Re: 2.0K vs. 0.2K

From: Stevan Harnad <harnad_at_COGLIT.ECS.SOTON.AC.UK>
Date: Fri, 7 May 1999 15:53:43 +0100

> Date: Thu, 6 May 1999 13:52:06 -0400 (EDT)
> From: "Arthur P. Smith" <>
> sh> (1) Is the true cost closer to $2000 per article or $200?
> Well, the true cost (for this specific question, for us) is
> a matter of calculation, not speculation.

Fair enough. But I of course did not mean the true cost of CURRENT paper
publication, minus printing/distribution, divided by the number of

What is being discussed is the true cost of producing ON-LINE-ONLY
pages, already distributed by a free public archive. So the only
remaining ESSENTIAL journal cost per page left to reckon is that of
quality control (peer review). Do you have a calculation for that?

> Taking our total
> costs for editorial and production and electronic services work
> (ie. not including printing and distribution) and dividing by
> the number of papers we publish in a year yields very close to
> $1500 per article.

As I have said in reply to this sort of (mis)calculation many times
before: this is like calculating the potential saving from switching
from horses to cars only in terms of how much less one will have to pay
for hay!

You are reckoning in a whole infrastructure and way of doing things
that evolved specifically for producing print on paper, supported by
S/L/P. A system dedicated only to quality control would look a lot

> As I've said elsewhere, this should drop as we implement new technologies
> - our editorial office {American Physical Society] is currently
> undergoing a complete workflow analysis to move to a more electronic
> environment and to determine where inefficiencies remain and
> can be improved.

Among the prospective improvements, are the following being
contemplated? (1) scaling down to online-only, (2) eliminating all
aspects of archiving/distribution, and (3) delivering only what is
essential in order to turn submitted drafts into refereed, accepted
drafts? Does that still come out closer to 2K than .2K per draft?

> And one of Mark Doyle's projects here (that I'm
> a bit involved with) is to help authors send us material that
> should cut our copyediting costs significantly too.

Offloading onto the author as much as possible of draft preparation,
revision, and mark-up is certainly the natural and sensible way to

> But those new things aren't there yet, and if we were to
> cover all our first-copy costs through page charges, that's what
> it would have to be for us right now.

That calculation is uncontested, but not the one in question!

> It will certainly be
> different for other publishers and from journal to journal - however
> I doubt there are many publishers with real first-copy
> costs right now as low as $200/published article.

Correct (with the possible exception of some of the new online-only
journals -- although their small scale is not yet comparable with that
of any of the established journals, and they are still involved in
archiving and distribution, whereas with universal
author-self-archiving they will no longer need to be) but irrelevant.

We are talking about what costs would be in a system scaled down to the
online essentials only. No needless "add-ons."

(Just as this anomalous literature [donated for free by its authors,
unlike any other literature] must not be held hostage to S/L/P, it must
also not be held hostage to inessential "add-ons" with which it is
tightly but needlessly wrapped! Once the entire corpus is seamlessly
online and free, the user community will spawn "add-ons" of undreamt-of
scope and power in the form of links, search-engines, classification
and alerting. Let primary publishers not cast their fate with the
prospect of upgrading their traditional products with secondary
and tertiary services, for this strategy is bound to lose in the
innovative shareware arena of the Web!)

> To take another example, the New Journal of Physics (
> is a totally new journal, funded by author charges. Their introductory
> price is $500/article. Having talked with them, I know that's not near
> covering their current costs (especially with only 7 published articles
> so far).

We all wish this new IOP journal well, but, as explained earlier in
this threaded discussion, page-charge journals are almost certainly
premature at this time. A large-scale transition to author
self-archiving will have to precede them, addicting readers and authors
to the new world of free access, thereby putting increased cancellation
pressure on S/L/P budgets, in turn creating an incentive for everyone
to find a new way of covering the only remaining essential cost:
quality control.

Not only is the NJP burdened with the current IOP paper-based
infrastructure and its overheads, but the archival IOP draft still has
many expensive but needless add-ons that the scaled-down online
journals of the future will be able to offload onto authors and

And NJP is hurting for papers for the usual reason new journals hurt for
papers, regardless of medium. In short, the fate of NJP, although all
our fingers and toes are crossed that it will make it, is in no way
representative of the fate of the page-charge model further downstream.

 sh> (2) Is the distinction between allowing free self-archiving of the final
 sh> draft on the "home" server and on the "global" server coherent and
 sh> enforceable?
> I just sent off a note requesting our official policy on this.
> To some extent it's probably a matter for copyright lawyers to decide.

It's up to the lawyers only if journals take a confrontational position
in this whopping conflict of interest between themselves and their
authors. If the interaction is collaborative, then what the lawyers say
is irrelevant: the publishers will let authors do what is in the best
interests of research and researchers. I think it is quite obvious what
that is, without the help of a copyright lawyer.

But the enforecability issue is not just a legal one in any case: it is
also a technological one. There is a slippery slope between preprints
and final drafts: Can the law specify a cut-off point, and is there
any way to detect it in each specific case, or even in most cases? Will
publishers have to become net-sleuths, trawling day and night for
suspicious look-alikes, and then examining each individually to decide
which side of the slippery slope they are on? Is there any way prevent
institutional archives from being "cached" and mirrored in more global
sites? Would we want to? Can a lawyer answer any of this?

Is there not something very revealing and natural in the fact that in
no other area of publishing does this potential conflict exist, where
the publisher must try to force the author not to give away his texts
for free, and must eternally spy on him to make sure he does not do it?
Is there not a more sensible and amicable way to resolve this conflict
of interest (that I have dubbed the "Faustian Bargain") between
research/researchers and those who are meant to help make their
findings PUBLIC?


> But please note that our copyright license (which allows authors to
> at least self-archive the final APS production version of their paper
> on their home server, and also allows them to at least leave earlier
> versions up on preprint servers) is viewed as flamingly liberal by
> most of our scientific publishing colleagues :-)

I know and profoundly appreciate (and unfailingly acknowledge) that!
APS sits on the side of the angels in this (and in many other things)!

> In fact, Tom Walker's
> journal does NOT allow this unless authors pay their electronic
> reprint fee.

And it is for this reason that I have suggested since the
beginning of the American Scientist Forum that although Tom's heart is
undeniably in the right place, his proposal is unrealistic. Not only
does it assume that S/L/P can persist, but it imagines that authors
will pay page charges on TOP of S/L/P even though they could achieve
the very same result for free, by self-archiving!

 sh> (3) Is the cost to be thought of as coming from authors' pockets or
 sh> from institutional S/L/P savings?
> What savings? Tom Walker's model (which was what the discussion quoted
> was about) still requires institutions to purchase the journal in order
> to get access to papers by authors who haven't paid the reprint charge.

Indeed, and this is part of why I believe Tom's proposal is
unfortunately a non-starter.

> Even if they could drop their subscriptions this doesn't make sense though.
> Aside from the authors we have at institutions too small to subscribe to
> our journals (which for us is I think really a very small fraction),
> a large number of our authors come from institutions that have
> many authors, but only one subscription: ie. if those institutions
> went to an "author-pays" system rather than a "reader-pays"
> system, they would be paying more, not less.

Do you think this would still be true if the page charges were only for
the true costs of quality control, with no "add-ons"?

> For example, a national
> lab (let's say Los Alamos) which pays $12,000 for a subscription
> to everything we publish would need to publish at most 8 papers
> per year with us to save money using our costs (or at most 60
> papers/year with Stevan's number) when in fact they send us something
> like 200 papers/year right now. Does Los Alamos's library have
> an extra $300,000 (or even $40,000) to throw at us?

Although I believe your figures are wrong (even for "Stevan's number"),
you are nevertheless putting your finger on one potential problem for
the specific model in which page charges are paid entirely from
institutional S/L/P savings:

I had pointed this out already in the AmSci thread on "The Urgent Need
to Plan a Stable Transition": Although, most institutions are net
CONSUMERS of journal papers, and would profit hugely from S/L/P
terminations even after paying for all their institutional author
page-charges, some institutions (a minority) may be net PROVIDERS of
journal papers. It seems to me that HERE is where a library consortial
pooling of resources WOULD make sense (i.e., in fairly sharing
author-end up-front costs, NOT reader-end global site licenses).

Any way you do the overall arithmetic, if the annual total global S/L/P
expenditure is G today, and scaling down to online-only quality control
costs would reduce it to g, then G-g represents a net saving. A
consortium could see to it that this annual windfall saving was fairly
distributed among net-consumer and net-producer institutions.

Here's another way to think of it: Currently, the net-provider
institutions are subsidizing the net-consumer institutions. With up-front
page charges, the excess savings of the net-consumers will be
redistributed to balance the excess expenses of the net-providers, but
all institutions will save something like (G-g)/G times their own
current serial budgets annually.

> Now maybe an author-pays system would be fairer on the smaller
> institutions - but let's not talk about saving institutions money
> when all we're doing is redistributing the same costs in different ways.

Not quite. I am betting (on, paradoxically, market-economic grounds)
that continuing to recover costs on the reader-end, access-blocking
S/L/P model will perpetuate an inelastic (if not monopolistic) inflation
of both prices and (inessential add-on wrap-in) services. The potential
savings will never be realized or passed on to institutions as long as
the cost recovery model is S/L/P. Only reader-end page charges --
forced by author-end subversive self-archiving, which makes the
literature free by bypassing S/L/P -- will give journals the incentive
to scale down their services and expenses to the lean essentials,
thereby maximizing G-g and ensuring that it is all passed on to

Stevan Harnad
Professor of Cognitive Science
Department of Electronics and phone: +44 1703 592-582
Computer Science fax: +44 1703 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 - 19:45:30 GMT