Access, Dissemination, and Peer-Review Costs

From: Stevan Harnad <>
Date: Thu, 3 Jan 2002 23:26:58 +0000

On Wed, 2 Jan 2002, Peter Suber wrote:
> Re: Excerpts from FOS Newsletter
> Dissemination fees, access fees, and the double payment problem

> First let's distinguish access fees from dissemination fees. Access fees
> pay for access. Subscriptions and licenses are the primary examples but
> not the only ones. There are secondary examples in micropayments and other
> forms of pay-per-view. By definition, access fees make access unfree. So
> if we want free online access, then we must find a way around access fees.

It is for this reason that for several years now I have telescoped all
three forms of access-blocking tolls under the portmanteau of "S/L/P"
tolls. It blocks the temptation to imagine that one can be the remedy
for the other!

However, Peter's partition is not quite complete: One can indeed
separate the costs of access and the costs of dissemination. But there
is a third cost, the cost of quality-control (peer-review), which is
not part of either cost, but is typically recovered through one or the
other (S/L/P access fees or dissemination fees -- mostly the first).

In addition, Peter describes the partition as if all the costs were
essential, and hence that the same amount must be recovered one way or
the other. This is incorrect. The access model is a PRODUCT acquisition
model: The product is the text, and that is what the S/L/P tolls
pay for accessing (either on-paper or on-line).

But there is an alternative to this if we consider only the on-line
text (as we should): The on-line text is (1) an OUTGOING product (and a
give-away one, at that), FROM the author (more accurately, from the
author-institution), rather than (2) an INCOMING product, TO the
reader-institution, as on the S/L/P access model. Let us call the model
in which the outgoing product is free (1) a "service" model, where the
service purchased is quality-control (peer review), and it is purchased
by the author-institution, as a 3rd party evaluation, to inspect,
improve and certify the quality of its (give-away) outgoing product:
its (peer-reviewed) research papers.

What about dissemination then? On an outgoing, service-based model, the
(free) product-provider-institution pays for the quality-control
service and provides the "dissemination" too. But what does this
dissemination amount to? Nothing more that the Web-archiving of the
quality-controlled product, freely accessible to all!

The token that has to drop in our minds if we are to understand this
PostGutenberg reality is that institutional self-archiving DISTRIBUTES
the "dissemination" burden across provider-institutions by covering it
for each outgoing article AT SOURCE, instead of bundling products
willy-nilly from disparate institutions and trying to "disseminate"
them all jointly as a 3rd party venture, as Gutenberg journals did.

That 3rd-party bundling was necessitated by the on-paper
dissemination-medium. But the PostGutenberg (on-line)
dissemination-medium no longer requires it at all! Gutenberg
dissemination, if you like, was analog and dynamic. It required
physically distributing a physical object worldwide. PostGutenberg
dissemination is digital and "static": The provider need merely archive
the text on the Web. Search engines, harvesting, alerting, and user
downloading take care of the rest (provided the text is made
interoperable, hence harvestable, through OAI-compliance).

So it is misleading to speak of access costs versus dissemination
costs. Dissemination costs shrink to almost zero (the archiving cost
per article is virtually zero) if the "dissemination" is done at
source, by the provider-institution (i.e., the researcher's own
university), leaving only the essential quality-control cost (per
article) to be paid for -- likewise by the provider-institution.

Nor is this apparent double-dipping the unjustified burden to the
long-suffering provider institution (which is already providing the
research for free, and always has been!) that it sounds like on first
hearing! For the provider-institution's annual windfall savings from
the huge and needless expenditures for S/L/P access tolls (once they no
longer need to be paid, because all costs have been covered at source)
will be more than enough to cover the much smaller quality-control (and
virtually-zero "dissemination) costs (at source)!

> Dissemination fees pay for dissemination (publication, distribution) rather
> than access. If they pay the full cost of dissemination, then the provider
> has completely covered its costs and can offer access free of charge. So
> dissemination fees solve the problem of free online access, if we can find
> the money to pay them.

True, but, in light of the above, this is an over-simplification,
omitting the most important details of the actual quid-pro-quo
involved: What is paid for? why? and by whom? Separate answers are
needed for the the Gutenberg and the PostGutenberg medium. And what
remains essential (peer review) and what (like paper, or the
publisher's PDF) becomes merely optional, PostGutenberg?

> Who will pay dissemination fees? There many potential sources and in the
> best future many different funding models might co-exist. Authors might
> pay dissemination fees,

On the face of it, this sounds like adding insult to injury! Authors
already give away their refereed research (to publishers as well as to
all users) for free: Should they now have to PAY for that privilege?

But of course, once we realize that the PostGutenberg "dissemination"
fees per peer-reviewed online article are almost zero, this question
becomes trivial. The nontrivial question is "Who will pay the
peer-review service fees?" (which are likely to be between $200 and
$500 per article)?

There are several possible candidates, but the poor author certainly
should not be one of them! As we are contemplating here a system in
which there are no longer any S/L/P access tolls, those who are no
longer paying those access tolls are now the beneficiaries of (their
subcomponent) of the collective institutional windfall savings of $2000
- $5000 per article (that is the average total amount that those of the
planet's institutions that can afford it, currently pay for their
access in S/L/P tolls, if you add all their contributions together, per

Who are making those windfall savings once there are no more S/L/P
tolls? The erstwhile payers of those annual S/L/P tolls were of course
the authors' institutions! Hence it would seem to make the most sense
that they should be the ones to redirect a small portion (10-30%) of
their 100% annual windfall savings on their now-obsolete purchases of
the peer-reviewed journal articles from authors at other institutions,
paid for as incoming products, to pay instead only for the peer-review
service on their own authors' outgoing articles.

> and so might their employers (universities and
> labs) or their funders (foundations and governments).

Authors' employers (institutions) are already subsidizing the
dissemination of their research by buying it (and everyone else's
research) back through their annual library serials budget. (This
"buy-back" scenario is often evoked by libraries, tearfully, but
without noticing the critical fact that it is mostly not their OWN
institution's research they are buying "back," but one another's!)
Hence the 100% savings on this annual S/L/P "buy-back" look like the
most natural pot from which to draw the smaller (10-30%) costs of the
up-front "buy-out" of the peer-review costs for each institution's
own outgoing research.

The other funders of the research -- when there is indeed a research
grant, for it is not clear what proportion of the annual [perhaps] 2-4
million articles published in the planet's 20,000 peer-reviewed
journals is actually supported by a research grant -- are indeed the
research foundation and government grants (from which the institution
is often also a receiver of overhead costs).

Yes, those research funders could cover the peer-review costs too, but
research funds are already scarce enough and shrinking, so one can
understand why researchers and their institutions might not wish to
have still more of them redirected elsewhere -- particularly when the
windfall savings for covering the $200-$500 per paper costs in question
are already in the hands of the institution (on the hypothesis of
terminating all S/L/P access expenditures)...

> Journal publishers
> might pay them out of the profits from the sale of add-ons.

Yes, but let us be more specific about this: What are the "add-ons"?
They are the paper version, and any enhanced on-line version
(publisher's PDF version, hypertext-linked version, etc.). Those
add-ons are optional -- and paid for by S/L/P tolls! As long as there
is still a market for them, publishers themselves can indeed
"subsidize" free on-line access for everyone. But will they want to?
Particularly if the competition from the freely accessible version
might erode their optional add-on market?

A much more promising and natural scenario (than that of expecting or
asking or waiting for publishers to provide free online access
themselves) is for the provider-institutions simply to go ahead and
provide it themselves, through author/institution self-archiving! And
then the chips can fall where they may. If the optional add-on S/L/P
market continues to cover all the essential costs, the publishers are
doing the de facto subsidization, and everyone wins (because the
for-free version is accessible to anyone who cannot afford the optional
for-fee version, which nevertheless keeps on paying all the bills).
Whereas if the optional S/L/P market shrinks, the institutional savings
will grow proportionately, ready for a part of them (10-30%) to be
redirected to cover the essential peer review costs -- if/when that
ever becomes necessary.

> For me one the
> most hopeful possibility is journal endowments. If a journal is endowed
> and can pay its dissemination expenses from the interest on its endowment,
> then it needn't hustle funds from any source ever again, creating a very
> stable, long-term solution. Since dissemination costs for online journals
> are low, the needed endowments are correspondingly low.

Yes, but where are those journal endowments (for 20,000 peer reviewed
journals worldwide, mind!) to come from?

> If there are many potential sources of funds to pay dissemination fees,
> then we shouldn't assume that authors must pay. I've argued against author
> fees in the past (FOSN for 9/6/01), but we shouldn't make the mistake of
> thinking that the objections to author fees apply to every kind of
> dissemination fee.

First, the "dissemination fee" is a red herring, as the dissemination
costs per peer-reviewed article are nearly zero, thanks to the Web.

It is the $200-$500 peer-review service costs per article that need to
be covered, and the only natural source for that (the only one that
does not envision redirecting committed funds from elsewhere) is the
prospect of a portion of the windfall savings (if and when they occur!)
-- from the $2000-$5000 per-article currently paid collectively by all
subscribing institutions as S/L/P access-tolls for each incoming
article -- at each individual institution being redirected to cover the
peer review costs for its own outgoing articles.

> BMC does not depend on authors to pay its processing fees, although it once
> called these fees "author charges"
> ( Its latest thinking
> emphasizes universities. By offering to waive processing fees for authors
> employed by universities with institutional BMC memberships, BMC is
> appealing to universities to see the economy of paying small processing
> fees for free online journals rather than paying large subscription
> fees. Despite this, I'm sure BMC would be delighted if foundations made it
> a practice to include funds for article processing fees in every research
> grant.

Alas, charging service fees for outgoing articles now, before
institutions have the S/L/P savings to pay them out of, may be
premature (although perhaps some foundations will be willing to tide
them over for a while, as an investment in the transition toward free
online access).

But to cast these charges as "dissemination" fees rather than the
peer-review fees that they really are would be to perpetuate the
obsolete Gutenberg practise of wrapping the optional add-ons in with
the essentials: Archiving and dissemination are trivial, and can be
done much more cheaply by the author-institutions themselves, on a
distributed basis, without paying journals any more money (no double
payment). It is only the peer review service costs that need to be
covered. (That may require some further downsizing by free-access
journals such as BMC: It is not enough to become online-only. The
archiving/dissemination may be best left to the distributed
provider-institutions, if it entails larger costs when covered instead
by the publisher.)

> But let's focus on the pitch to universities. It's true that universities
> would save significantly if they paid for journals with up-front
> dissemination fees (subsidizing their authors) rather than back-end access
> fees (subsidizing their libraries). However, this savings will only
> materialize in the long run, after priced-access journals have been driven
> out of the field by competition with free-access journals. Until then,
> universities will have to pay twice, first in processing (dissemination)
> fees and then in subscription (access) fees. The problem facing BMC is to
> find a way to help institutions get past the short term loss to the long
> term gain.

Exactly. So tide-over subsidy may be the only hope for new start-up
online-only free-access journals like BMC (while self-archiving
subverts the rest of the system until either the institutional windfall
savings become available or free online access becomes the norm for
all). I am afraid all this means is that the alternative-journals route
to free-access faces certain financial obstacles that the
self-archiving route to the same goal does not.

> Before looking at solutions, let me restate the problem. Funding journals
> through dissemination fees is the only way to make journal access free for
> readers.

Except that it continues to be misleading to call them "dissemination
fees" in the Web-era! Dissemination is a Gutenberg problem. Molecules are
heavy and expensive to disseminate; bits are not!

> So insofar as FOS is embodied in journals, dissemination fees are
> the long-term solution. But since traditional journals charging access
> fees will not disappear immediately, and since no university will want to
> cancel all its subscriptions to them immediately, there will be a
> transition period in which universities face a daunting double payment
> --first, for the desirable new journals representing the future, and
> second, for the important existing journals.

Again, there is the implicit assumption here that the future we desire
-- all 20,000 peer-reviewed full-text journals accessible online for
free for all -- requires a transition to new journals, rather than
freeing access to the existing journals! It is only the new-journals
route that entails the problem of double-payment. The institutional
self-archiving of all existing peer-reviewed research does not. And the
possible expense that might (perhaps) need to be covered eventually by
this second (self-archiving) route is not dissemination but peer
review. And it need not be covered until when (and if) the S/L/P
expenditures no longer cover them. Until then, there is only the (old)
single S/L/P payment from those institutions already paying it now; and
if that time (S/L/P cancellation) ever does come, then, by definition,
the shrinking S/L/P revenues for publishers will have their exact
counterparts in growing S/L/P savings by institutions, which will then
be available for paying the remaining essential cost (peer review).

It is important not to make the "double payment" problem appear to be a
problem for freeing access online: It is only a problem for freeing
access online by the route of creating new alternative free-access
online-only journals (or instantly converting existing journals into
free-access online-only journals). It is not a problem for the route of
institutions self-archiving their own peer-reviewed research, right
now. The latter transition route will continue to be "subsidized" by
ongoing S/L/P expenditures until such a time as competition from the
free versions causes those to begin to shrink. That time may or may not
come, and it may come sooner or later. But, unlike the problem of
covering alternative free-access, online-only journals' expenses now,
it is not a problem institutional self-archiving needs to worry about
now -- and may never be. Hence no "double payment."

(I have repeated myself repeatedly in the foregoing passages! But from
its history of not being understood, the point does seem to bear

> So until the competition with
> free journals drives priced journals from the field, universities will pay
> for both, or want to do so, and their costs will be higher than before or
> after the transition period. The question, then, isn't the economic
> feasibility of the long-term solution, but the economic feasibility of the
> transition. How can we eliminate the double payments?

We can try to find subsidies for the new journals -- but we can also
simply accelerate our institutional self-archiving of our own articles
appearing in the existing journals.

> First note that there is only a double payment here because (during the
> transition period) universities will have reason to support two sets of
> journals when it now supports only one, not because universities would ever
> pay twice for the same journal. When a university pays dissemination fees
> through its authors, it would support journals that provide free online
> access. When it pays access fees through its library, it would support
> traditional journals that limit access to paying subscribers. The new
> dissemination fees must come out of a budget already strained to cover
> access fees.

To put it another way: Their S/L/P budget currently only pays for that
small subset of the annual 20,000 that they can afford to buy in. If
that same money could buy in all 20,000, that would be all the better.
If the 20,000 were accessible without having to be bought in, and could
be had by simply redirecting each institution's total current S/L/P
expenditures toward only its own annual outgoing papers -- and perhaps
even saving 70% or more -- all the better. But paying out MORE money
now, in order to have "free" access to a few more (new) journals just
does not seem like a very inviting proposition...

> If all universities simultaneously agreed to pay processing fees through
> their authors, and stop paying subscription fees through their libraries,
> then the funding paradigm would shift as quickly as institutional inertia
> allows.

Alas it would not. Instead, it would lead to a rather confusing
cost-redirection problem, because paying per-outgoing-article
per-journal is not at all the same as paying S/L/P per incoming
article/journal! The payment is not for the same product, even though
the consumer (the institution) happens to be the same. Earlham might
subscribe to, say, 2000 of the world's 20,000 journals, paying, let's
say, E dollars annually. Now how does it switch from paying for those
2000 annual incoming products to paying for its own outgoing Earlham
articles? First, do those submissions go to the same journals? If the
two are not identical (and not simultaneously duplicated at all the
other institutions that are currently subscribing to and submitting to
those same journals) then we have have a rather complicated
coordination and redirection problem.

And what, exactly, would be paid, for what? Currently, in exchange for
S/L/P tolls, institutions get an on-paper and often also an on-line
product -- the annual full-text articles in each journal subscribed
to. If journals instantly re-did their arithmetic, and redistributed
their annual collective S/L/P charges of $2000-$5000 per article to the
submitting author-institutions instead of to the subscribing
reader-institutions (let's call this OC for outgoing charges), then
imbalances from institution to institution (in terms of which journals
they had subscribed to and which journals they submit to) could produce
huge local upheavals in the size of their OC budgets, sometimes making
them much smaller, but sometimes also much larger than what their S/L/P
budgets had been.

Moreover, this simple redirection would not produce any pressure for
cutting expenses: Every publisher would simply be paid through OC
exactly what he had been paid through S/L/P. So no savings from
downsizing, and separating the essential costs (peer review) from the
inessential ones (paper, PDF, archiving, dissemination), would be
realized. For although downsizing to the essentials ($200-$500 per
article, instead of $2000-$5000) would not only save all the consumers
money, but would buffer against the institutional imbalances
mentioned earlier, an instant redirection would not lead to such
downsizing, saving or buffering (why/how should it?).

Moreover, even the dream of such an instant redirection is itself a
very far-fetched one. And the dream of such an instant redirection
along with an instant downsizing to the essentials is a still more
far-fetched one. We researchers really risk falling into long-term Zeno's
Paralysis, if we continue to contemplate such unrealistic abstract
scenarios, instead of doing what we can already do immediately, namely,
institutional self-archiving of our own outgoing peer-reviewed

> Of course, this won't happen and BMC isn't counting on it to
> happen. But because this won't happen, the BMC strategy faces a classic
> freeloader problem. If affluent and far-seeing universities A, B, and C,
> join the revolution and start paying dissemination fees, then less affluent
> or less far-seeing universities D, E, and F will have reason to benefit
> from the ABC investment without joining it.


> Another way to put this is that the strategy faces a prisoner's
> dilemma. Universities will have a motive to defect against cooperators
> (freeload on early adopters and lengthen the transition period),
> cooperators will be vulnerable to defectors (early adopters will subsidize
> freeloaders and pay double payments for a longer period), and yet mutual
> cooperators will be best off (by shortening the transition to single
> payments).

Correct, so why are we even contemplating it, instead of doing and
promoting something doable that does not face this dilemma?

> Perhaps the simplest way to put the dilemma is that all institutions may
> prefer low dissemination fees to high access fees (partly to save money and
> partly to get free access), but none wants to adopt this model first. All
> want to wait until enough others have adopted it to make double payments
> unnecessary. This pattern looks familiar. There are many other situations
> in which everyone wants to make a certain choice but no one wants to go
> first. For example, all merchants in a town may want a day of rest (say,
> on Sundays), but the first to close on Sundays will lose customers to those
> who do not. Or, all the states in the U.S. may want a relief fund for the
> poor, but the first to raise taxes first in order to provide one will lose
> businesses, hence taxes, to those that do not.

It is more complicated, for even in prisoner's-dilemma form, the best
outcome can only be the redirection of 100% of S/L/P, not downsizing to
the 10-30% essentials. But if we sit and wait for the benign resolution
of a prisoner's dilemma here, we will be waiting a long, long time!
Institutional self-archiving has the advantage of being free of these
restraining trade-offs and leading naturally and directly to the
optimal outcome, now, with no prisoner's dilemma to resolve first!

(Researchers and their institutions are being exceedingly slow about
getting down to self-archiving, but this is not because they are facing
a prisoner's dilemma; they just have not yet seen and understood the
benefits to research and researchers from freeing online access to
their peer-reviewed output. Remedying that requires campaigns of
information for researchers and their institutions, facilitation and
support of institutional self-archiving, and digitometric evidence of
the benefits of free online access. No prisoner's dilemma needs to be

> Finally, if only to show that there are many paths to FOS, remember that we
> can have FOS without journals. Peer review is essential for reliable
> scholarship, and journals traditionally provide peer review, but journals
> are not essential for providing peer review. There are many other
> conceivable ways to do so, especially in an era interactive digital
> networks. If we get peer review without journals, then self-archiving can
> give us free online access to the peer-reviewed literature at essentially
> no cost beyond the network infrastructure already present at every
> university --plus the cost, if any, of the peer review service provider.

Unfortunately, there are some misleading assumptions here too: A
journal, medium-independently (i.e., independently of whether it is
on-paper, on-line, or what have you) IS just a peer-review service
provider and certifier! Speculations about whether we can retain
peer-review but dispense with "journals" become incoherent once
we think of journals as being online-only!

And for the rest, it's just the partitioning problem once again:
What is an "entity X" that takes a text, performs peer review on it,
and then certifies the outcome as "peer-reviewed to the established
standards of entity X" if not a journal? That simply takes us back
where we were before, above: Do these entities have to be new ones,
like BMC, or can they simply be the old 20,000, but with online access
to them freed? And are these entities to perform (and be paid for) only
their essential service (peer review) or are they to continue to wrap
in the costs of options like paper, PDF, links, dissemination, and

> The double payment problem is a real one, but two conclusions give grounds
> for hope. First, it can be solved by any of several strategies, including
> coordinated action, external funding, and journal conversion. Second, it
> can be bypassed to the extent that dissemination fees are paid by anyone
> other than the libraries and universities that now pay access fees.

The double-payment problem is an artifact of the alternative-journals
route to free access. It can be bypassed either by finding tide-over
subsidies for all or some or most of the 20,000 established journals or
their new alternative-journal successors, in exchange for free online
access, or it can be bypassed by institutional self-archiving.

> The economics of FOS are easy compared to the economics of the
> transition. But the transition doesn't have or need to have a single,
> optimal strategy. Let it be a patchwork of makeshifts. If different
> disciplines, nations, or decades have different resources and constraints,
> let local experimentation adapt these general strategies locally.

Let's hope it won't be another decade!

Stevan Harnad

Harnad, S. (2001) The Self-Archiving Initiative. Nature 410: 1024-1025
Nature WebDebatesversion:
Fuller version:

Harnad, S. (2001) For Whom the Gate Tolls? How and Why to Free the
Refereed Research Literature
Online Through Author/Institution Self-Archiving, Now.

NOTE: A complete archive of the ongoing discussion of providing free
access to the refereed journal literature online is available at the
American Scientist September Forum (98 & 99 & 00 & 01):

You may join the list at the amsci site.

Discussion can be posted to:
Received on Thu Jan 03 2002 - 23:30:44 GMT

This archive was generated by hypermail 2.3.0 : Fri Dec 10 2010 - 19:46:22 GMT