Implementing a policy

• This is a section within Good practices for university open-access policies.

Launching a repository

• The institution must have an institutional repository, or participate in a consortial repository. Most schools launch a repository before adopting a policy to fill it, but some do it the other way around.

Individualized writing

• Institutions implementing the kind of policy recommended here will want the policy to prevail over a later publishing contract inconsistent with the policy. Merely passing the policy may attain that goal. However, to be more certain, practically and legally, that the policy license survives any later transfer, US institutions should get authors to sign a "written instrument" affirming the policy.
  • Here's why: Under US copyright law (17 USC 205.e) a "nonexclusive license...prevails over a conflicting transfer of copyright ownership if the license is evidenced by a written instrument signed by the owner of the rights licensed or such owner's duly authorized agent."
  • This provision doesn't say that in the absence of a written instrument, the nonexclusive license will not prevail over a later contract inconsistent with the policy. A university might take the position that the nonexclusive license in the policy will prevail in any case, and will probably never have to test its position in court. But to be safe, it's best to get a written affirmation of the grant of rights (or license) as specified by 17 USC 205.e.
  • We don't know how to accomplish this goal outside the US, and welcome advice from people who do know.

• Harvard uses several methods to get the written affirmation of the policy. When faculty deposit their own articles, a dialog box in the deposit process asks them to affirm the grant of rights (the license) in the policy. When someone else (an administrative assistant or the Office for Scholarly Communication) deposits articles on their behalf, the faculty member must first have signed a one-time assistance authorization form containing an affirmation of the grant of rights. Thus, whatever route an article takes into the repository, the institution obtains a written affirmation of the license.
  • Here's Harvard's language for affirming the license is: "[I]f I am a member of a Harvard Faculty or School that has adopted an open access policy found at http://osc.hul.harvard.edu/, this confirms my grant to Harvard of a non-exclusive license with respect to my scholarly articles as set forth in that policy."

• Finally, the individualized writing documents the consent of faculty hired by the institution after the adoption of the policy.

Facilitating waivers

• The institution should create a web form through which faculty can obtain waivers. This not only streamlines bookkeeping, but proves to faculty that the process is easy and automatic. Harvard can share code for such a web form.

• Some publishers may require faculty to obtain a waiver as a condition of publication. Institutions need not try to prevent this. Accommodating these publisher policies proves that publishers have the means to protect themselves, if they choose to use them, and that fact makes it unnecessary for faculty to protect or "paternalize" their favorite publishers (e.g. society publishers) by voting against a proposed policy. On the other hand, the institution may want to talk with publishers who take this position, to see whether they can work out an accommodation.

Author addenda

• An author addendum is one way for authors to retain rights that a standard publishing contract would otherwise give to the publisher. For policies of the kind we recommend, author addenda are unnecessary for rights retention, for the same reason that individual author-publisher negotiations are unnecessary. The institution has the rights needed for OA directly from the grant of rights in the policy. Hence, faculty need not obtain those rights from publishers.

• However, author addenda may be desirable for other reasons.
  • An addendum alerts the publisher that the author's institution already possesses certain non-exclusive rights. This can prevent misunderstandings on each side.
  • An addendum also modifies the contract to make it consistent with the university's OA policy. (Note that some but not all unmodified contracts are already consistent with the policy.) Without this modification, some authors could sign contracts inconsistent with the policy.
  • See the section on individualized writing above for the reasons why a well-implemented institutional OA policy would take priority over a later publishing contract inconsistent with the policy. Because the policy takes priority, authors who sign publishing contracts inconsistent with the policy may be unable to live up to those contracts. Because the policy takes priority and retains key rights, the risk is not copyright infringement but breach of contract. An addendum modifying the contract completely eliminates the risk of breach.
    • Note, however, that there may be no risk to eliminate. Under some legal theories, a widely-known prior license would protect the author from a claim of breach of contract, even in the absence of an addendum. This is one more reason to publicize the university's OA policy.

• Also see the entry below on working with publishers.

Multiple deposits

• If a faculty member deposits a paper in a non-institutional repository (e.g. arXiv, PubMed Central, SSRN), the institutional repository should harvest a copy.

• To avoid diluting the traffic numbers at the several repositories, all should comply with the (evolving) PIRUS and PIRUS2 standards for sharing traffic data.

• If a faculty member is subject to two OA policies (e.g. one from the university and one from the funder), the university should offer to make the deposit required by the funder.
  • For example, most faculty at Harvard Medical School are subject to the NIH policy. If they deposit in the HMS repository, then HMS will insure that a copy is deposited in PubMed Central. If faculty think that an institutional policy will double their administrative burden, they will vote against it.

Dark deposits

• Faculty should always deposit suitable versions of new scholarly articles in the institutional repository. If they obtain a waiver for a given article, then the deposit will at least initially be "dark" (or non-OA). But the author should still deposit the manuscript.
  • One reason for repositories to allow dark deposits is to support the message that faculty should always deposit their new work.

• If a deposit is dark, at least the metadata should be OA.
  • Another reason to allow dark deposits is to facilitate search indexing and discovery for work which, for one reason or another, cannot yet be made OA.

• If a deposit is only intended to be dark temporarily, for a known embargo period, then dark deposits should be set to open up automatically at the future date determined by the author decision or embargo period. Most repository software today supports this option.

• If an author deposited a manuscript and obtained a waiver, then the institution does not have permission under the policy to make that manuscript OA. At least initially, that deposit must be dark. However, the repository may switch the manuscript to OA if it can obtain permission from another source, such as a standing policy of the publisher's to allow OA after a certain embargo period. See the entry on waiver options. Repositories should make dark deposits OA whenever they are legally allowed to do so.

• For seven reasons why repositories should allow dark deposits, see Stuart Shieber, The importance of dark deposit, The Occasional Pamphlet, March 12, 2011.

Deposited versions

• Some authors will deposit the published version of an article instead of the final version of the author's peer-reviewed manuscript.
  • Some will mistakenly believe it is the version the policy asks them to deposit. Some will simply prefer it and demand to make it the OA version.
  • Unless the publisher consents to the open distribution of the published version, ask the author for the final version of the author's peer-reviewed manuscript. If the author can't find the right version or insists on depositing the published edition, then make it a dark deposit and open it up if and when the repository can obtain permission to make it OA.

Internal use of deposited versions

• When the institution reviews faculty publications for promotion, tenure, awards, funding, or raises, it should limit its review of research articles to those on deposit in the institutional repository. Or it should use the institutional repository as the mechanism for submitting articles for use or review by internal committees.

• Versions of this policy have been adopted at the Université de Liège, Edinburgh Napier University, the University of Oregon Department of Romance Languages, the Catholic University of Louvain, China's National Science Library, the Chinese Academy of Sciences, India's International Center for Tropical Agriculture, Canada's Institute for Research in Construction, the University of Salford, and the University of Luxembourg.

• This type of policy has been recommended in many major reports and analyses of best practices for university policies (here listed in chronological order):
  • The May 2010 Alhambra Declaration on Open Access recommended that universities should "consider[] repository-deposited material for evaluation processes and research assessment."
  • The September 2012 tenth-anniversary statement from the Budapest Open Access Initiative recommended (1.6) that "Universities with institutional repositories should require deposit in the repository for all research articles to be considered for promotion, tenure, or other forms of internal assessment and review....[This policy should not] be construed to limit the review of other sorts of evidence, or to alter the standards of review."
  • A September 2013 report from the UK House of Commons Select Committee on Business, Innovation and Skills acknowledged (paragraph 26) that "authors are much more likely to archive their research papers in their institutional repositories if they are required to do so as a condition of funding compliance and if deposit is linked to institutional performance evaluation, research grant applications and research assessment."
  • A November 2013 report from the Mediterranean Open Access Network (MedOANet) concluded (p. 12) that an OA "requirement should be linked to professional advancement and evaluation. Authoritative researcher, departmental and institutional publication lists should be directly drawn from the institutional repository for evaluation purposes, thus making clear to authors that this is the source that will be used for this purpose and that they therefore have a personal interest in making sure their work is fully represented in the repository."

• An analogous policy has been recommended for national research-assessment policies:
  • In a April 2003 article, Stevan Harnard argued that UK funding agencies "should mandate that in order to be eligible for Research Assessment and funding, all UK research-active university staff must maintain (I) a standardised online RAE-CV, including all designated RAE performance indicators, chief among them being (II) the full text of every refereed research paper, publicly self-archived in the university's online Eprint Archive and linked to the CV for online harvesting, scientometric analysis and assessment."
  • A February 2013 report from the Higher Education Funding Council for England recommended (Sections 11 and 12) that works not deposited in an OA repository immediately upon publication should not be eligible for the new Research Excellence Framework (REF), the UK's national research-assessment program. HEFCE adopted this policy in March 2014.

• When properly written and implemented, these policies would not alter the kinds of evidence that committees are willing to consider, and would not alter the standards they use in awarding promotion, tenure, or funding.

• Institutions not ready to change their process for promotion and tenure could change the form by which faculty apply for promotion and tenure and list their publications. The new form could simply add fields for the URLs of OA editions of the faculty member's research articles.

Associating articles with their definitive versions

• The author manuscript deposited in the repository is typically not identical to the definitive published version, and its provenance should be made clear. This can and should be done in at least two ways.

• First, each deposited article should be associated with the full citation for the published article. This may be done in a free text citation metadata field using any suitable citation style, or the equivalent information may be made available through a set of metadata fields providing journal name, volume, number, pages, etc.

• Second, it is a good idea to provide links from the repository to the online definitive version of the deposited article where available. For example, Harvard provides links to definitive versions...
  1. on search results pages associated with each search result,
  2. on item metadata pages, and
  3. on a cover page added to the front of the deposited PDF of the article.

Repository indexing

• The repository should be configured to support crawling by search engines.
  • See for example the JISC InfoNet recommendations and Google Webmaster Guidelines.

• Repository managers should check to see whether the contents are discoverable through major search engines, and follow up any indexing failures.

• This is not just a technical detail. Faculty who vote for an OA policy want to know that the resulting works will be discoverable through ordinary search engines. If faculty believe that deposit in the repository only benefits the rare user who makes a special visit to the repository and runs a local search, then many would vote against the policy or not bother to deposit their work.

Repository withdrawals

• If a publisher sends a reasonable takedown request to the repository, the repository should always comply.

• If the author wishes to withdraw an article already on deposit (e.g. because it is mistaken, embarrassing, superseded by a newer version, etc.), then the repository should withdraw the article. The author can always obtain a waiver, and then the university would no longer have the rights to distribute it under the policy. That's one reason why repositories should follow author wishes on distribution. Another is that repositories depend on faculty cooperation and good will. In any case, experience suggests that authors rarely ask to withdraw their own articles.

Content beyond the policy

• The institution should welcome the deposit of types of scholarly content above and beyond the types covered by the policy. For example, if the policy focuses on scholarly articles, the repository should welcome deposit of other genres as well, such as theses and dissertations, books or book chapters, datasets, and digitized work from other media for which it has permission to provide OA. If the policy covers articles published after a certain date, it should welcome the deposit of articles published before that date.

• Even if the policy only covers work by faculty, the repository should welcome deposits from scholars at the institution who are not faculty, such as students, research fellows, post-docs, staff, and administrators.

• Even if the policy only gives the institution permission to make certain kinds of content OA, the repository could accept dark deposits where it doesn't have permission for OA. In those cases it could at least provide OA to the metadata.

Treaties with publishers

• Some publishers may concur with the policy so long as certain aspects are clarified concerning how the policy will be implemented. Providing such clarifications may be entirely reasonable, given that the policy language itself can't possibly cover all aspects of its implementation. For example, publishers may want to be sure that for manuscripts published in their journals the repository entry will include a complete citation and link to the published edition, or that the university will not distribute the publisher's version of the article, or that the license will not be used to sell articles. If the institution is comfortable with these clarifications (indeed, the clarifications in the treaty may well be aspects of implementation to which the university is already committed), it may make these explicit in return for an explicit statement of the publisher's cooperation with the policy, for instance, by not requiring waivers or addenda to publication agreements. These agreements may contain any provisions consistent with the policy and agreeable to both sides. (At Harvard they are called "treaties".)

• We strongly recommend against treaties requiring universities to respect a given embargo period for all articles from a given journal or publisher. Such a treaty would essentially give the journal or publisher a blanket opt-out of a significant provision of the university OA policy, and violate the express interest of the faculty in adopting a policy to shift the default to immediate OA.
  • However, when authors rather than publishers seek an embargo, and seek it case by case rather than for all articles from a certain journal or publisher, the policy can accommodate them. See the entry on embargo options.

• Here's an example of treaty language used at Harvard.

Learning the denominator

• An institution can easily tell how many articles are on deposit in its repository. But it cannot easily tell how many articles ought to be on deposit. If it wants to calculate the deposit rate (the number deposited divided by the number that ought to be deposited), then it must determine the denominator. This is a critical piece of information in measuring the effectiveness of the policy and its implementation.

• Some institutions ask faculty to submit an annual list of their publications. If so, the information should be shared with the repository managers. The raw list of publications is less helpful than one broken down by categories, such as books, journal articles, and so on. If the policy only covers journal articles (for example), then the relevant denominator is the number of journal articles.

Working with publishers

• See the entry on author addenda. A well-written author addendum can explain to publishers what rights the author has already assigned to the institution. Hence it can prevent authors from signing publishing contracts they cannot fulfill and prevent misunderstandings on all sides. However there are other ways to achieve some of the same goals.

• Publishers who normally require the transfer of exclusive rights, but who do not demand waivers from authors at your institution, can modify their publishing contracts to facilitate cooperation with the institution. For example, it would help both sides if publishers included a sentence like this one from the Science Commons addendum: "Where applicable, Publisher acknowledges that Author's assignment of copyright or Author's grant of exclusive rights in the Publication Agreement is subject to Author's prior grant of a non-exclusive copyright license to Author's employing institution and/or to a funding entity that financially supported the research reflected in the Article as part of an agreement between Author or Author's employing institution and such funding entity, such as an agency of the United States government."
  • Such a clause would make addenda unnecessary for authors and publishers, and cost the publisher nothing.

Tracking usage stories

• MIT pioneered a technique for tracking stories about how articles they provide from their repository are being used. Harvard and perhaps others have copied the technique as well. The technique is to inject an extra page at the front of the PDF of a distributed article (which the repository may already be doing to provide citation and licensing information), which provides a statement requesting information about how the article was used with a link to a web form to provide the statement. The MIT language is:

The MIT Faculty has made this article openly available. Please share how this access benefits you. Your story matters.

• The stories can then be compiled and shared. For example, see the stories from MIT and Harvard.
• In the web form, you may want to request information about the article's user as well as the identity of the article itself (or this latter can be automatically provided in the link, as in the Harvard implementation). However, all of this information should be supplied only optionally.

---

Return to the table of contents.