From Peter Suber's July 2011 Issue of the SPARC Open Access Newsletter.
From the beginning, OA struggled against the widespread assumption that it must violate copyright law. But this has been a struggle against perception, not reality. In fact, steering clear of infringement has always been easier than steering clear of this false assumption and the harm it has caused.
The assumption has made some authors fear OA. It has made some institutions skittish about committing to OA. It has needlessly weakened some OA policies, for example, by creating loopholes for dissenting publishers. It has even been a dishonest pretext for bad legislation.
There are bullet-proof methods for OA publishers and repositories to avoid copyright problems. These methods are better known today than they were five years ago, but we still struggle against the same false assumption, the same fear, the same skittishness, the same needless capitulation, and the same dishonesty. Here's an attempt to clarify the situation in a dozen propositions.
(1) Fair use is not enough to authorize OA. Gratis OA exceeds fair use by distributing a full-text work to a worldwide audience. Libre OA exceeds fair use by permitting users of the work to exceed fair use
Clarifying fair use, pushing the envelope on fair use, and expanding fair use are all desirable. But they're insufficient for OA.
Note that fair use isn't enough to authorize conventional TA publishing either.
(2) The public domain (PD) is more than enough to authorize OA. But it only authorizes OA for works in the PD, and most of the works for which we want OA are under copyright. Hence, most of the time the PD is not enough either.
Protecting the PD, preventing retroactive copyright (piracy from the PD), and expanding the PD are all desirable. But they're insufficient for OA to works under copyright, a category that covers nearly all new research articles.
Note that the PD is not enough to authorize conventional TA publishing of works under copyright either.
(3) If fair use and the public domain aren't enough to authorize OA for full-text works under copyright, then we need permission from the copyright holder.
Notice where we've arrived. OA for full-text works under copyright avoids infringement in exactly the same way that conventional TA publishing avoids infringement. For sufficiently old work, OA and TA can rely on the public domain. For newer work under copyright, they rely on rightsholder consent.
This is the first point on which the widespread assumption goes wrong. It presupposes that conventional TA publishers have easy and natural methods to avoid infringement, and that OA providers have nothing but dubious or convoluted methods. But the two camps use exactly the same methods.
If a journal publishes a new article without the author's permission, it is guilty of infringement, whether the journal is OA or TA. The publishing agreement is the means by which the author, as the original copyright holder, grants permission to the publisher. Rightsholder consent frees the publisher to proceed without liability or fear of liability.
OA journals, like TA journals, use contracts with authors to secure the needed permissions and avoid infringement. OA repositories obtain permission from the rightsholders too, although sometimes that means obtaining it from authors before they transfer rights to a journal, and sometimes it means obtaining it from publishers after they acquire rights from authors.
(4) Authors of new work are the copyright holders until or unless they transfer copyright to someone else, like a publisher. Publishers only hold the rights that authors voluntarily transfer to them.
This is the second point on which the pernicious widespread assumption goes wrong. It presupposes that publishers always hold all rights. A related third error is to presuppose that publishers typically use the rights they hold to oppose OA.
Initially publishers don't hold any rights. Even after authors sign publication agreements, authors may retain some rights and transfer others. And even after publishers acquire rights from authors, many publishers choose to permit OA. In fact, most surveyed TA publishers permit green OA and a growing number of publishers are experimenting with gold OA (more in #8).
(5) Authors are much more likely than conventional publishers to consent to OA. This may not be true for novels or journalism, but it's true for scholarly journal articles. Part of the reason is that scholarly journals do not pay authors. Hence, authors of scholarly journal articles can consent to OA without losing revenue. Another, related part of the explanation is that authors of scholarly journal articles write for impact, not for money, and OA increases their impact by enlarging their audience.
Consequently, the best strategy for obtaining rightsholder consent to OA is (1) to keep the key rights in the hands of authors, or (2) to transfer rights to OA publishers rather than TA publishers, or (3) to obtain permission from authors *before* they transfer rights to TA publishers. The worst strategy is to transfer key rights to a corporation that won't permit OA and then hope to get permission.
(6) For similar reasons, it's easier to obtain copyright holder consent to OA for still-unpublished literature than already-published literature. When we seek consent to OA for future literature, we ask the authors before they transfer any rights to publishers, and authors are more likely to consent than publishers. When we seek consent for past literature, we usually have to ask publishers, not authors.
When we put ##4-6 together, we see the wisdom of rights-retention OA policies. Funders like the Wellcome Trust and NIH require grantees to retain the non-exclusive right to authorize OA to future articles arising from their research grants. Faculty at universities like Harvard and MIT vote to grant their institution the non-exclusive right to make their future work OA, though with an easy opt-out.
Both kinds of policy obtain permission from authors, when authors are still the rightsholders. Both obtain permission for future, still-unpublished works rather than for past, already-published works.
(7) Rights retention works best when authors do not act alone.
Individual authors acting alone may try to retain key rights to their still-unpublished articles, but it's not always easy or successful. If they don't have the aid of funder or university policies to back them up, at least they have the aid of lawyer-drafted author addenda to their publishing agreements. Author addenda help authors frame their requested contract modifications in legally precise and enforceable language.
But author addenda are merely proposed contract modifications which publishers may take or leave, and individual authors have less bargaining power than funders or universities. Publishers reject author addenda more often than they refuse to publish work by authors subject to OA institutional OA mandates.
By contrast, all surveyed publishers accommodate the rights-retention policy at the NIH, even if they pay lobbyists to oppose it and even if they have an unquestioned right (#12) to refuse to publish NIH-funded authors. This is one of the strongest arguments for rights-retention OA mandates at funders and universities.
(8) Most TA publishers (64%) already consent to green gratis OA. But we want OA for all articles, regardless of where authors choose to publish. And we want OA regardless of how publishers may modify their access policies in the future. These are reasons for universities and funders to require rights retention, or at least to make rights retention the new default.
For the journals and publishers granting advanced permission for green gratis OA, see the SHERPA RoMEO database and its statistics page.
In addition to green OA, more and more publishers are experimenting with gold OA. SHERPA currently lists 92 publishers offering hybrid OA options, including all of the largest publishers.
(9) Most OA journals (78%) don't offer libre OA. That is, they publish under all-rights-reserved copyrights and don't allow uses beyond fair use.
When I checked last week (June 24, 2011), 1,448 out of 6,647 journals in the DOAJ, or 21.8%, used some kind of CC license.
As of the same date, 747 or 11.2% had the SPARC Europe Seal of Approval, which requires CC-BY.
OA repositories are rarely in a position to obtain the permissions needed for libre OA. Hence, we can't criticize or complain when most of their deposits are gratis, not libre. But OA journals can easily obtain the permissions needed for libre OA. When they don't offer libre OA, they have no excuse. This is one of the largest missed opportunities of the OA movement to date.
Today most libre OA is gold OA. But unfortunately it's not yet the case that most gold OA is libre OA, and unfortunately it's not even close.
(10) OA does not depend on copyright reform, let alone on the violation or abolition of copyright. However, OA would benefit from reforms of the right kind and many dedicated people are working on them.
For example, here are some copyright reforms that would help the cause:
* Shorten the term of copyright, or at least prevent it from becoming even longer every time Mickey Mouse is about to fall into the public domain.
* Ban the retroactive extension of copyright to works in the public domain.
* Allow OA for orphan works, with a takedown requirement if the rightsholder steps forward and complains.
* Permit the circumvention of DRM in pursuit of non-infringing uses.
* Recognize that some creative works generate revenue for creators, and some don't, and that creators of the former type are harmed by unauthorized copying while creators of the latter type are harmed by the default prohibition of copying. That is, stop making royalty-free literature collateral damage in the war against revenue leaks.
* Allow green OA, at least for royalty-free literature, within a certain time after publication, regardless of the publishing contract the author signed with a publisher.
* Allow digitization and search indexing without permission when they result in no dissemination, or when the dissemination consists of nothing more than fair-use snippets.
* Make the penalties for copyfraud (false claim of copyright) at least as severe as the penalties for infringement; that is, take the wrongful decrease in the circulation of ideas at least as seriously as the wrongful increase in the circulation of ideas.
While OA would benefit from any of these reforms, it doesn't require any of them. OA is compatible with copyright law as it is, despite its grotesque imbalance in favor of publishers. OA needn't wait for copyright reform and hasn't waited. Why? Because rightsholder consent suffices, and rightsholders are consenting in growing numbers.
(11) It follows that the OA movement needn't, and shouldn't, support tactics that depend on deliberate infringement or knowing winks at infringement.
I spelled out the reasons in a blog post from September 2009:
I don't accept that advancing OA through deliberate violations of copyright law would do more good than harm. I have three basic reasons: (1) OA is already lawful and doesn't require the reform or violation of copyright law, even if it could leap forward with the right reforms. (2) OA activists will never match the publishing industry's funds for litigation. (3) One of the most persistent and harmful misunderstandings of OA is that it violates copyright law. We've come a long way in educating policy-makers out of that misunderstanding....A campaign to give the publishing lobby its first valid evidence that OA violates copyright is the last thing we need.
For more detail, see Not Napster for science (October 2003).
(12) Publishers have the right to refuse to publish any work for any reason. This is good and we should all celebrate it. Without this fundamental right, publishing would be a propaganda arm of the state, and all publishing would lose credibility.
However, this fundamental right has nothing to do with copyright. If we had to identify its legal basis, in the US, we could point to the First Amendment and freedom of the press, or the Thirteenth Amendment and abolition of slavery.
It is this background right, not copyright, which allows publishers who dislike the NIH policy to refuse to publish NIH-funded authors. But as noted (#6), all surveyed publishers accommodate the NIH policy.
The relevance of this background right to copyright is strategic, not legal. OA policies that go too far or too fast might encounter this kind of pushback from publishers. For example, a green *libre* mandate (as opposed to a green *gratis* mandate) might simply trigger publisher rejections of covered work, which would hurt authors and not advance OA. However, when large organizations or many smaller organizations adopt strong policies, publishers have little choice but to accommodate them.
When publishers own the key rights, then they can block OA and regard OA without their consent as infringement. They can lobby Congress and argue that an OA mandate without rightsholder consent is "inconsistent with copyright". But when they don't own the key rights, because authors have retained them, then they cannot block OA or regard it as infringement. They can't object that it's "inconsistent with copyright". They can't complain that it violates a right they possess, only that it would violate a right they wish they possessed.
If they oppose a rights-retention policy, like the NIH policy, they can't ground their opposition in copyright law. But if they choose, they can ground their opposition in the background right to refuse to publish any work for any reason. They don't have the easy opt-out of deciding around their conference-room table that they will withhold consent to OA. They only have the hard opt-out of refusing to publish NIH-funded authors. If they do not exercise this fundamental right, then they should admit that they have chosen not to, and they should not allow their lawyers and lobbyists, who should know better, to pretend that the policy is a standing violation of copyright law.