Drafting a policy
From Harvard Open Access Project
- This is a section within Good practices for university open-access policies.
What an OA policy can achieve
- In this guide, we present our understanding of good practices for university open-access policies. An effective OA policy can build support for OA, as an academic and social good, into standard university practice.
- As we discuss below, we recommend a policy that provides for automatic default rights retention in scholarly articles and a commitment to provide copies of articles for open distribution. Policies of this sort have many benefits: they allow authors to retain extremely broad use and reuse rights with a minimum of effort; they allow universities to help authors in openly distributing articles for maximum impact; they allow other researchers and the general public to obtain broader access to articles; and they support these benefits without the need to negotiate with publishers and while preserving academic freedom, author choice, and consistency with copyright law.
- Although we find this kind of policy preferable, alternative sorts of policies can also be effective, and we discuss them as well. Some kinds of policies we find counterproductive, and we recommend avoiding them.
Statement of goals of the policy
- Many policies open with some statement of the policy goals. There is no "best practice" statement of the benefits of OA or the goals of promoting OA. But there are some mistakes to avoid.
- Don't say that the purpose of the policy is "only", "solely", or "exclusively" to achieve one benefit of OA, or some particular list of benefits. Leave the door open to achieve all the benefits of OA, even if you are not ready to enumerate them all.
- If you want to permit all the benefits of OA, then a narrow statement of the policy's purpose could give unwanted support to a plaintiff, court, or future administrator at your own institution trying to force a narrow reading on the policy. Even an innocent-seeming phrase like "for the purpose of open dissemination" could be interpreted later to prevent text mining, or to prevent the institution from transferring rights back to the author. Finally, any clause limiting the range of non-exclusive rights that authors grant to the university will in turn limit the range of rights that the university could later transfer back to the author.
Types of policy
- There are at least six types of university OA policy. Here we organize them by their methods for avoiding copyright troubles.
- The policy grants the institution certain non-exclusive rights to future research articles published by faculty. This sort of policy typically offers a waiver option or opt-out for authors. It also requires deposit in the repository.
- We recommend type #1 in this guide. Most of the good practices collected here are about that sort of policy.
- The policy requires faculty to retain certain non-exclusive rights when they publish future research articles. Whether or not it offers a waiver option for authors, it requires deposit in the repository.
- We do not recommend #2 because it requires faculty to negotiate with publishers in order to retain the needed rights. That is difficult to do. Many faculty are intimidated by the prospect and will not to do it. Even if all tried it, some will succeed and some will fail. Some will get one set of rights and some will get another. That will make access uneven and multiply implementation headaches.
- The policy seeks no rights at all, but requires deposit in the repository. If the institution already has permission to make a work OA, then it makes it OA from the moment of deposit. Otherwise the deposit will be "dark" (non-OA) until the institution can obtain permission to make it OA. During the period of dark deposit, at least the metadata will be OA.
- When type #1 policies are politically unattainable on a certain campus, then we recommend type #3. We prefer #1 to #3 because #1 provides permission to make articles OA through the repository and #3 does not.
- The policy seeks no rights at all and does not require dark deposits. It requires repository deposit and OA, but only when the author's publisher permits them.
- We do not recommend #4 because it allows recalcitrant publishers to opt out at will. Some institutions believe that a loophole for recalcitrant publishers is the only way to avoid copyright infringement. But that is mistaken. All six approaches listed here, properly implemented, avoid copyright infringement.
- Similarly, some institutions believe that an opt-out for authors, as in #1, is the same as an opt-out for publishers, as in #4. But that is also mistaken. Publishers have reasons or incentives to opt out far more often than authors.
- The policy does not require OA in any sense, but merely requests or encourages it.
- When #1 and #3 are both politically unattainable on a certain campus, we recommend either a type #5 policy or waiting until the community is ready for a type #1 or #3 policy.
- The policy does not require OA in any sense, but asks faculty to "opt in" to a policy under which they are expected to deposit their work in the repository and authorize it to be OA.
- We do not recommend #6 because it is equivalent to no policy at all. Faculty may already opt in to the practice of self-archiving and OA. This sort of policy differs little from #5 except by leaving the impression that asking faculty to opt in to an OA policy is somehow different from requesting or encouraging OA itself.
- The policy grants the institution certain non-exclusive rights to future research articles published by faculty. This sort of policy typically offers a waiver option or opt-out for authors. It also requires deposit in the repository.
- For independent analyses concluding that type #1 policies are lawful, and provide legally sufficient permission for OA through the institutional repository, at least in the United States, see:
- Simon Frankel and Shannon Nestor, Opening the Door: How Faculty Authors Can Implement an Open Access Policy at Their Institutions, a white paper from SPARC and Science Commons, August 2010. The paper shows how OA policies can avoid legal pitfalls, and uses the Harvard and MIT policies as a model.
- Eric Priest, Copyright and the Harvard Open Access Mandate, Northwestern Journal of Technology and Intellectual Property, preprint August 1, 2012, published version forthcoming. Also see Stuart Shieber's blog post on Priest's article, Is the Harvard open-access policy legally sound? The Occasional Pamphlet, September 17, 2012.
- On our preference for type #1 and type #3 policies over the other four types, see Recommendation 1.1 from the BOAI-10 statement (September 2012): "When publishers will not allow OA on the university’s preferred terms, we recommend either of two courses. The policy may require dark or non-OA deposit in the institutional repository until permission for OA can be obtained. Or the policy may grant the institution a nonexclusive right to make future faculty research articles OA through the institutional repository (with or without the option for faculty to waive this grant of rights for any given publication)."
Grant of rights to the institution
- The policy should be worded so that the the act of adopting the policy is the same as the act of granting the university certain non-exclusive rights. The policy should not merely ask, encourage, or require faculty to retain certain rights in the future, when they sign publishing agreements. It should say, "Each faculty member grants...", or "hereby grants...", not "will grant..." or "must grant...."
- By granting the rights at the time of the vote for the policy, in advance of future publications, the policy frees faculty from the need to negotiate with publishers. It secures the rights even when faculty fail to request them. It secures the same rights for every faculty member, not just the rights that a given faculty member might succeed in obtaining from a given negotiation with a given publisher.
- Some policies start with the grant of rights that we recommend, but then muddy the waters with confusing or even inconsistent additional language.
- One mistake is to accompany the grant of rights with a provision encouraging faculty to negotiate with publishers to retain some or all of the same rights already granted to the institution. This is confusing because one purpose of the grant of rights is to make that kind of negotiation unnecessary. The two clauses might even be inconsistent, one making negotiation unnecessary for OA, and the other implying that negotiation is necessary. (A negotiation clause would be more justified if it aimed to insure that authors only sign contracts consistent with the policy; for more on this, see our entry on author addenda.)
- Another mistake is to accompany the grant of rights with a provision creating a loophole for publishers whose publication agreements, or in-house copyright policies, do not allow OA on the university's terms. This is confusing because one purpose of the grant of rights is to close exactly that sort of loophole. The two clauses might even be inconsistent, one implying that publishers have no opt-out (except by requiring authors to obtain waivers), and other implying that publishers may opt out at will.
- Another mistake is to grant rights to "published scholarly articles" rather than to "scholarly articles" more broadly. This language could easily be interpreted to mean that the author grants no rights to the institution until the article is published. By then, of course, many authors will have already signed publishing contracts, and will have far fewer rights to grant to the institution. Often they will not have enough rights to authorize OA through the institutional repository. The same problem could arise if the grant of rights is limited to "peer-reviewed scholarly articles", because by the time an article is peer-reviewed, many authors will already have signed copyright transfer agreements with publishers. The key purpose of the rights-granting provision of the policy is to grant a wider rather than a narrower set of non-exclusive rights to the institution, and to do so before the author signs a publishing contract and loses the ability to grant such a wide range of rights. If the institution wishes to limit repository deposits to a certain subset of scholarly articles, such as those that are peer-reviewed and/or published, it can say so elsewhere in the policy or in its implementation plan.
- For reasons to grant a wider rather than a narrower range of non-exclusive rights to the institution, see the entry above on stating the goals of the policy.
- Note that in what follows we'll often refer to the grant of rights as the "license" or "permission" for OA.
- For the relationship of this grant of rights to the work-for-hire doctrine, see our entry on academic freedom.
Deposit in the repository
- The policy should either require deposit of relevant work in the institutional repository, or require making relevant work available to the institution for deposit.
- The waiver option should apply only to the grant of rights, not to deposit in the repository. (More under waivers below.)
- The policy needn't require faculty to make deposits themselves. The deposits may be made by others (such as student workers) on behalf of faculty, provided that faculty make the appropriate versions of their articles available for deposit. For simplicity in what follows, we will refer to depositors as faculty, but will mean to include others acting on behalf of faculty.
- The policy should specify that the deposited version should be the final version of the author's peer-reviewed manuscript, sometimes called the accepted author manuscript (AAM). This version contains the text approved by peer review. It should also include all the charts, graphics, and illustrations which the author has permission to deposit. It should include post-review copy-editing done collaboratively between author and journal. It need not include any post-review copy editing done unilaterally by the journal, the journal's pagination, or the journal's look and feel.
- If the publisher consents, then the institution should deposit the published version of an article to complement the final version of the author's peer-reviewed manuscript already on deposit.
- This could be mentioned in the policy itself or simply made an implementation practice.
- The published version should only replace the author's manuscript when the published version allows at least as many reuse rights as the author's manuscript. Some publishers will be happy to make this substitution in order to prevent the circulation of multiple versions. However, when the published version carries a more restrictive license than the author's manuscript, then the published version should be deposited alongside the accepted author manuscript, and the latter should not be removed from the repository.
- SHERPA RoMEO keeps a list of publishers willing to allow deposit of the published version.
- The policy should require faculty to deposit their peer-reviewed manuscripts at the time of acceptance for publication, or no later than the date of publication.
- If the policy allows authors to specify an embargo on a given article, the deposit should still be made between the time of acceptance and the time of publication. But it will be a dark deposit until the embargo period expires.
- The policy should make clear that the institution will always grant waivers, no questions asked. Faculty needn't offer a justification or meet a burden of proof. To prevent needless fear or confusion on this point, the policy should refer to "obtaining" a waiver, or "directing" that a waiver be granted, rather than "requesting" a waiver.
- To allay potential faculty concerns that an institution may override a waiver in the future, the waiver should contain language that it may not be revoked by the institution.
- The waiver option should apply only to the license or grant of rights to the institution, not to the deposit in the repository. Faculty should deposit their articles in the repository even when they obtain waivers. At least initially, these would be dark or non-OA deposits.
- Faculty who want waivers for separate publications should obtain separate waivers. Institutions should not offer "standing waivers" that apply to all future publications from a given faculty member. Standing waivers would defeat the purpose of shifting the default to permission for OA.
- A waiver for a particular article means that the institution does not receive the policy's usual bundle of non-exclusive rights for that article. Hence, for that article the university will not have permission from the policy to provide OA. But the university may have permission from another source, such as the author (who may have retained rights from the publisher) or the publisher (who may give standing permission for repository-based OA after a certain embargo period).
- For example, if the publisher allows repository-based OA six months after publication, then the university will eventually have permission from the publisher even if it doesn't have permission from the policy. If the university has a copy of the article on dark deposit in the repository, then it may make the repository copy OA as soon as the embargo runs or the new permission takes effect.
- Hence, the waiver provision of the policy should not promise that the university will never make a copy OA. On the contrary, the policy might say that the university will make faculty work OA whenever it has permission to do so.
- Some supporters of OA worry that a waiver option will make the policy ineffective. They worry that the waiver rate will be high, for example, above 50%. However, the experience at every school with a waiver option is that the waiver rate is low. At both Harvard and MIT it's below 5%.
- Omitting a waiver option would limit faculty freedom to submit new work to the journals or publishers of their choice. Including a waiver option restores that freedom but without impeding OA. The kind of policy we recommend shifts the default to OA. It uses faculty inertia to support OA rather than to support standard copyright transfers which give the OA decision to publishers. Faculty who worry that a waiver option entails a high waiver rate should not underestimate the power of shifting the default. It can and does change behavior on a large scale.
- In this guide we use the terms "waiver" and "opt-out" interchangeably.
- The policy may also give authors the right to specify an embargo period (a delay in the open distribution of an article).
- The Duke policy is a model here: "The Provost or Provost's designate will waive application of the license for a particular article or delay access for a specified period of time upon written request by a Faculty member."
- Harvard's Model Open Access Policy incorporates the Duke language with this annotation: "Duke University pioneered the incorporation of an author-directed embargo period for particular articles as a way of adhering to publisher wishes without requiring a full waiver. This allows the full range of rights to be taken advantage of after the embargo period ends, rather than having to fall back on what the publisher may happen to allow. Since this is still an opt-out option, it does not materially weaken the policy. An explicit mention of embargoes in this way may appeal to faculty members as an acknowledgement of the prevalence of embargoes in journals they are familiar with."
- When faculty specify an embargo period, they should still deposit their articles in the repository on the usual timetable. The embargo option allows a delay in making a deposited article OA, not a delay in the initial deposit.
- For schools and authors, embargoes are much better than waivers.
- When the author obtains a waiver, neither the school nor the author may exercise the large bundle of non-exclusive rights granted by the policy. When the author obtains an embargo and not a waiver, both the school and the author may exercise those rights, with the temporary exception that they may not provide OA for the length of the embargo. Needless to say, the best situation is when authors obtain neither a waiver nor an embargo.
- If you wonder why avoiding waivers is better authors, and not just for institutions, see our entry on transferring rights back to the author. Under the kind of policy we recommend, institutions with non-exclusive rights granted by the policy may grant them back to the authors, giving authors far more rights to reuse their own work than they would have had without this type of policy. Authors who want to maximize their rights to reuse their own work should be the first to try to avoid obtaining a waiver.
- In our experience, many authors and publishers who want waivers really want embargoes, or would be satisfied with embargoes. Hence, when possible, see whether those seeking waivers would accept embargoes instead.
- We recommend against any policy language, or implementation practice, requiring the university to respect a given embargo period for all articles from a given journal or publisher, at least without a significant concession from the journal or publisher in exchange. For more details, see the entry on treaties with publishers.
Scope of coverage, by content category
- The policy should specify what categories of content are covered by the license and the expectation of deposit. In particular, the policy should cover scholarly articles, or the kinds of writings typically published in peer-reviewed journals and conference proceedings.
- The policy should not cover writings not considered scholarly in the field (in most fields, op-ed pieces, popular articles) or scholarly writings that generate royalties (textbooks, monographs).
- The Harvard model policy covers "scholarly articles" alone, and explains in this annotation:
- What constitutes a scholarly article is purposefully left vague. Clearly falling within the scope of the term are (using terms from the Budapest Open Access Initiative) articles that describe the fruits of scholars' research and that they give to the world for the sake of inquiry and knowledge without expectation of payment. Such articles are typically presented in peer-reviewed scholarly journals and conference proceedings. Clearly falling outside of the scope are a wide variety of other scholarly writings such as books and commissioned articles, as well as popular writings, ﬁction and poetry, and pedagogical materials (lecture notes, lecture videos, case studies).
- Often, faculty express concern that the term is not (and cannot be) precisely deﬁned. The concern is typically about whether one or another particular case falls within the scope of the term or not. However, the exact delineation of every case is neither possible nor necessary. In particular, if the concern is that a particular article inappropriately falls within the purview of the policy, a waiver can always be obtained.
- One tempting clarification is to refer to scholarly articles more specifically as "articles published in peer-reviewed journals or conference proceedings" or some such specification. Doing so may have an especially pernicious unintended consequence: With such a definition, a "scholarly article" doesn't become covered by the policy until it is published, by which time a publication agreement covering its disposition is likely to already have been signed. Thus the entire benefit of the policy's nonexclusive license preceding a later transfer of rights may be vitiated. If clarifying language along these lines is required, simultaneously weaker and more accurate language can be used, for instance, this language from Harvard's explanatory material (also used above): "Using terms from the Budapest Open Access Initiative, faculty's scholarly articles are articles that describe the fruits of their research and that they give to the world for the sake of inquiry and knowledge without expectation of payment. Such articles are typically presented in peer-reviewed scholarly journals and conference proceedings."
- Works not covered by the policy can still be placed in a repository, and with permission they can be made OA. The policy or separate implementation documents can encourage deposit of other kinds of work that fall outside the scope of the license and deposit requirement.
Scope of coverage, by time
- Neither the grant of rights nor the deposit requirement should be retroactive. Under the kind of policy we recommend here, faculty can only make the desired grant rights to the institution for future, still-unpublished works, not for previously published works.
- However, the policy or separate implementation documents might encourage deposit of works completed prior to the adoption of the policy.
- The kind of policy we recommend here not only grants rights to the institution, but allows the institution to grant those rights to others. Here's the key language (from the Harvard model policy): "More speciﬁcally, each Faculty member grants to [university name] a nonexclusive, irrevocable, worldwide license to exercise any and all rights under copyright relating to each of his or her scholarly articles...and to authorize others to do the same" (emphasis added).
- The primary purpose of this language is to allow the institution to grant rights back to the author. The effect is that authors retain or regain certain rights to their work, including rights that they might have transferred away in their publishing contracts.
- This gives authors far more rights to reuse their own work than (1) they have under standard publishing contracts or (2) they have under other types of OA policy.
- This not only helps access, use, and reuse. It promotes author freedom. Hence, when well-explained, it also helps muster faculty support for the policy in the first place.
- For this reason, the set of rights transferred to the institution should be as broad as possible. That enables the author to retain or regain the broadest possible set of rights.
- Although the kind of policy we recommend here can correctly be called a rights-retention policy, it doesn't provide direct or simple rights retention by authors. Instead it provides direct rights retention by institutions, and indirect rights retention by authors.
Transferring rights to others
- Authors subject to this kind of policy may still sign publishing contracts with publishers. The policy grants certain non-exclusive rights to the institution, and authors should not sign contracts giving the same rights to publishers (or other parties). However, they will never need to do so. The vast majority of publishers agree that they can obtain the rights they need for publication without requiring authors to obtain waivers. But when authors wish to publish with a publisher who thinks otherwise, they may obtain a waiver, no questions asked.
- For detail on alerting publishers to the rights already granted to the institution, see the entry on author addenda. For detail on waiving the grant of rights to the institution for a given work, see the entry on waivers.
Enhancing user rights
- Authors subject to this kind policy may use open licenses, such as Creative Commons licenses, to enhance user rights. The kind of policy we recommend here is compatible with the use of open licenses but does not require them. Institutions may adopt this kind of policy and decide afterwards when or whether to make use of open licenses. Similarly, it may adopt this kind of policy and leave authors free to make these decisions on their own, case by case.
- The policy should include a provision making a certain office or committee responsible for implementing the policy.
- A policy is more likely to pass if it only says what it has to say. Other details can be left to the office charged with implementing the policy.
- When it's desirable to share both the draft policy language and the implementation plan, make sure to keep the two distinct. That way the policy itself is not enlarged to include the implementation plan, and can remain brief and minimal. In addition, it gives the implementation group the flexibility to adjust its plan, within the guidelines of the policy, to suit changing circumstances.
Separating the issues
- A university requiring green OA (deposit in OA repositories) may also encourage gold OA (publishing in OA journals). But it should be careful about doing both in the same document. Where it has been tried, faculty tend to assume that the policy requires gold OA, or publishing in OA journals, and thereby limits their freedom to submit new work to the journals of their choice.
- Part of the background here is that many people still mistakenly believe that all OA is gold OA, and therefore that a policy trying to assure OA must be trying to assure gold OA or to require publishing in OA journals.
- This is such a serious problem that if the policy document mentions gold OA at all (using any terminology, such as "OA journals", "OA publishers", or "OA publishing"), then it should only be to make clear that the university is not considering a gold policy, and that the policy will preserve faculty freedom to publish wherever they wish.
- A university with a green OA policy may (and we think, should) also launch a fund to help faculty pay publication fees at fee-based OA journals. But the green OA policy should make clear that it is separate from the journal fund. Otherwise faculty may think that the policy itself requires faculty to submit new work to OA journals, a common and harmful misunderstanding.
- We offer some other recommendations on separating the issues in the section on Adopting a policy.
Return to the table of contents.