Skip to main content
SearchLoginLogin or Signup

7-5. The Open Access Mandate at Harvard

Published onMar 14, 2019
7-5. The Open Access Mandate at Harvard

From “The open access mandate at Harvard,” SPARC Open Access Newsletter, March 2, 2008.

Harvard's new OA policy is not the first university-level OA mandate, but it's the first in the US, the first to be adopted by faculty rather than administrators, the first adopted policy to focus on permissions rather than deposits, and the first to catch the worldwide attention of the press and blogosphere.

Here's the heart of it:

Each Faculty member grants to the President and Fellows of Harvard College permission to make available his or her scholarly articles and to exercise the copyright in those articles. In legal terms, the permission granted by each Faculty member is a nonexclusive, irrevocable, paid-up, worldwide license to exercise any and all rights under copyright relating to each of his or her scholarly articles, in any medium, and to authorize others to do the same, provided that the articles are not sold for a profit. The policy will apply to all scholarly articles written while the person is a member of the Faculty except for any articles completed before the adoption of this policy and any articles for which the Faculty member entered into an incompatible licensing or assignment agreement before the adoption of this policy. The Dean or the Dean's designate will waive application of the policy for a particular article upon written request by a Faculty member explaining the need.

To assist the University in distributing the articles, each Faculty member will provide an electronic copy of the final version of the article at no charge to the appropriate representative of the Provost's Office in an appropriate format (such as PDF) specified by the Provost's Office. The Provost's Office may make the article available to the public in an open-access repository.

The Office of the Dean will be responsible for interpreting this policy, resolving disputes concerning its interpretation and application, and recommending changes to the Faculty from time to time.

Text of the resolution adopted by the Harvard Faculty of Arts and Sciences, February 12, 2008


(Disclosure: I consulted informally with proponents of the Harvard policy several times over the past few years.)

Here are some notes and reflections:


So far, it applies only to Harvard's Faculty of Arts and Sciences (FAS), not to the business school, law school, medical school, or other schools within the university. However, the Harvard library is actively taking the policy to the rest of the institution. As Robert Darnton told Library Journal Academic Newswire: “My position is to spread the FAS motion … throughout the whole university. That is going to be one of my top priorities in the weeks and months ahead. I will be discussing this with the law school, medical school and business schools.” Stay tuned.

If the Harvard policy is not the first university-level OA mandate, then what was?

Preceding the Harvard mandate are at least 12 other institutional mandates and three departmental mandates in nine countries: Australia, Belgium, France, India, Portugal, Russia, Swizterland, Turkey, and the UK.

The pioneers on this frontier are Queensland University of Technology in Australia, whose mandate took effect on January 1, 2004, and the University of Minho in Portugal, whose mandate took effect one year later on January 1, 2005.

If a dozen earlier university mandates came first, why was Harvard's was the first to grab the attention of the press and blogosphere? The answer is pretty clear: because it's Harvard and because this policy was adopted by the faculty itself.

Did I mention that the Harvard policy was adopted by the faculty itself? Did I say that the vote was unanimous?

The policy has critical support beyond the faculty, for example from Provost Steven Hyman and University Librarian Robert Darnton. But Professor of Computer Science Stuart Shieber, the policy's chief strategist and advocate, deserves special kudos for choosing to make the policy rest on faculty support and for his patient campaign of information and persuasion, which culminated in the unanimous faculty vote.

This is strong: Imagine any faculty voting unanimously for any interesting policy.

The publishing lobby has often argued that the call for OA mandates is a sign that researchers oppose OA and must be coerced. This argument always flew in the face of the evidence, but the unanimous Harvard vote should be the last nail in the coffin in which we bury the idea. For the same reason, the Harvard vote decisively confirms Alma Swan's finding that the overwhelming majority of researchers do not resent OA mandates and would willingly comply with one from their funder or university.

But even OA supporters can learn something from this vote. We knew that faculty didn't oppose OA or need to be coerced. We knew that sluggish faculty support for OA initiatives was due more to lack of familiarity than informed opposition. While an administrative mandate is an effective shortcut to a good outcome, and one with which most researchers would willingly comply, Stuart Shieber and the Harvard Faculty of Arts and Sciences have shown that there's another route to the same goal. It may take longer, but it directly addresses the problems of faculty unfamiliarity and misunderstanding, and brings informed consent with it.

I've said many times that I only support OA mandates that are conditions on voluntary contracts. And while I believe that even administrative mandates fit this description, the voluntariness of the Harvard policy is conspicuous and unparalleled.

(For the fullest version of my argument that mandates should be conditions on voluntary contracts, and that administrative mandates at funding agencies and universities qualify, see my July 2006 article on mandating OA for electronic theses and dissertations.)

The policy requires two things of FAS faculty: (1) that they give Harvard non-exclusive permission “to exercise any and all rights under copyright” over their scholarly articles, which includes permission to disseminate OA copies through the institutional repository, and (2) that they send digital copies of their articles to the Provost's Office.

I call these requirements, but the resolution approved by the faculty avoids words like “requirement,” “mandate,” and “must.” Instead, it says that “[e]ach Faculty member grants … permission … ,” “[t]he policy will apply to all scholarly articles written while … ,” and “each Faculty member will provide. …” These are direct statements of what is and will be the case, much as statutes describe legal obligations with the word “shall.” If there's any give in the policy, and there is, it lies in the fact that the policy allows opt-outs.

Faculty may opt out of both requirements. From the resolution: “The Dean or the Dean's designate will waive application of the policy for a particular article upon written request by a Faculty member explaining the need.” […]

Will it be easy or hard to get a waiver? In one sense, it will be very easy. Stuart Shieber elaborated for Nature News: “If the author requests a waiver, the dean will provide a waiver.”

But the request and explanation must be in writing. Moreover, faculty members may not ask for one waiver to cover all their articles, but must submit separate requests for separate articles.

This mix of ease and difficulty is clearly deliberate. On the one hand, FAS faculty wanted the freedom to make exceptions. Either they could foresee cases in which they would want exceptions for themselves or they foresaw the vote without that freedom built in. On the other hand, they wanted to give themselves an additional incentive to push back against the publishers who will ask faculty to ask for waivers.

Does the existence of an automatic opt-out, even with a small administrative hurdle, vitiate the policy? Not at all. Think of a classroom in which teachers require students to sign out before leaving the room. The “waiver” is automatic and students use it. But it's the exception and most students are in their seats most of the time. The policy sends the signal about what is expected, and the expectation alone, perhaps with a small administrative hurdle, makes the scene very different from one with no policy at all.

Is a policy with an automatic opt-out on request still a mandate? This is an unfruitful question which devolves quickly into a verbal dispute. A better question is whether opt-outs will be rare or common.

The opt-out will certainly reduce compliance from 100% to something less. But if opt-outs are rare and the compliance only drops to 95%, give or take, then the policy will still be a huge success. Moreover, even if the initial gap is large, it will very likely shrink over time. For example, if some faculty are inclined to opt out today because they believe that OA archiving will bypass peer review or violate copyright, then they'll drop that objection when they understand the facts. If they are inclined to opt out because they want to publish in a certain journal, which asks them to opt out, then over time they may be more willing to ask the journal to accommodate Harvard, rather than the other way around, and over time the journal itself may be willing to do so.

In December 2006, the Australian Research Council (ARC) adopted a policy requesting—not requiring—its grantees to self-archive, but requiring non-complying grantees to justify their non-compliance. You could call that a mandate or not. But as I pointed out at the time, it's the functional equivalent of a mandate. “It creates a strategic consideration that is not a sanction but more consequential than anything to be found in some of the policies that use mandatory language.”

The Harvard policy is stronger than the ARC policy for going beyond the language of mere requests, but weaker for granting exemptions automatically. And on top of this mix of strengths and weaknesses, the Harvard policy uses a strategy similar to the ARC's for shrinking the exemption loophole and creating the functional equivalent of a mandate. Nevertheless, I'm the first to admit that English has a limited vocabulary for the shades and hues of regulatory strength. If we had a better word than “mandate” for this particular shade, I would use it.

(For more on the regrettable and sometimes misleading connotations of the word “mandate” in this context, see this dialog between Jan Velterop and me a year ago this month.)

Pat Schroeder, President of the Association of American Publishers (AAP), told Science Magazine that Harvard's willingness to grant opt-outs means that the policy is not a mandate. OK. But even Pat Schroeder knows that shifting the default and requiring dissenters to opt out can be a game-changer. Otherwise she wouldn't object to Google's opt-out Library Project or make her organization the lead plaintiff in a lawsuit to stop it.

Even at schools with more mandatory mandates, or no opt-outs, successful implementation depends (as I've often argued) on expectations, education, assistance, and incentives, not coercion. For example, I don't know of any university that exacts a punishment for failure to comply with an OA “mandate.” Universities with mandates achieve their objectives, as Harvard plans to, by communicating a firm expectation, discussing the benefits, offering assistance, and creating incentives, even while they allow exceptions and avoid sanctions.

The Harvard policy shifts the default from non-archiving to archiving, and shifts the burden from OA cooperators to OA dissenters. That's much more than a mere request or encouragement. Before the Harvard vote, the default for faculty was non-archiving and the problem was to persuade them to do something they were not already doing (even if it takes very little time and brings a documented boost in citation impact). The new policy makes archiving the default, mediated by the Provost's Office, and faculty who want to do anything else must bear the burden of explaining their desire to the Dean.

It's not a heavy burden, but then neither is OA archiving. As we well know from long experience on the other side of this line, even a light burden can change behavior on a large scale. This is even more likely when we understand that the new default amplifies faculty research impact, rather than curtailing it, and that the Harvard policy nudges faculty toward their interests rather than away from them.

Daniel Pollock understood the policy well in Outsell Insights: “The faculty is using researchers’ inertia to its advantage. …”

Before anyone puts too cynical an interpretation on this shrewd strategy, remember that it has unanimous faculty support. Harvard faculty now understand that faculty inertia works against self-archiving and have agreed to shift the default and let inertia work against publisher exclusivity instead.

Was the opt-out provision necessary to pass the resolution? Apparently yes. As Shieber told Robin Peek, “There was legitimate concern that there could be particular cases in which the license granted by the motion could work against the interest of a particular faculty member. The provision was certainly important in assuaging some faculty members’ worries that they could be held hostage by the policy in cases where it wasn't serving their best interests.”

When Harvard authors refuse to request waivers, how many journals will refuse to publish their work? Let's tighten the question even further. Journals will know from the submission cover letter that an author is affiliated with Harvard, but they will not know whether the author is requesting an opt-out from Harvard until after the peer review process is over, the article has been accepted, and the journal asks the author to sign the standard copyright transfer agreement. How many journals will refuse to publish works by Harvard authors when those works have already been approved by their peer review process?

The number must be very small today and will shrink steadily over time as journals consider their options and other universities adopt similar policies. Journals don't want to throw away the sunk costs of completed peer review; they don't want to turn away work they've already decided is good; and they don't want to become known as journals that reject good work because of the author's institutional affiliation.

But in the short term, journals can demand that their Harvard authors request waivers and many Harvard authors may comply with these requests. I won't regard that as a failure of the policy. To me the question is what happens over time, not what happens in the early stages.

If it's true that few journals will refuse to publish peer-reviewed work by Harvard authors, or that the number will decline over time, or both, then Harvard faculty have much to gain by refusing to request waivers. This remains the case even if it's also true that a waiver will sometimes be the only way to be published in a certain prestigious journal. The question is not how many journals want faculty to request waivers, but how many will refuse to publish an approved work in the absence of a waiver.

Because faculty have their own reasons for refusing to request waivers (the request process is a nuisance and OA will enlarge their audience and impact), and because journals have their own reasons for publishing already-accepted work by Harvard authors, Robert Darnton was right to say that faculty “will have the collective weight of Harvard behind them if they resist a journal's demand for exclusive rights.” Harvard's collective weight will help faculty get published when they refuse to opt out, and (for that reason) it will also help them refuse to opt out.

Most schools don't have Harvard's weight, of course. But as more universities adopt similar policies, they will increase their collective weight and increase the likelihood that publishers will have to accommodate them. In that sense, the Harvard policy, and the earlier university-level OA mandates, are standing invitations to other universities to gain strength through common purpose and create a critical mass that will change journal policies for all authors. Think of the current university mandates as the first wave of a consumer movement in a journal marketplace distorted by monopolists.

Universities trying to estimate the benefits of an OA mandate should take into account that the benefits will be amplified by synergy with other institutions. Universities that wanted an OA mandate but didn't want to be at the front of the pack, fearing that publishers could refuse to publish work by their faculty, will now find it safer to act. Harvard has increased the bargaining weight of authors demanding the rights to self-archive, and every institution that now adopts an OA mandate will increase it even further, for themselves and for those still to follow.

Under the policy, Harvard faculty will still own the copyrights to their research articles. Harvard is not acquiring ownership, just a non-exclusive license. However, once faculty grant that non-exclusive license to the university, they will not be able to transfer the usual package of exclusive rights to a publisher. Under the new policy, then, faculty still own their own work “subject to” the Harvard license (as the Harvard press release puts it), and may still transfer their rights to publishers, subject to the same condition.

This has two important legal consequences when faculty do not request opt-outs. First, Harvard will have an express license from the copyright holder to host and disseminate OA copies of these works. Second, publishers will never acquire the rights which would allow them to forbid OA archiving at Harvard or to claim that it infringes their copyright.

This elegantly solves every copyright-based objection to OA archiving. The strategy was pioneered among funding agencies by the Wellcome Trust and among universities by the University of California (see the Postscript below). The idea is for authors to reserve the right to self-archive, even if they transfer everything else to a publisher. The less elegant and less effective alternative adopted at some funding agencies and universities is to require OA archiving except when publisher policies don't allow it, thereby giving the opt-out to publishers rather than authors.

If publishers dislike the Harvard policy, they must either ask the Harvard author to request a waiver, and hope the author will agree, or refuse to publish the work. It won't be enough to adopt an in-house policy against OA archiving, as it would have been in the old days when authors gave publishers the full package of exclusive rights.

This may be obvious, but let me elaborate for one more second. Publishers have two very different legal grounds to deny author preferences for OA: (1) the rights transferred to them by the author and (2) their own background right to refuse to publish any work for any reason. The first is based on copyright—the public statute and the private contract transferring rights from author to publisher. The second is utterly independent of copyright. When Harvard authors follow the new OA default, rather than request an opt-out, publishers have no copyright-based grounds on which to protest or prohibit the resulting OA. All they have left is the right to refuse to publish. There's no doubt that this power counts for much, especially at high-prestige journals. But there's also no doubt that authors have their own power in this game, from their ability to submit their work to other journals and from the growing number of authors who will demand the right to self-archive. At some journals, publishers will have the upper hand, but at others authors will have the upper hand. For Harvard authors who follow the new OA default, it's a power game now, not a rights game. And the balance of power just tipped toward authors because Harvard decided to throw its weight on that side. It will tip further toward authors every time another university adopts a similar policy.

The dozen pre-Harvard university mandates require faculty to deposit their eprints in the institution's OA repository. By contrast, Harvard requires faculty to give permission for OA archiving but not to make the deposits itself. That's why I call it a permission mandate rather than a deposit mandate.

Instead of making the actual deposits, Harvard faculty must send copies of their work to the Provost's Office, which will then deposit them. This approach raises at least three questions.

First, what will the Provost's Office actually do with the eprints? The resolution says only that “[t]he Provost's Office may make the article available to the public in an open-access repository” (emphasis added). I wish it had said “will” rather than “may” here. Will the Provost's Office be selective, depositing some but not others?

Second, how quickly will the Provost's Office make the deposits? If it works quickly, it could provide OA at or before the moment the article is published. If it works slowly, it will miss a beautiful opportunity to use the permissions it already has in hand. If it delays 6–12 months, then it may as well defer to publisher embargoes.

Third, if the faculty was willing to self-impose the expectation that they should send their eprints to the Provost's Office, why not self-impose the expectation that they should deposit them directly in the IR?

I don't know. But I suspect the answer is that the faculty wanted to lower the barrier to compliance, or reduce the burden on faculty, and understood that it's easier to email an eprint than to deposit one in a repository, even if the difference is small. For example, St. Andrews University takes this difference into account and boosts the deposit rate in its IR by asking faculty email their eprints to a librarian, who then makes the deposits.

Les Carr and Stevan Harnad have shown that repository deposits are less time-consuming than faculty fear, averaging about 10 minutes per paper.

Similarly, Alma Swan and Sheridan Brown have shown that the process is simpler than faculty fear, and that even the few who face difficulties the first time face fewer difficulties the second time.

However, if faculty fear that self-archiving is difficult and time-consuming, even if their fears are groundless, then they aren't likely to vote to mandate self-archiving. If this is why the Harvard resolution didn't include a direct deposit requirement (and it may not be), then it's possible that further faculty education will reduce the anxiety level and allow a revision that ensures direct deposits in the IR. But it's also possible that no revision is needed. It's possible that the slightly reduced burden for authors will actually boost compliance, and it's possible that the Provost's Office will make the deposits at least as quickly and surely as faculty would under a direct deposit mandate.

Here's how I described the advantages of a permission mandate in SOAN for December 2007:

This model reduces the demands on faculty and increases the certainty about permissions. As long as the university is willing to pay people … to make the actual deposits, it could be a faster and more frictionless way to move the deposit rate toward 100%.

The two key variables in implementing the Harvard policy are the rate of opt-outs and the speed of deposits. If the opt-out rate is low and the deposit speed is high, then it's hard to imagine a better university-level OA policy. But if the opt-out rate is high, or the speed is low, or both, then the policy won't meet its own objectives and Harvard faculty should consider revisions to fine-tune it.

Stevan Harnad is already calling for revisions. In particular, he'd like to see a deposit requirement (without an opt-out) alongside the existing permission requirement (with an opt-out). The particular deposit policy he'd like to see is what he calls immediate deposit / optional access (ID/OA), or what I call the dual deposit/release strategy: all articles are deposited immediately upon publication, but only the metadata are made OA at the same time; the texts become OA only when the university acquires permission or publisher embargo runs. Stevan and I both support the dual deposit/release strategy. But he argues that it is needed now to strengthen the Harvard policy, and I argue that it is not needed now and may never be needed. See our two blog posts encapsulating this debate,

My argument in a nutshell is that the policy won't need revision if the rate of opt-outs is low and the speed of deposits is high. Harvard has adopted the world's first permission mandate and it deserves time to work out implementation procedures that will deliver on its own objectives. If the implementation doesn't deliver, then I'm confident that Harvard insiders will be at least as concerned as outsiders like me and will already be thinking about fine-tuning.

Harvard's permission mandate will assure that Harvard has the right to make every faculty eprint OA (except when faculty request opt-outs). Is this unnecessary in light of the fact that about two-thirds of surveyed TA journals already allow OA archiving?

No; it's still necessary and desirable. At most green journals, the Harvard [license] will be accepted as a matter of course. But it will clearly be necessary for the ungreen third of TA journals. It will even be necessary for some of the green two-thirds, as insurance against later changes to the journal's access policy. It will also be necessary for greenish journals that aren't green enough, for example, because they prohibit deposit in certain repositories, impose fees or embargoes on self-archiving, or limit re-use rights.

If you had to drive across a friendly border every day to get to work, and knew that two-thirds of the time you would be waved through, it would still make good sense to carry your passport. Likewise, Harvard's permission mandate makes good sense precisely because Harvard wants OA for all of its research output, not just for the fraction for which publishers are already granting permission.


I've already seen a rash of misunderstandings in the press and blogosphere about the Harvard policy. Here are the seven most common:

  • “This is the first OA mandate at a university.” Nope.

  • “Harvard faculty may only submit articles to green journals (those that permit OA archiving).” Harvard faculty may submit to any journal. When they submit to a journal that does not ordinarily permit OA archiving, then either the author will ask it to do so in this case (through an author addendum) or the publisher will ask the author to request a waiver from Harvard. No journals will be off-limits to Harvard authors unless the journals themselves refuse to publish work by Harvard authors. This misunderstanding is closely related to the next:

  • “The Harvard policy conflicts with policies at journals that do not allow OA archiving.” First, either Harvard authors will request opt-outs or they won't. If they do, there's no conflict. If they don't, then either a journal will accept a particular paper (with the author's request to retain key rights) or it won't. If it does, then there's no conflict. If it doesn't, then the author will look for another publisher. Either the author and journal end up agreeing on terms, or the Harvard author is in the same position as a non-Harvard author with a rejected manuscript.

  • “Harvard will own faculty writings.” Faculty will continue to own the rights to their own work, until or unless they transfer those rights to a publisher. Harvard is a getting non-exclusive license, not ownership.

  • “Harvard faculty will have to pay a fee to comply with this policy (either to publishers or to Harvard).” This is deeply confused. First, it falsely assumes that the Harvard policy is about publication in OA journals (gold OA), when it's really about distribution through an OA repository (green OA). Second, it falsely assumes that all or most OA journals charge publication fees, when only a minority do so.

  • “The purpose of the policy is to bypass peer review.” Half the sources making this mistake then express horror and half say “good riddance.” But both the horror and joy are misplaced. We may not know all the versions to which the policy applies, but it clearly applies to the “final version of the article” and therefore to some version of the peer-reviewed postprint. Hence, the purpose of the policy is not to bypass peer review, but to provide OA to peer-reviewed articles.

  • “The policy will undermine peer review.” If this one is based on a sloppy reading of the policy, concluding that its purpose is to bypass peer review, then it collapses into the previous misunderstanding. But otherwise it's based on a hasty prediction of doom for subscription-based journals, heedless of the counter-evidence, and a careless conflation of peer review with peer review at subscription-based journals. For a full answer to the objection that OA will undermine peer review, see my article in SOAN for September 2007.

Harvard's example will spread. Some universities will feel a competitive urge: “This policy will help Harvard and we have to keep up.” Some will feel a cooperative urge: “The more universities adopt OA mandates, the more we will accelerate research and guarantee that publishers will adapt.” Some will feel both.


At Open Access News, Gavin Baker and I put together five collections of comments on the Harvard policy:

February 13, 2008

February 14, 2008

February 15, 2008

February 17, 2008

February 22, 2008


While Harvard is the first university to adopt a permission mandate, the University of California was the first to draft one.

California approved its first draft policy in December 2005 and distributed it in January 2006. It approved a revised draft in January 2007 and sent it to the 10 UC campuses for review in May 2007.

The California draft policy says that faculty “shall routinely grant” to the university a non-exclusive license to provide OA to their works through the institutional repository. When faculty sign copyright transfer agreements with publishers, they “must retain” the right to comply with the university policy. The draft policy also allows faculty to opt out, one work at a time.

According to the Chronicle of Higher Education for February 21, 2008, “Comments on [the California] draft last year reflected ‘almost universal support for the concept,’ says Gary S. Lawrence, director of systemwide library planning, ‘but a great deal of concern about the implementation details.’ Harvard's success in creating an arrangement that faculty members agreed on, he says, ‘provides us a lot of encouragement.’ ”

University of California OA policy home page (updated as new steps are taken)

The latest draft of the policy, January 29, 2007

Policy FAQ, February 2007

Faculty Senate review of the draft policy, July 9, 2007

Administrative and other non-Senate reviews of draft policy, July 26, 2007

A study showing little faculty familiarity with OA or with the California draft policy: Faculty Attitudes and Behaviors Regarding Scholarly Communication: Survey Findings from the University of California, August 2007

This article focuses on the OA policy at Harvard's Faculty of Arts and Sciences, the university's first. But today all Harvard schools have OA policies: Faculty of Arts and Sciences (February 2008), Law School (May 2008), Kennedy School of Government (March 2009), Graduate School of Education, Business School (June 2009), Divinity School (November 2010), Graduate School of Design (March 2011), School of Public Health (November 2012), and the Medical School (June 2014).

Subsequent policies refined the FAS policy in three major respects. First, the waiver does not require a written justification, just a check in a web form. Second, the waiver only applies to the license, or permission for OA, not to the deposit requirement. Third, the policy introduced an embargo option alongside the waiver option.

I've been a Faculty Fellow at Harvard's Berkman Center for Internet & Society since June 2009, and Director of the Harvard Office for Scholarly Communication since July 2013. But I wrote this piece before I had a Harvard connection and before any Harvard connection was on the horizon.

No comments here
Why not start the discussion?