Elsevier's new postprint archiving policy, continued
SPARC Open Access Newsletter, issue #75
July 2, 2004
by Peter Suber
Last month I praised and thanked Elsevier for adopting its new policy permitting postprint archiving.  I've been hearing about it ever since.

A quick look at my mail would suggest that OA proponents are dividing on the helpfulness and significance of the new policy.  But a slower read suggests that there are fewer disagreements here than meet the eye, even though when all is said and clarified, there are still some disagreements.  Let me try to untangle some of the controversies.

(1) Is Elsevier providing open access?

No.  It's letting authors provide OA.  (This is assuming that the new policy lets authors provide full OA, not just near-OA.  More on this in section 8 below.)  Elsevier is letting authors have OA if they want it, but authors have to take the next step.  Elsevier is only providing the opportunity. 

The bad news is that authors have been slow to seize existing archiving opportunities.  The good news is that if authors do act, then they can have OA even if no one else lifts a finger.

For recent evidence that it's hard to persuade authors to deposit their work in institutional archives, even when the archives exist and the permissions have been cleared, see Andrea Foster's article in the June 25, 2004, Chronicle of Higher Education.

For more on the primacy of author-actions for achieving OA, see my article from last month's issue, abridged and edited for the Nature OA debate.

Because Elsevier didn't previously provide the same wide-open door for authors, we're a lot further along than we were before.  But because authors need to be stirred to action, we still have a long way to go.

(2) Is it really necessary to get publisher permission for postprint archiving?

Yes and no.  It depends on whether the author has already transferred copyright to the publisher.  If so, then postprint archiving requires the publisher's permission; if not, then the author's own permission suffices. 

The author never needs the publisher's permission if the author retains copyright (or the right of postprint archiving).  The same is true if the author is willing to archive the preprint and the corrigenda (the differences between the preprint and the postprint) rather than the actual postprint.  Thinking of the preprint+corrigenda method, I've often said myself that authors can achieve OA to their own writings without anyone else's permission or cooperation.  So I'm not surprised that some readers were thinking of it when they said that authors don't need Elsevier's permission for archiving the postprints of their Elsevier articles.  The rub is that preprint+corrigenda archiving is not the same as "postprint archiving", even if it's a close second.  It's much less convenient for both authors and readers.  We can disagree about how much true postprint archiving is worth, compared to the next best thing, but we should agree that it requires the consent of the copyright holder.

Should authors of research articles transfer copyright to publishers, making it necessary to get their publishers' permission to archive their own articles?  No, but that's another story for another day.

If you are an Elsevier author, then you will have transferred copyright to Elsevier.  If you want to provide OA to your postprint, and not just OA to the preprint+corrigenda, then you'll need Elsevier's permission.  Starting now, you'll have it, and you'll know that you have it when you submit your work to an Elsevier journal.

Adding Elsevier's 1800 journals to those that already permitted postprint archiving brings the percentage of journals with this policy up to 83%.  This is according to Stevan Harnad's journal-level supplement to Project SHERPA's publisher-level data on copyright and archiving policies.

(3) Why praise Elsevier for acting in self-interest?

I stand by my analysis last month that Elsevier gains from adopting its new policy.  This is not a cynical interpretation; on the contrary, it acknowledges that OA really has benefits to offer even subscription-based publishers.  Elsevier journals are now more attractive to authors than journals that don't preauthorize postprint archiving.  Until all publishers adopt the same policy, every publisher who follows suit will reap the same advantage in attracting submissions.

When I said that Elsevier deserves our thanks, I didn't mean "thanks for making a sacrifice for our sakes".  I meant "thanks for seeing that this was in your interest" or "thanks for finally acting on your interest" or "thanks for seeing that, in the tug of interests pro and con, this policy deserved a try" or simply "thanks for helping OA even if you're also helping yourself".

Let's not hold a double standard.  It's in the interest of BioMed Central and the Public Library of Science to offer OA, yet we praise them for doing so.  Elsevier took a step in the same direction and deserves a measure of praise.  Let's not adopt the theory that OA has to be a sacrifice for the provider before we praise it. 

Some of my correspondents overlooked the fact that my praise was highly qualified.  Preauthorized postprint archiving is good for authors who want OA, and to that extent good for OA.  It doesn't live up to the public definitions of "open access" (more below).  It doesn't help libraries that want OA as relief from exorbitantly priced journals.  It doesn't necessarily trump other reasons authors might have for shunning Elsevier journals. 

One question I received in several forms was this:  What if this admittedly welcome step helps Elsevier's economic health and academic reputation?  Couldn't that be a net loss for the coalition of stakeholders working for OA?  Yes, it could.  But praising one policy is entirely compatible with criticizing other policies.  Let's be vigilant and aggressive about Elsevier's pricing policies, bundling terms, and negotiating tactics.  But at the same time, let's be accurate and honest about what helps OA.

(4) If postprint archiving becomes widespread, what's left to do?

I didn't address this last month, but I've seen disagreements swirling around the question.  Let's suppose that most other publishers adopt a similar policy and that most authors seize the opportunity.  What if 80-90% of authors start providing OA to their postprints?  Will we have reached our goal? 

It depends on your goal.  If you want OA to the literature, then we'll essentially have reached the goal.  We'll only have to tweak 80% into 100%.

If you want the kind of OA that helps libraries, not just the kind that helps authors and readers, then we either have to wait for a compliance rate very close to 100% or we have to shift our attention (which shouldn't have lagged) to OA journals.  Widespread postprint archiving probably isn't enough to justify libraries in cancelling the expensive subscriptions that are breaking their budgets. 

We don't know what widespread postprint archiving will do to subscription-based journals.  It has not killed them in physics, but we don't know yet whether this kind of coexistence will transfer from physics to other fields.  We do know that widespread archiving will not "kill journals" without qualification, because it will not kill OA journals.  Hence we do not have to worry that the success of the OA vehicle that doesn't perform peer review (the OA archive) will kill the OA vehicle that does perform peer review (the OA journal).  However, let's assume that widespread postprint archiving will harm some subscription-based journals and not others.  In that case, we'll have a reason to continue our campaign for OA journals.

In short, we'll have at least two reasons to continue to work for OA journals even in a world of widespread OA archiving:  first, to help libraries solve the pricing crisis, and second, to ensure the survival of peer-review providers.  (BTW, this is compatible with the view that in the continuing evolution of scholarly communication peer review needn't be provided by "journals".)

(5) Is the new Elsevier policy a big step or a small one?

Compared to what?  If we compare the new Elsevier policy to what authors need to provide OA to their own work, then it's a big step.  It's just about everything authors need.  (If there's an exception, see section 8 below.)  If we compare it to what libraries need to cancel expensive journals, or what the entire range of stakeholders need to reform scholarly communication, then it's small. 

If we compare it to Elsevier's previous policy, then it's somewhere in the middle.  Previously Elsevier required case-by-case requests for postprint archiving, but it routinely granted them.  If we put the accent on the elimination of the permission barrier, then the step is large.  If we put the accent on the "yea rate", then it's smaller.

We may agree on all three of these judgments --that from these three standpoints the policy-change is large, small, and middling.  But we might still disagree on the preferred standard of comparison because we disagree in our goals.  Any step that is large for authors but small for libraries will bring out these differences.

(6) Is the glass half-empty or half-full?

Some OA proponents acknowledge that Elsevier's policy is helpful as far as it goes, but they want to focus on what this policy doesn't do, what remains to be done, or other Elsevier policies that are clearly harmful.  OK.  This is a matter of accent again.  We can agree on all the facts.

I have no objection if someone wants to acknowledge the benefits of this step for authors and then focus attention and energy elsewhere, wants to question or dispute these benefits, or wants to disregard the whole set of issues.  But I do object if someone acknowledges the benefits but wants us to deny them in the name of strategy.  I have a very different picture of strategy.  I think it's good strategy to show that we are just as willing to acknowledge positive steps as negative ones.  It's good strategy to show that we welcome positive steps, and do not merely demand them as our right.  It's good strategy to mean what we say and to say what we mean.

(7) Is the apparent helpfulness of the policy a deceptive appearance?

Some say "yes".  Some who say "yes" seem to assert this line of reasoning:  "This policy seems good but must really be bad, since it comes from a bad company."  This an odd line to take in the face of a change of policy that seems to be for the good.  Do you judge the policy by the company or the company by the policy?

Some assert a slightly different line of reasoning:  "This company has been bad in the past and continues to be bad on other fronts.  Therefore what looks like a good step might really be bad.  We have to be careful here and look very closely."  If the conclusion is that we should be suspicious, I can accept that.  The question is whether we use suspicion as shorthand for dismissal or as a goad to look more closely.  If we look more closely, then the question is what we find when we look.

I've corresponded with several readers who called for suspicion.  But I haven't yet found anyone who is ready to say that the new policy is really harmful to OA and only appears to be helpful.  If you think it is really harmful, I'd like to hear your thoughts, either privately or through our forum. 

SPARC Open Access Forum

(8) Is this full OA or just an approximation?

Elsevier allows deposit in some OA archives but not in others, for example, in your institutional repository but not in a disciplinary archive.  Yet one part of every public definition of OA is the freedom to make and distribute copies.  Can readers of a properly archived copy of an Elsevier postprint make a copy and deposit it where Elsevier doesn't want it to go?  Can readers make copies at all? 

I pointed out these problems in my article last month and don't disagree with anyone who wants to withhold the label of "open access" from this policy.  At least the policy permits postprint archiving.  While I want to save the term "open access" from dilution, I also want to acknowledge forward steps when they occur, even if they stop short of full OA (which for me is the BOAI definition). 

In any case, any final decision about whether Elsevier's policy permits full OA depends on how Elsevier clarifies the policy.  For this, see the next section.

(9) How should we interpret the new policy?  What does it actually permit?

Elsevier still hasn't posted an official version of the new policy.  When it does, I hope it will clarify the following points. 

* Karen Hunter's May 27 email on the new policy said that authors could post the final version of the text in a "Word or Tex file, not a PDF or HTML downloaded from ScienceDirect".  Everyone understands that authors won't be allowed to post the Elsevier-published PDF or HTML.  But I've seen some listserv postings and journalists say that authors will be restricted to Word or Tex files.  My reading is that authors could use any format, even PDF or HTML, provided they don't use copies of Elsevier's own files.  Which reading is right?

* Elsevier's June 3 press release said that the author's archived postprint may not be "used for commercial purposes -- such as systematic distribution...."  What does this mean?  It's clear enough that authors cannot sell access to their postprints (not that they would want to) and that other publishers may not sell access to copies of the postprints.  But depositing the postprint in an institutional repository, which the policy specifically allows, is already a kind of "systematic distribution".  What kind of systematic distribution is the policy trying to disallow? 

Note that if the criterion here is so vague that authors can't apply it themselves, then they'll have to check with Elsevier case by case, negating the benefit of permitting postprint archiving in advance.

* The same press release said that the archived postprints could not be "used for commercial purposes -- such as...creating links for commercial customers to articles."  What does this mean?  One journalist has said that it bars all links to the archived postprint.  Another says it only bars links from "centralized databases".  A third says it only bars links from commercial sources.  If the policy trying to bar some kinds of links, then what kinds exactly?  Does Elsevier want to bar authors from linking to their own postprints (say) from their home pages or listserv messages?  Does it want to bar people beyond the author's control from linking to the author's copy of the postprint?  Does it want to bar search engines from crawling archives and offering links on their hit pages?

* Elsevier's "Practical Guide" to how authors may use their Elsevier articles (undated except as 2004) says that authors may put their postprints "on pre-print servers and the authors' personal or institutional Web sites."  By including pre-print servers, this version of the policy embraces archives like arXiv and CogPrints, not just institutional repositories.  This would be a welcome addition, but will it make the next cut?  It seems inconsistent with the Karen Hunter email, which limited the permission to institutional repositories and personal web sites.

* Some have read the Elsevier language about the "final version of the text" as excluding charts, images, and other non-textual elements.  I read it to include everything in the final edition of the article, and merely to exclude the Elsevier-produced file with its special look and feel.  Which reading is correct?

My praise for the new policy rests on some guesswork about how these questions will be resolved.  If I misunderstood the policy, or if it's clarified in the wrong direction, I promise to curb my enthusiasm.

Elsevier permits postprint archiving, my article from SOAN for 6/2/04

Elsevier's page on rights retained by authors
Not yet updated to reflect the new policy.  Checked July 2, 2004.

Elsevier's page on electronic preprints
Not yet updated to reflect the new policy.  Checked July 2, 2004.

Karen Hunter's May 27 email first disclosing the new policy

Elsevier's June 3 press release on the new policy

Elsevier's document, Ways to Use Journal Articles Published by Elsevier:  A Practical Guide (2004)
See esp. Section 2 on "how authors can reuse their own articles"

* Postscript.  One point I made last month was that Elsevier doesn't have, and needn't have, the same position on OA archiving that it has on OA journals.  While Elsevier was receiving wide press coverage for its new support for OA archiving, CEO Crispin Davis reiterated the company's position on OA journals for the company's in-house newsletter.  His critique was summarized by Richard Wray in the June 30 issue of The Guardian.

Richard Wray, Open access jeopardises academic publishers, Reed chief warns, The Guardian, June 30, 2004.
("The rise of open access publishing of scientific research could jeopardise the entire academic publishing industry, according to the chief executive of Reed Elsevier....Sir Crispin Davis warned that [OA journals] 'could jeopardise the stable, scalable and affordable system of publishing that currently exists.' ")

It's no contradiction for Elsevier to permit OA archiving and oppose OA journals, just as OA archiving might take off first in physics and OA journals take off first in biomedicine.  While archives and journals can both provide genuine OA, they offer different sets of benefits that appeal to different constituencies in different ways.  Sometimes this is a disadvantage:  the success of OA archiving in physics is slow to translate to other disciplines.  But sometimes it's an advantage:  Elsevier's opposition to OA journals doesn't stop it from supporting OA archiving. 
The problem is not inconsistency, but the deplorably weak arguments Elsevier uses to attack OA journals.  The best compendium of these arguments, and the best answers to them, is by BMC.

However, the Davis argument summarized by Wray on June 30 contains a point of interest that may be new in Elsevier's rhetoric.  It is starting to acknowledge that one of Elsevier's objections to OA journals is based on the company's economic interest and survival.  This is credible.  It's only when Elsevier praises its own service as "stable, scalable, and affordable", or pretends to represent the interests of science, that its arguments conspicuously miss the target.  Could it be that Elsevier is starting to make the argument that its own survival as a company is in the interest of science?  If so, then it would be easy to answer.  Certain services, like peer review, are indispensable, but no particular publisher is indispensable.  We don't need Elsevier for science any more than we need Disney for entertainment or Microsoft for software. 

The differences between OA archiving and OA journals mean that Elsevier's helpful new policy on the former needn't moderate any of its arguments against the latter.  But it would be a shame if the good will that Elsevier created or could create with its archiving policy were squandered through overblown denunciations of OA journals that are daily proving their viability, quality, and impact.


Read this issue online

SOAN is published and sponsored by the Scholarly Publishing and Academic Resources Coalition (SPARC).

Additional support is provided by Data Conversion Laboratory (DCL), experts in converting research documents to XML.


This is the SPARC Open Access Newsletter (ISSN 1546-7821), written by Peter Suber and published by SPARC.  The views I express in this newsletter are my own and do not necessarily reflect those of SPARC or other sponsors.

To unsubscribe, send any message (from the subscribed address) to <SPARC-OANews-off@arl.org>.

Please feel free to forward any issue of the newsletter to interested colleagues.  If you are reading a forwarded copy, see the instructions for subscribing at either of the next two sites below.

SPARC home page for the Open Access Newsletter and Open Access Forum

Peter Suber's page of related information, including the newsletter editorial position

Newsletter, archived back issues

Forum, archived postings

Conferences Related to the Open Access Movement

Timeline of the Open Access Movement

Open Access Overview

Open Access News blog

Peter Suber

SOAN is licensed under a Creative Commons Attribution 3.0 United States License.

Return to the Newsletter archive