Open Access Archivangelism

---

SUMMARY: Which Green OA Mandate should an institution adopt?
    ID/OA: The Immediate Deposit, Optional Access-setting (ID/OA) mandate immediately guarantees at least 63% OA plus 37% Almost-OA, moots all objections on copyright grounds, and does not put the author's choice of journal at risk by requiring individual licensing negotiations by the would-be author with the publisher (with no guarantee of a successful outcome). The other alternative candidate mandates are:
    ID/IA: The Immediate Deposit/Immediate Access (ID/IA) mandate is stronger than ID/OA. But how can such a mandate manage to reach consensus on adoption as long as 37% of journals don't endorse immediate OA self-archiving? (Invariably this has meant having to allow an author opt-out or waiver for such cases, in which case the policy is no longer a mandate at all -- i.e., it is weaker than ID/OA. Hence not one of the existing mandates to date is ID/IA.)
    ID/DA: The usual compromise, therefore, is to allow access embargoes, with or without a cap on the maximal allowable length. But an Immediate Deposit/Delayed Access (ID/DA) mandate, with no cap on the allowable delay (embargo) is simply identical to ID/OA! Adding a cap on the maximal allowable embargo delay is splendid, but that's just ID/OA with an embargo cap. (So if an institution can reach successful consensus on this stronger mandate (capped ID/DA), they should by all means adopt it; but if not, they should just go ahead and adopt ID/OA.)
    DD/DA: Next there is Delayed Deposit/Delayed Access (DD/DA), in which the deposit itself may be delayed until the embargo elapses, instead of being done immediately upon acceptance for publication, as in ID/OA. But with or without an embargo cap, DD/DA is in fact needlessly weaker than ID/OA, because it arbitrarily loses the 37% Almost-OA that authors can provide semi-automatically via the button, until the date at which each embargo elapses. (DD/DA further risks needlessly losing a lot of the 63% OA as well, by not requiring immediate deposit in any case.)
    Author Licensing Mandate ("author addendum", "rights reservation"): Last, and almost as strong as the (nonexistent) ID/IA mandate is a negotiated author-licensing mandate ("author addendum"). But how can such a mandate reach consensus on adoption and compliance with authors who are concerned that it would put their papers at risk of not being accepted by their journal of choice whenever the licensing negotiation fails? As a consequence, there exist no ID/IA mandates either, only ID/IA with an optional author-opt-out (waiver) (as in Princeton's mandate) -- which again loses the 37% Almost-OA during any embargo or opt-out, and is hence needlessly weaker than ID/OA (if a mandate with waiver/opt-out is indeed a mandate at all! This has been remedied in Harvard's mandate, which adds copyright reservation [with a waiver option] with ID/OA [with no waiver option]).
    It is because of this logic and these pragmatics that ID/OA is the default baseline mandate: Anything weaker than ID/OA is gratuitously weaker than necessary (and generates less OA than ID/OA). Anything stronger (such as ID/IA without opt-out, or mandatory licensing without opt-out) is more than welcome, if an institution can successfully reach consensus on adoption and compliance. But no institution (or funder) has yet managed that, hence holding out for an over-strong mandate leads to the adoption of no mandate at all. (And designate deposit as the sole means of submitting refereed publications for performance evaluation.)

---

The following are replies to some queries I received about Elsevier's Authors' Rights Policy on Green OA Self-Archiving:

Q: "The Elsevier Authors' Rights policy states that authors may...

'Post a personal manuscript version of the article on the author's personal or institutional website or server... This means an author can update a personal manuscript version (e.g., in Word or TeX format) of the article to reflect changes made during the peer-review process. Note [that] such posting may not be for commercial purposes... and may not be to any external, third-party website.'

"Does this not exclude posting to a repository? And if so, can Elsevier still qualify for Green status given such a restriction?"

The Green OA author-rights policy of Elsevier (and of many other Green publishers) explicitly excludes posting to a 3rd-party repository (in order to avoid sanctioning free-riding from rival publishers). The author's own institutional repository, in contrast, is of course simply an institutional website/server.

And that is just fine for OA -- in fact, better than fine, for the following reasons:

All Open Access repositories -- whether institutional or institution-external -- are interoperable, because they comply with the OAI metadata harvesting protocol. So if you deposit your full-text in any one of them, it is functionally as if you had deposited it in any or all of them, because central harvesters (such as OAIster and Elsevier's own Scirus) can and do harvest the metadata from all those repositories, and then provide a search interface that collects the links to all the distributed repository contents into one searchable global OA repository (or several, by subject, or nation, or funder, or what-have-you) -- just as Google and Google Scholar likewise do.

So not only does Elsevier's Green author-self-archiving policy acknowledge the right of every single Elsevier author to make (the author's refereed final draft of) every single one of his Elsevier articles OA, but it helps correct an extremely counterproductive and ill-thought-out -- though well-intentioned -- bug in many of the current OA self-archiving funder-mandates, in which direct central deposit (in PubMed Central) is mandated, needlesly, instead of direct institutional deposit (which can be followed by central metadata-harvesting or automatic SWORD-based export to PubMed Central).

It is quite natural to ask: "If all repositories are interoperable, what difference does it make whether mandates specify central self-archiving or institutional self-archiving?"

To understand the reply, one must first clearly apprehend that:

-- not all research is in one field
-- not all research is funded by any given funder
-- not all research is funded at all
-- all research originates from institutions
-- yet most institutions have not yet mandated self-archiving

As all research originates from a university or other research institution, institutions are the universal providers of all refereed research output. Hence, to systematically ensure that all refereed research -- from all institutions, in all fields, whether funded or unfunded -- is self-archived, it is essential that all self-archiving mandates -- whether from institutions of funders -- be convergent and mutually reinforcing, rather than divergent and competitive. 

In other words, both funders and institutions need to mandate self-archiving in the author's own institutional repository. That way funder and institutional mandates reinforce one another: For example, NIH's mandate, which technically applies only to NIH-funded biomedical research, would vastly increase its scope if it stipulated institutional deposit; that would immediately extend the reach of the NIH mandate to all institutions receiving NIH funding, thereby encouraging each of them to adopt institutional mandates of their own to cover all their research output, and not just the NIH-funded subset. And it would of course also encourage universities to create institutional repositories, if they have not done so already. (Institutions -- already quite obsessively involved in preparing and vetting their research grant proposals, are also the natural ones to monitor and ensure compliance with the funder's deposit mandate as part of the grant's fulfillment conditions, and especially if they have a institutional deposit mandate of their own. And again, one convergent deposit and locus -- institutional -- is enough for an author; the rest can be handled by automatic harvesting or exportation.)

Nothing whatsoever is lost with convergent mandates: NIH can still harvest all of its funded research into PubMed Central; so can any other harvesters, into their own collections. (And import/export can also be done automatically, via the SWORD protocol.) There is absolutely no need for -- and many reasons against -- authors depositing directly in PubMed Central or any other central collection: Not only does that fail to encourage and reinforce the adoption of institutional mandates for all the rest of each institution's (non-NIH) research output, but, apart from creating needless confusion and ambiguity about locus of deposit, it even puts authors at (perceived) risk of having to do multiple deposits in order to fulfill both funder and institutional mandates. (Here too, the risk is perceived rather than real, because the SWORD protocol can of course transfer contents in both directions, whereas the real problem is inducing the adoption of mandates from both funders and institutions, and particularly from the universal providers, the institutions: It is precisely there that collaborative, convergent mandates from funders are a great help whereas divergent, competitive mandates are a gratuitous hindrance.)

So some of the funder mandates to date, welcome though they were and are, have failed to think things through (and failed to fully understand IRs, and OAI-interoperability and OA itself) in needlessly mandating direct central deposit. (That's rather like requiring providers to deposit directly in google, instead of depositing on their own distributed websites, from which google can then harvest.)

This inadvertent error can and will still be corrected, of course, but the correction needs to happen sooner rather than later, because meanwhile divergent funder mandates are slowing the growth of the potential institutional mandates that represent the real lion's share of global research output across all disciplines (and research access and impact continues to be needlessly lost).

NIH has so far been unresponsive about reconsidering its unnecessary and counterproductive requirement to deposit directly in PubMed Central (and a number of CopyCat funders elsewhere have been simply aping NIH, equally unreflectively). The work of those who are trying to create a systematic synergy that will result in the rapid global growth of OA mandates by the sleeping giant (the world's universities and research institutions), is thereby made more difficult than necessary -- but here, Green publishers' restrictions on depositing in 3rd-party central repositories may prove to be a serendipitous aid in converging on convergent deposit mandates and practices.

Q: "Having made the case for Immediate Deposit/Optional Access-setting (ID/OA), it's not clear how you can be so complacent about the Optional Access-setting part. Surely in order to achieve the goal of genuine open access, articles must not only be deposited immediately, but also made freely available. Now clearly, the weaker the mandate, the more likely is its adoption; this is a matter of tactics. But what if all institutions adopt ID/OA, but then do not Open Access on the deposits? Has anything been gained? (Perhaps you’d reply yes, because one has to win the first battle before being free to fight the second?)"

    (1) I am not at all worried about the (37% of) articles that might, under the ID/OA mandate, be deposited as Closed Access (hence not OA but only "Almost-OA," thanks to the repositories' semi-automatic, almost-instantaneous, "email eprint request" button). I have been at this for over a decade and a half now, and for that decade and a half, we have not had the at-least 63% OA plus 37% Almost-OA that we could have had, if we had adopted ID/OA mandates. Instead, all we had is what we have now: the <15% of all articles that are being self-archived spontaneously (unmandated), plus the even smaller percentage of all articles that are being published in Gold OA journals.

    (2) So I am unworried about universal ID/OA mandates, but definitely not unworried about continuing to do without them, just waiting instead in the hope that (a) either spontaneous self-archiving will somehow increase to 100% on its own, without the need of any mandate at all, (b) or that agreement on even stronger OA mandates will somehow be reached, or (c) that all publishers will somehow agree to convert to Gold OA publishing!

    (3) Just continuing to advocate and wait for (a) or (c) is, I hope all will agree, not a sensible strategy after all these years.

    (4) As for stronger mandates, consider the options: In place of the ID/OA (Immediate Deposit, Optional Access-setting) mandate -- which immediately guarantees at least 63% OA plus 37% Almost-OA, moots all objections on copyright grounds, and does not put the author's choice of journal at risk by requiring individual licensing negotiations by the would-be author with the publisher -- we have the following alternative candidates:

(4a) There is the stronger Immediate Deposit/Immediate Access (ID/IA) mandate. But how can such a mandate manage to reach consensus on adoption as long as 37% of journals don't endorse immediate OA self-archiving? (Invariably this has meant having to allow an author opt-out for such cases, in which case the policy is no longer a mandate at all -- hence weaker than ID/OA. Not one of the existing 58 mandates is ID/IA.)

(4b) The usual compromise, therefore, is to allow embargoes, with or without a cap on the maximal allowable length. But such an ID/DA (Immediate Deposit/Delayed Access) mandate, with no cap on the allowable delay (embargo) is simply identical to ID/OA! Adding a cap on the maximal allowable embargo delay is splendid, but that's just ID/OA with an embargo cap. (So if an institution can reach successful consenus on this stronger mandate (capped ID/DA), they should by all means adopt it; but if not, they should just go ahead and adopt ID/OA.)

(4c) Then there is Delayed Deposit/Delayed Access (DD/DA), in which the deposit itself is delayed until the embargo elapses, instead of being done immediately upon acceptance for publication, as in ID/OA. But with or without an embargo cap, that is in fact weaker than ID/OA, because it loses the 37% Almost-OA accessible via the button, until the date at which each embargo elapses.

(4d) Last, and almost as strong as the (nonexistent) ID/IA mandate (4a) is a negotiated author licensing mandate ("author addendum"). But how can such a mandate (for authors to use the author addendum) reach consensus on adoption with authors who are concerned that it would put their papers at risk of not being accepted by their journal of choice whenever the licensing negotiation fails (because the publisher declines)? As a consequence, there exist no ID/IA mandates either, only ID/IA with an optional author-opt-out (as in Princeton's mandate) -- which again loses the 37% Almost-OA during any embargo or opt-out, and is hence weaker than ID/OA (if a mandate with an opt-out is indeed a mandate at all! This has been remedied in Harvard's mandate, which adds copyright reservation (with a waiver option) with ID/OA (with no waiver option)).

It is because of this logic and these pragmatics that ID/OA is the default baseline mandate: Anything weaker than ID/OA is gratuitously weaker than necessary (and generates less OA than ID/OA). Anything stronger (such as ID/IA without opt-out, or mandatory licensing without opt-out) is more than welcome, if an institution can successfully reach consensus and compliance -- but no institution (or funder) has yet managed that, hence holding out for an over-strong mandate leads to the adoption of no mandate at all.

The most common mandates are DD/DA, with and without embargo caps, but those are all weaker than ID/OA, and hence should be upgraded to at least ID/DA, whether with or without an embargo cap.

Licensing mandates are rarer, but as they have opt-outs, they too are weaker than ID/OA. Again, however, they can easily be upgraded to add ID/OA (without opt-out) alongside the "mandatory" licensing (with opt-out), and then they are fine.

So the reason we can all be relaxed and satisfied with ID/OA is that it is the default baseline for all OA mandates. Where a stronger one can be agreed, it should of course be adopted. And it makes no sense to arbitrarily adopt a still weaker mandate -- DD/DA or author-licensing with opt-out -- since upgrading either of those to ID/OA is free of objections, either in institutional copyright worries or in author worries about journal choice. But on no account should no mandate at all be adopted, in favor of an unending quest for a mandate stronger than ID/OA: Adopt ID/OA now, and then try to agree on a stronger upgrade thereafter (and meanwhile, designate deposit as the sole means of submitting refereed publications for performance evaluation).

(Note that I did not list a universal conversion to Gold OA publishing on the part of publishers as being among the mandate options, because, on the one hand, institutions and funders can only impose mandates on their own employees and fundees; not on publishers. And, on the other hand, continuing to sit around trying to persuade publishers to convert to Gold OA of their own accord is likely to extend our waiting time even closer to the time of the heat death of the universe than continuing to sit around trying to reach agreement on stronger mandates, instead of just going ahead and adopting ID/OA right now.)

Q: "How do you know that the university servers/websites in which Elsevier allow authors will be OAI compliant?"

Because just about all Institutional Repositories today are OAI-compliant! OAI-interoperability was part of the rationale for the institutional repository movement -- a movement that was itself set in motion by the OA movement (in 2000) but has alas since moved off in other directions of its own (often vague and aimless ones, not carefully thought through in advance).

Q: "Why do you think Elsevier has adopted a Green policy on self-archiving?"

I cannot second-guess the motivation. Elsevier is a commercial company. Unlike (some) Learned Society and University publishers, its primary allegiance is not to science but to the bottom line.

My guess is that Elsevier realized that OA is inevitable, and unstoppable. So they have settled on just trying to slow it down, if possible. At first, I think Elsevier's Green policy on author self-archiving was based on three calculations: (1) that a Green policy on author self-archiving might quiet the growing clamour for OA, perhaps even (2) reducing the widespread hostility against Elsevier for their aggressive pricing policies, while (3) not many individual authors would be likely to actually go ahead and self-archive. If this was indeed Elsevier's reasoning, then they were right on all three counts.

But what Elsevier did not reckon on was the emergence and growth of the movement for institutional and funder self-archiving mandates. (While it was adopting a Green Policy on author self-archiving, in order to satisfy the author demand for OA, Elsevier was also lobbying -- vigorously, but in the end unsuccessfully -- against Green OA self-archiving mandates. (Elsevier, along with the American Chemical Society [the non-Green publisher giving Learned Societies such a bad name] had been among the prime supporters of the notorious "prism"/pitbull anti-OA coalition that backfired so badly when exposed in Nature.] So I am quite ready to forgive and forget Elsevier's failed attempt to lobby against Green OA self-archiving mandates; their Green policy on author self-archiving (though it too was not necessary for Green OA to prevail in the end) has nevertheless been more helpful than the lobbying has been harmful, and historians will make due note of that.

Finally, before anyone asks (again), my own view of the way things will go in the future for publishing (about which I no longer speculate publicly, because speculation and counterspeculation on this question has been for far too long yet another distraction from practical action toward providing immediate Green OA) is described here, here, and here. Speculations aside, all empirical evidence to date has been that conventional subscription-based publishing and Green OA self-archiving can co-exist peacefully, even in fields where Green OA self-archiving has already reached 100% years ago.

Stevan Harnad
American Scientist Open Access Forum