged
across net providers and net consumers) will be less (perhaps
considerably less) than 1/3 of the current
Subscription/Site-License/Pay-Per-View (S/L/P) expenditure ($NN),
namely, $NN/3.

Now, how does this translate to the individual institution level? Does
Harvard publish so much more than it currently consumes via S/L/P that
it will not be about 2/3 better off, like everyone else? I think 1/3 is a
sufficient buffer so the answer is that it WILL be much better off than
it is now, with its current S/L/P budget. It might not have as much
windfall saving as the less research-active, net-consumer-instititions
(which may well do better than 2/3).

But there are benefits to all that research publication (grants,
prestige, prizes, impact), and the freeing of the research literature
itself will benefit research-active "consumers" much more than inactive
ones (although the enhanced access may also help draw the consumers
into greater productivity -- this will certainly be true in the third

So the arithmetic has to be done, and the causality has to be thought
through, and not narrowly, but broadly, to include all the
ramifications of learned research impact for a learned research

Stevan Harnad
Professor of Cognitive Science
Department of Electronics and phone: +44 23-80 592-582
Computer Science fax: +44 23-80 592-865
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 - 19:45:37 GMT