On Wed, 14 Aug 2002, Jan Velterop (BioMed Central) wrote:

> Research articles must be peer-reviewed

Research articles must be peer-reviewed by a "peer-review service provider
and certifier" (also known as a "refereed journal") with an established
and known quality standard. Peer review is not generic. The literature is
and needs to be sign-posted for quality, as it is now, in the hierarchy
of journals varying in their degrees of peer-review rigor and hence their
quality levels so that papers can be weighed by the known quality-level
(often correlated also with the impact factor) of the journal in which
they were published. This filtration and the resulting sign-posted
quality landscape are created in the service of navigation by the
reader/researcher, application by the user/researcher, and evaluation
by the research institution/funder. Journals have to be plural,
independent, and of known quality standard.

> Copyright must not be allowed to play a role in the business model

In the special case of this anomalous, give-away literature:
peer-reviewed research.

Copyright in the open-access era (when all peer-reviewed research is
freely available online to all would-be users) is retained by the
give-away author and used only to protect the authorship and the
integrity of the text. (It was only part of the "business model" of the
toll-access era because it was needed to protect publishers' toll-access
revenues. With open-access, there are no longer toll-access revenues,
hence peer-review service-provision costs must be covered another way.
The most natural way, it seems to me, is out of institutional toll-access
windfall savings, if/when they occur.)

> We do not currently charge for peer review if the article is not accepted.

This is quite reasonable. This is the model where the peer-review
service costs for non-accepted articles are factored into the
peer-review service costs for the accepted articles. For a
high-rejection-rate journal (usually correlated with high quality) this
can mean that the accepted 5% are paying for the processing of the
rejected 95%.

There is nothing wrong with this, and this is what keeps peer review
costs at between $50 and $500 per accepted paper, depending on the
journal, its submission rate, acceptance rate, and quality standards
(though the actual intercorrelations have not been analyzed and might
prove surprising).

An alternative model that might be tested is to charge a smaller
submission fee per paper as a flat rate, to be credited toward the
full peer-review service fee should the paper be accepted. This could help
discourage the author practice of submitting the same paper to a
descending series of journals, taking up their refereeing resources
and time, until the paper finds a journal at a quality-level willing
to accept it: Submission costs might encourage authors to go from the
outset directly to the journal at the appropriate quality level (and to
take revision according to referees' recommendations, resubmission and
re-refereeing more seriously).

> There are costs to organising peer review, but we have been, and still
> are, building tools to keep those costs to a minimum.

The costs are indeed minimal, especially if the entire process is
implemented online; but they are non-zero, hence need to be recovered. And
as implementing the peer-review is the only journal publisher service
whose necessity is still certain in the open-access era, it is important
to make sure that the coverage of the peer-review service costs is
ensured by the "business model."

With all the other traditional publication functions it is not at all clear
which of them -- if any at all -- is still essential in the open-access
era, particularly on the assumption that the institutions that will be
covering the peer-review service costs for the research output of their
researchers will also be self-archiving the pre-refereeing preprints and
the post-refereeing postprints.

In a distributed institutional archiving system like that, interlinked
and interoperable through compliance with the OAI-protocol, it is clear
that the hard-copy edition provided by the publisher is no longer an
essential. Nor is it clear whether any mark-up services are still needed,
with authors already preparing so many of the phases of the digital
text themselves. XML templates and software editors are on the way. As
much substantive editing as is needed is already part of the peer-review
cost. Perhaps a little copy-editing needs to be factored into that
cost as well.

But are there any other essentials?

A proxy self-archiving service of the kind Ingenta is planning to provide
-- http://www.ecs.soton.ac.uk/~harnad/Hypermail/Amsci/2108.html --
is obviously a potential value, but it is clearly an option and not an
essential; it is only for those institutions that prefer to outsource
their self-archiving.

BioMed Central is considering providing such a proxy self-archiving
service too, and that is to be applauded.

But, again, is there any OTHER essential service, over and above peer
review (with or without the optional proxy self-archiving service)?

> The articles we publish are not just marked up and converted
> into HTML and PDF and put online (and made available via Avant-Go),
> they are also firmly and actively embedded in the web-like structure of
> the scientific literature, among others via CrossRef linking, indexing
> in PubMed, Medline, BIOSIS, and other services, rather than just relying
> on them being found by search engines or harvesters (all our material is
> OAI-compliant)....
>
> Articles of exceptional interest are also press-released to a wide and
> global variety of specialist and general media, giving the authors maximum
> exposure.

I would like to make a (confident) prediction. The classical indexing
services are going to become superfluous and vanish as soon as enough
of the literature becomes open-access, and the alerting/visibility
functions they used to perform will be completely taken over by OAI
harvesters like OAIster and ARC, which will simply be the Googles of the
peer-reviewed literature. (Perhaps Google searches themselves will be
limitable to the peer-reviewed, OAI-tagged corpus.) The main sign-posting
function will continue to be performed by the "JOURNALNAME" tag, just
as it always has been, specifying the known peer-review standard that
that paper has met. And, unlike in most of today's indexing services, the
full-text will be accessible too (and will also be full-text inverted
and searchable), likewise at no cost.

Making the obsolescence of the old forms of alerting/visibility
complete will be new OSI scientometric search-engines like citebase --
http://citebase.eprints.org/cgi-bin/search -- which will provide newer and
far richer online performance indicators for the entire citation-linked
open access corpus, ranking papers, authors and journals on a variety
of measures, including citation impact and download impact.
http://www.ecs.soton.ac.uk/~harnad/Tp/thes1.html

So I doubt that alerting/visibility services will be among the
essentials provided by either primary journal publishers of secondary
index publishers for much longer.

> We do more than
> just archive, but our archiving 'component' is typically organised along
> disciplinary lines rather than geographical ones, the latter being
> the case for the typical institutional self-archiving facility. With
> the majority of articles in the life sciences and medicine written by
> multiple authors with multiple affiliations, this approach offers some
> advantages over institutional archives, although sophisticated indexing
> and OAI-compatibility of the material levels the differences to a degree.

I couldn't follow this. The point of OAI-compliant archiving is that
it no longer makes any difference which archive a paper happens to be
in. If it has the right OAI tags (e.g., journalname) it can be
harvested, searched, citation-linked, citation-ranked and retrieved as
if everything were in one global virtual archive. The tags can also be used
to partition the archives or the searches along any lines one may wish
-- disciplinary, geographic, or institutional.

> The fees we charge for our services (the APCs) are, on a per-article
> basis, materially lower (up to a factor 10) than the aggregate cost per
> article to academia of conventionally published material.

The amount that institutions will be willing to pay will depend on what
they feel they need. They will always need peer review. They can do
their own self-archiving or they can outsource it. It's not clear,
though, whether they will need or want to pay for anything else.

> We are finding,
> however, that scientists are rarely concerned with that. They do certainly
> not seem concerned if their institution picks up the bill. What they
> value most in open access journals is the vastly increased visibility and
> findability of their articles, leading to increased use and, reportedly,
> to increased citations to their articles.

Which is a good reason for scientists to do the one sure thing that
is in their own hands, right now, namely, to self-archive their own
research output in their institutional Eprint Archives (whether self-run
or outsourced). That is the primary way to maximize visibility and
retrievability. The rest will take care of itself, as this open-access
corpus grows.

> Because the new open access journals do not have an impact factor (IF)
> assigned by the ISI impact factory yet (they are too young to have one;
> the formula calls for several years of track record), some authors
> are reluctant to trade off prestige (which comes with publishing in a
> journal with a reasonable IF) for increased exposure. They need this
> prestige in order to get tenure or grants, as many tenure committees
> and funding bodies are not yet ready to recognise articles published
> in non-conventional journals. This is likely to change once there is
> a track record of the new open access journals that enables IFs to be
> calculated. Early indications of citations to articles in the open access
> journals published by BioMed Central point to very good IFs in due course.

There is nothing unconventional about open-access journals, scientifically
speaking. One of the biggest of them -- and only 3 years old: JHEP
http://jhep.sissa.it/ --  already has an impact factor higher than
that of most other journals of whatever kind -- and we can be 100%
confident that its authors have no problem whatsoever in getting
those articles recognized and duly credited by tenure
and grant committees! (Alas, JHEP has recently reverted to
toll-access to cover its costs, fearing, apparently, to try the
author-institution peer-review service cost model, and preferring
to recover expenses through the traditional toll-access model:
http://www.ecs.soton.ac.uk/~harnad/Hypermail/Amsci/1812.html ).

But most new open-access journals are not JHEP, and are struggling with
what all new journals, lacking established track records, must struggle
with (though perhaps the online-only medium -- NOT the cost-recovery model
-- adds an extra handicap).

The increased visibility and impact provided by open-access may help to
surmount these start-up barriers for the viable ones among these new
journals, but it will take time. (Meanwhile, however, self-archiving
can provide immediate open-access to everything without any start-up
barriers to overcome.)

> The changes open access imposes on the conventional publishing edifice
> are fundamental, and there will always be scepticism and reluctance to
> embrace new models that are so far-reaching. Fortunately, however,
> there is a growing corpus of authors who see the importance, benefits
> and opportunities of open access and who are willing to 'blaze the
> trail'. 'Established' scientists are in the best position to give open
> access a boost; their careers don't depend on impact factors. We are
> engaged in efforts to reach especially these echelons of authors.

All true, but remember that there are two paths to open access, BOAI
Strategy 1 (self-archiving) and BOAI Strategy 2 (open-access journals),
and the first does not require embracing new models...

Stevan Harnad