Self-Archiving Mandates

---

SUMMARY: Eight supplemental points based on Peter Suber's excellent talk (and the audience discussion) on "What Can Universities Do To Promote Open Access?" at Harvard's Berkman Center for Internet and Society.
(1) Journals vs. books: OA is only about author give-away work. Peer-reviewed journal articles are all, without exception, author give-aways, but most scholarly books are not. OA can only be mandated for give-away work. (Once OA for journal articles prevails, more authors will undoubtedly want the same for their monographs too.)
(2) Versions and Citability: The canonical version of a journal article is the final, peer-reviewed, accepted version (the "postprint"). That is what researchers need, though not necessarily in the form of the publisher's PDF. What is cited is always the published work. Researchers are infinitely better off if those who cannot afford the publisher's official PDF can always access the author's self-archived postprint.
(3) First OA Self-Archiving Mandate: Queensland University of Technology's was the world's first university-wide OA self-archiving mandate, as Peter notes, but the very first OA self-archiving mandate of all was that of the School of Electronics and Computer Science at Southampton University.
(4) Prior Evidence of Probability of Compliance With OA Self-Archiving Mandates: Swan & Brown's author surveys found that 95% of authors would comply with an OA self-archiving mandate (over 80% willingly) but authors were not asked whether they would comply with a copyright-retention mandate. The same is true of Arthur Sale's data on actual mandate compliance rates.
(5) Deposit Mandates vs. Copyright-Retention Mandates: NIH's is not a copyright-retention mandate. It is a no-opt-out deposit mandate plus a no-opt-out requirement to negotiate with the 38% of journals who don't endorse immediate OA, so as to be able to make the deposit OA within a year. Harvard's is a copyright-retention mandate, with opt-out.
(6) Mandate Implementation Mechanisms: There are no sanctions on deposit mandates, as Peter notes; there are administrative incentives and contingencies: The IR is made the official locus for submitting publications to be assessed for performance review.
(7) Peer Review, Journals and Repositories: Journals provide peer review; IRs provide access to peer-reviewed postprints. The issue of IRs providing peer review is a red herring (raised by others, not Peter).
(8) Journal Weighting in Researcher Performance Evaluation: [added 5 April] The credit and weight accorded for publishing in a given journal in a researcher's performance evaluation should depend only on the journal's track-record for quality, not on its OA policy or status.

---

Peter Suber gave an excellent talk at Harvard's Berkman Center for Internet and Society entitled: "What Can Universities Do to Promote Open Access?" and the discussion was very interesting too. The video and powerpoints are here.

I append eight comments stimulated by the talk and accompanying discussion; many are just elaborations on points Peter made:

(1) Journals vs. books: For the OA movement it is ever so important to clearly separate the case of journal articles from the case of books. The reason is simple: OA is and has to be only about author give-away work. Peer-reviewed journal articles are all, without exception, author give-aways, written only for uptake, usage and impact, not for royalty revenues.

Now although this may also be true of some scholarly books, it definitely is not true of all or most of them now. Hence not only can OA not be mandated for non-give-away books, but, at a time when OA itself is still so widely misunderstood, it is important to treat the clearcut, exception-free give-away case of peer-reviewed journal articles first, and separately, rather than to conflate it with the complicated hybrid case where the majority of the content is not author give-away at this time.

Once OA for journal articles prevails, more authors will undoubtedly want the same for their monographs too.

(2) Versions and Citability: The problems raised during the question period concerning the versions and citability problem are mostly a matter of misunderstanding:

First, we are talking about journal articles only.

The canonical version of a journal article is the final, peer-reviewed, accepted draft (the "postprint"). That is what researchers need, not necessarily the publisher's PDF.

What is cited is always the published work (unless one is explicitly and deliberately referring to unrefereed, unpublished prior drafts [preprints] or corrected, revised postpublication updates [which we could call "post-postprints"]). The version and citation issue pertains only to the specific case where one deliberately wishes to cite either an unpublished draft or an unpublished revision.

Otherwise, all citations of the peer-reviewed article itself -- whether based on reading the author's self-archived postprint of it, or the publisher's PDF of it -- are citations to the canonical published work itself (and point to the bibliographic data for the published work).

In other words: there is no special version or citation problem for postprints.

Now, as to the separate scholarly question of whether an author can be trusted if he says that "this draft of my published article is indeed the refereed final draft" -- that is a matter for scholarly practice and integrity. It is not an OA issue. It is not even a technical issue (although there are technical ways of computationally comparing versions to check whether and how a given draft diverges from the published PDF or XML text).

The only relevant point is that the scholarly and scientific research world is infinitely better off if all those scholars and scientists who cannot access the publisher's official PDF of any given article can always access the author's self-archived postprint of it. The possibility that some authors may sometimes be either untruthful or sloppy can be handled on a case by case basis if/when it ever comes up. But that possibility is definitely no reason to call into question the basic principle that what researchers need today is access to the postprint. And that it is in the authors' and their institutions' and their funders' best interests that authors should provide access to their postprints, by self-archiving them in their IRs. Inasmuch as self-archiving the publisher's PDF creates obstacles to self-archiving (because the publisher does not allow it, or because access to it is embargoed), that should definitely not be grounds for delaying the immediate provision of the author's postprint -- or for delaying the adoption of mandates to deposit it.

(3) First OA Self-Archiving Mandate: For the record: Queensland University of Technology's was indeed the world's first university-wide OA self-archiving mandate, as Peter noted, but not the world's first OA self-archiving mandate.

The very first OA self-archiving mandate was that of the School of Electronics and Computer Science at Southampton University in 2003: That was also the model for the first formal description of an OA self-archiving mandate in the BOAI Self-Archiving FAQ and the OSI EPrints Handbook.

(4) Prior Evidence of Probability of Compliance With OA Self-Archiving Mandates: It is not correct to say that the Swan & Brown author surveys found that authors would comply, and comply willingly, with any OA mandate at all. Authors were asked specifically about a mandate to deposit, and 95% said they would comply, with 81% complying willingly. They were not asked, however, about how they would feel about a mandate to retain copyright (let alone a mandate with an opt-out). Harvard's is the first copyright-retention mandate, and there is no evidence at all on how many faculty would comply, or comply willingly.

Arthur Sale's data on actual compliance rates likewise apply only to deposit mandates, not to copyright-retention mandates.

(5) Deposit Mandates vs. Copyright-Retention Mandates: NIH's is not a copyright-retention mandate. It is a deposit mandate. It can be fulfilled by simply depositing the postprint of an article that was published in any of the 62% Green journals that have endorsed immediate OA self-archiving, or any of the remaining journals that have endorsed embargoed access within NIH's time limit. That is a deposit mandate -- plus a requirement (with no opt-out option) to negotiate deposit within the embargo period for any remaining articles, published in journals that don't endorse immediate OA self-archiving. (Harvard's in contrast, is, in its present form, purely a copyright-retention mandate, with an opt-out.)

(6) Mandate Implementation Mechanisms: There are no sanctions on deposit mandates, but there are administrative incentives and contingencies, as Peter noted. Basically, if you wish to have your publications considered for institutional performance review, the official mechanism for doing so is to deposit them in your IR. (There is a similar rationale for fulfilling the Harvard copyright-retention mandate (for the non-opt-outs): one way to meet the condition of transmitting the postprint to the Provost is to deposit it directly in Harvard's IR.)

(7) Peer Review, Journals and Repositories: Very minor point: OA is about providing OA to peer-reviewed journal articles. OA journals can provide OA to their articles, or authors can provide OA to their articles (by depositing them in their IR). Singling out the fact that an IR does not provide peer review is a bit of a red herring. It risks encouraging people to speculate about alternatives to journals, instead of just focussing on providing OA to peer-reviewed journal articles. (Peter himself is of course very clear on this.)

(8) Journal Weighting in Researcher Performance Evaluation: In evaluating research performance, there need be no extra credit or weight accorded for publishing in a journal based on its OA policy or status. The weight should depend only, as always, on the journal's track-record for quality. OA can be provided by self-archiving articles published in any journal; there is no need for a researcher to select journals for any other reason than research quality.

Stevan Harnad
American Scientist Open Access Forum