Authors control the volume and growth of OA. They decide whether to submit their work to OA journals (gold OA), whether to deposit their work in OA repositories (green OA), and how to use their copyrights. But scholarly authors are still largely unfamiliar with their OA options. It’s pointless to appeal to them as a bloc because they don’t act as a bloc. It’s not hard to persuade or even excite them once we catch their attention, but because they are so anarchical, overworked, and preoccupied, it’s hard to catch their attention.
Fortunately, funding agencies and universities are discovering their own interests in fostering OA. These nonprofit institutions make it their mission to advance research and to make that research as useful and widely available as possible. Their money frees researchers to do their work and avoid the need to tie their income to the popularity of their ideas. Above all, these institutions are in an unparalleled position to influence author decisions.
Today, more than fifty funding agencies and more than one hundred universities have adopted strong OA policies. Each one depends on the primacy of author decisions.2
One kind of policy, better than nothing, requests or encourages OA. A stronger kind of policy requires OA or makes it the default for new work. These stronger policies are usually called OA mandates and I’ll use that term for lack of a better one (but see section 4.2 on how it’s misleading).
Request or encouragement policies These merely ask faculty to make their work OA, or recommend OA for their new work. Sometimes they’re called resolutions or pledges rather than policies.3
Encouragement policies can target green and gold OA equally. By contrast, mandates only make sense for green OA, at least today when OA journals constitute only about one-quarter of peer-reviewed journals. A gold OA mandate would put most peer-reviewed journals off-limits and seriously limit faculty freedom to submit their work to the journals of their choice. This problem doesn’t arise for green OA mandates.
Fortunately, this is well understood. There are no gold OA mandates anywhere; all OA mandates are green. Unfortunately, however, many people mistakenly believe that all OA is gold OA and therefore mistake proposed green OA mandates for proposed gold OA mandates and raise objections that would only apply to gold OA mandates. But as more academics understand the green/gold distinction, and understand that well-written green OA mandates are compatible with academic freedom, more institutions are adopting green OA mandates, almost always at the initiative of faculty themselves.4
At universities, there are roughly three approaches to green OA mandates:
Loophole mandates These require green OA except when the author’s publisher doesn’t allow it.5
Deposit mandates These require deposit in an OA repository as soon as the article is accepted for publication, but they separate the timing of deposit from the timing of OA. If the author’s publisher doesn’t allow OA, then these policies keep the deposited article dark or non-OA. If the publisher allows OA, immediately or after some embargo, then the deposit becomes OA as soon as the permission kicks in. Because most publishers allow OA on some timetable, this method will provide OA to most new work in due time.
Deposit mandates generally depend on publisher permission for OA, just like loophole mandates. The difference is that they require deposit even when they can’t obtain permission for OA.6
Rights-retention mandates These require deposit in an OA repository as soon as the article is accepted for publication, just like deposit mandates. But they add a method to secure permission for making the deposit OA. There’s more than one way to secure that permission. At the Wellcome Trust and NIH, which pioneered this approach for funding agencies, when grantees publish articles based on their funded research they must retain the nonexclusive right to authorize OA through a repository. At Harvard, which pioneered this approach for universities, faculty members vote to give the university a standing nonexclusive right (among other nonexclusive rights) to make their future work OA through the institutional repository. When faculty publish articles after that, the university already has the needed permission, and faculty needn’t take any special steps to retain rights or negotiate with publishers. Nor need they wait for the publisher’s embargo to run. Harvard-style policies also give faculty a waiver option, allowing them to opt out of the grant of permission to the university, though not out of the deposit requirement. When faculty members obtain waivers for given works, then Harvard-style mandates operate like deposit mandates and the works remain dark deposits until the institution has permission to make them OA.7
Many OA policies are crossbreeds rather than pure types, but all the policies I’ve seen are variations on these four themes.
First note that none of the three “mandates” absolutely requires OA. Loophole mandates allow some work to escape through the loophole. Deposit mandates allow some deposited work to remain dark (non-OA), by following publisher preferences. Rights-retention mandates with waiver options allow some work to remain dark, by following author preferences.
Loophole and deposit policies defer to publishers for permissions, while rights-retention policies obtain permission from authors before they transfer rights to publishers. For loophole and deposit policies, permission is contingent, because some publishers are willing and some are not. For rights-retention policies, permission is assured, at least initially or by default, although authors may opt out for any publication.
When loophole policies can’t provide OA, covered works needn’t make it to the repository even as dark deposits. When deposit and rights-retention policies can’t provide OA, at least they require dark deposit for the texts, and OA for the metadata (information about author, title, date, and so on). Releasing the metadata makes even a dark deposit visible to readers and search engines. Moreover, many repositories support an email-request button for works on dark deposit. The button enables a reader to submit a one-click request for a full-text email copy and enables the author to grant or deny the request with a one-click response.8
We could say that rights-retention policies require OA except when authors opt out, or that they simply shift the default to OA. Those are two ways of saying the same thing because, either way, faculty remain free to decide for or against OA for each of their publications. Preserving this freedom and making it conspicuous help muster faculty support, indeed, unanimous faculty votes. Because shifting the default is enough to change behavior on a large scale, waiver options don’t significantly reduce the volume of OA. At Harvard the waiver rate is less than 5 percent, and at MIT it’s less than 2 percent.
Loophole policies and rights-retention policies both offer opt-outs. But loophole policies give the opt-out to publishers and rights-retention policies give it to authors. The difference is significant because many more authors than publishers want OA for research articles.
Many institutions adopt loophole policies because they believe a blanket exemption for dissenting publishers is the only way to avoid copyright problems. But that is not true. Deposit policies don’t make works OA until publishers allow OA, and rights-retention policies close the loophole and obtain permission directly from authors at a time when authors are the copyright holders.
OA policies from funding agencies are very much like OA policies from universities. They can encourage green and gold OA, or they can require green OA. If they require green OA, they can do so in one of the three ways above. If there’s a difference, it’s that when funders adopt a rights-retention mandate, they typically don’t offer waiver options. On the contrary, the Wellcome Trust and NIH require their grantees to make their work OA through a certain OA repository on a certain timetable and to retain the right to authorize that OA. If a given publisher will not allow grantees to comply with their prior funding agreement, then grantees must look for another publisher.9
There are two reasons why these strong funder policies don’t infringe faculty freedom to submit work to their journals of their choice. First, researchers needn’t seek funds from these funders. When they choose to do so, then they agree to the OA provisions, just as they agree to the other terms and conditions of the grant. The OA “mandate” is a condition on a voluntary contract, not an unconditional requirement. It’s a reasonable condition as well, since public funders, like the NIH, disburse public money in the public interest, and private funders, like the Wellcome Trust, disburse charitable money for charitable purposes. To my knowledge, no researchers have refused to apply for Wellcome or NIH funds because of the OA condition, even when they plan to publish in OA-averse journals. The OA condition benefits authors and has not been a deal-breaker.
Second, virtually all publishers accommodate these policies. For example, no surveyed publishers anywhere refuse to publish work by NIH-funded authors on account of the agency’s OA mandate. Hence, in practice grantees may still submit work to the journals of their choice, even without a waiver option to accommodate holdout publishers.10
We should never forget that most toll-access journals already allow green OA and that a growing number of high-quality, high-prestige peer-reviewed journal are gold OA. From one point of view, we don’t need OA mandates when authors already plan to publish in one of those journals. But sometimes toll-access journals change their positions on green OA. Sometimes authors don’t get around to making their work green OA even when their journals allow it. And sometimes authors don’t publish in one of those journals. The final rationale for green OA mandates, then, is for institutions to bring about OA for their entire research output, regardless of how publishers might alter their policies, regardless of author inertia, and regardless of the journals in which faculty or grantees choose to publish.
Green OA mandates don’t assure OA to the entire research output of a university or funding agency, for the same reason that they don’t require OA without qualification. But implementing them provides OA to a much larger percentage of the research output than was already headed toward OA journals or OA repositories, and does so while leaving authors free to submit their work to the journals of their choice.
I’ve only tried to give a rough taxonomy of OA policies and their supporting arguments. For detailed recommendations on OA policy provisions, and specific arguments for them, see my 2009 analysis of policy options for funding agencies and universities.11
I’ve also focused here on OA policies for peer-reviewed research articles. Many universities have adopted OA mandates for theses and dissertations, and many funder OA policies also cover datasets. A growing number of universities supplement OA mandates for articles with a sensible and effective policy to assure compliance: When faculty come up for promotion or tenure, the review committee will only consider journal articles on deposit in the institutional repository.12
The strongest OA policies use words like “must” or “shall” and require or seem to require OA. They’re commonly called OA “mandates.” But all three varieties of university “mandate” above show why the term is misleading. Loophole mandates don’t require OA without qualification: when publishers dissent, articles are either not deposited in the repository or not made OA. Deposit mandates don’t require OA without qualification: when publishers dissent, articles are deposited in a repository but are not made OA. Rights-retention mandates with waiver options don’t require OA without qualification: authors may obtain waivers and sometimes do. I haven’t seen a university OA “mandate” anywhere without at least one of these three kinds of flexibility.
That’s the main reason why no university policies require OA without qualification. There are a few more. First, as Harvard’s Stuart Shieber frequently argues, even the strongest university policies can’t make tenured faculty comply.14 Second, as I’ve frequently argued, successful policies are implemented through expectations, education, incentives, and assistance, not coercion. Third, even the strongest policies—even the no-loophole, no-deference, no-waiver policies at the Wellcome Trust and NIH—make OA a condition on a voluntary contract. No policy anywhere pretends to impose an unconditional OA requirement, and it’s hard to imagine how any policy could even try. (“You must make your work OA even if you don’t work for us or use our funds”?)
Unfortunately, we don’t have a good vocabulary for policies that use mandatory language while deferring to third-person dissents or offering first-person opt-outs. Nor do we have a good vocabulary for policies that use mandatory language and replace enforcement with compliance-building through expectations, education, incentives, and assistance. The word “mandate” is not a very good fit for policies like this, but neither is any other English word.
By contrast, we do have a good word for policies that use mandatory language for those who agree to be bound. We call them “contracts.” While “contract” is short, accurate, and unfrightening, it puts the accent on the author’s consent to be bound. That’s often illuminating, but just as often we want to put the accent on the content’s destiny to become OA. For that purpose, “mandate” has become the term of art, for better or worse.15
I use “mandate” with reluctance because it can frighten some of the people I’m trying to persuade and can give rise to misunderstandings about the policies behind the label. When we have time and space for longer phrases, we can talk about “putting an OA condition” on research grants, in the case of NIH-style policies, or “shifting the default to OA” for faculty research, in the case of Harvard-style policies. These longer expressions are more accurate and less frightening. However, sometimes we need a shorthand term, and we need a term that draws an appropriately sharp contrast with policies that merely request or encourage OA.
If anyone objects that a policy containing mandatory language and a waiver option isn’t really a “mandate,” I won’t disagree. On the contrary, I applaud them for recognizing a nuance which too many others overlook. (It’s depressing how many PhDs can read a policy with mandatory language and a waiver option, notice the mandatory language, overlook the waiver option, and then cite the lack of flexibility as an objection.) But denying that a policy is a mandate can create its own kinds of misunderstanding. In the United States, citizens called for jury duty must appear, even if many can claim exemptions and go home again. We can say that jury duty with exemptions isn’t really a “duty,” provided we don’t conclude that it’s merely a request and encouragement.
Finally, a common misunderstanding deliberately promulgated by some publishers is that OA must be “mandated” because faculty don’t want it. This position gets understandable but regrettable mileage from the word “mandate.” It also overlooks decisive counter-evidence that we’ve had in hand since 2004. Alma Swan’s empirical studies of researcher attitudes show that an overwhelming majority of researchers would “willingly” comply with a mandatory OA policy from their funder or employer.16
The most recent evidence of faculty willingness is the stunning series of strong OA policies adopted by unanimous faculty votes. (When is the last time you heard of a unanimous faculty vote for anything, let alone anything of importance?) As recently as 2007, speculation that we’d soon see more than two dozen unanimous faculty votes for OA policies would have been dismissed as wishful thinking. But now that the evidence lies before us, what looks like wishful thinking is the publishing lobby’s idea that OA must be mandated because faculty don’t want it.17
Finally, the fact that faculty vote unanimously for strong OA policies is a good reason to keep looking for a better word than “mandate.” At least it’s a good reason to look past the colloquial implications of the term to the policies themselves and the players who drafted and adopted them. Since 2008, most OA “mandates” at universities have been self-imposed by faculty.
Some kinds of strong OA policy that are politically unattainable or unwise today may become attainable and wise in the future. Here are three examples.
Today, a libre green mandate (say, one giving users the right to copy and redistribute, not just access for reading) would face serious publisher resistance. Even if the policy included rights retention and didn’t depend on publishers for permissions, publisher resistance would still matter because publishers possess—and ought to possess—the right to refuse to publish any work for any reason. They could refuse to publish authors bound by a libre green policy, or they could insist on a waiver from the policy as a condition of publication. Policies triggering rejections hurt authors, and policies driving up waiver rates don’t do much to help OA. However, publisher resistance might diminish as the ratio of OA publishers to toll-access publishers tilts toward OA, as spontaneous author submissions shift toward OA journals, or as the number of institutions with libre green mandates makes resistance more costly than accommodation for publishers. When OA policies are toothless, few in number, or concentrated in small institutions, then they must accommodate publishers in order to avoid triggering rejections and hurting authors. But as policies grow in number, scope, and strength, the situation could flip over, and publishers will have to accommodate OA policies in order to avoid hurting themselves by rejecting too many good authors for reasons unrelated to the quality of their work.19
Today, a gold OA mandate would limit faculty freedom to submit work to the journals of their choice. But that’s because today only about 25 percent of peer-reviewed journals are OA. As this percentage grows, then a gold OA mandate’s encroachment on academic freedom shrinks. At some point even the most zealous defenders of faculty freedom may decide that the encroachment is negligible. In principle the encroachment could be zero, though of course when the encroachment is zero, and gold OA mandates are harmless, then gold OA mandates would also be unnecessary.
Today, faculty voting for a rights-retention OA mandate want a waiver option, and when the option is available their votes tend to be overwhelming or unanimous. But there are several circumstances that might make it attractive for faculty to abolish waiver options or make waivers harder to obtain. One is a shift in faculty perspective that makes access to research more urgent than indulging publishers who erect access barriers. Another is a significant rise in publisher acceptance of green OA, which gives virtually all authors—rather than just most—blanket permission for green OA. In the first case, faculty might “vote with their submissions” and steer clear of publishers who don’t allow author-initiated green OA. In the second case, faculty would virtually never encounter such publishers. In the first case, they’d seldom want waivers, and the second they’d seldom need waivers.
It’s understandable that green gratis mandates are spreading faster than green libre mandates, that green mandates in general are spreading faster than gold mandates, and that rights-retention policies with waiver options are spreading faster than rights-retention policies without waivers. However, there is modest growth on one of these fronts: green libre mandates.20
The case against these three kinds of OA policy is time-sensitive, not permanent. It’s circumstantial, and circumstances are changing. But the strategy for institutions wanting to remove access barriers to research is unchanging: they should adopt the strongest policies they can today and watch for the moment when they could strengthen them.
As researchers become more familiar with OA, as more institutions adopt OA policies, as more new literature is covered by strong OA policies, as more toll-access journals convert to OA, as more toll-access journals accommodate OA mandates without converting, and even as more OA journals shift from gratis to libre, institutions will be able strengthen their OA policies without increasing publisher-controlled rejection rates or author-controlled waiver rates. They should watch the shifting balance of power and seize opportunities to strengthen their policies.
The moments of opportunity will not be obvious. They will not be highlighted by objective evidence alone and will call for some self-fulfilling leadership. Institutional policy-makers will have to assess not only the climate created by existing policies, and existing levels of support, but also the likely effects of their own actions. Every strong, new policy increases the likelihood of publisher accommodation, and when enough universities and funders have policies, all publishers will have to accommodate them. In that sense, every strong new policy creates some of the conditions of its own success. Every institution adopting a new policy brings about OA for the research it controls and makes the way easier for other institutions behind it. Like many other policy issues, this is one on which it is easier to follow than to lead, and we already have a growing number of leaders. A critical mass is growing and every policy is an implicit invitation to other institutions to gain strength through common purpose and help accelerate publisher adaptation.