Some personal thoughts on the necessity and optimal length of any embargo periods applied to open-access deposits of articles published in toll-access academic journals. 

Conventional academic publishers have argued that it is necessary to place an embargo period on open-access deposits of authors’ articles in order to protect their income received through journal subscriptions. One can summarise their argument as follows:

1. It is a given that some measure of private return to the publisher is necessary to underpin the public good of research publication.
2. Short embargo periods will cause people to cancel subscriptions.
3. Longer embargo periods are therefore necessary to protect publisher income through subscriptions, and therefore, the public good of research publication.

If we accept that argument, the focus shifts from whether we should have embargo periods to how long they should be. This is desirable territory for conventional publishers, because they know that policymakers can be persuaded that the how long question may only satisfactorily be answered when there is evidence that short embargo periods have caused damage to publishers, and is therefore straightforward to argue for a risk averse policy strategy. However, the publisher reasoning above is flawed, and can be unpicked in its entirety.

Anyone prepared to conduct a thought experiment to reimagine scholarly communication would conclude that  effective research dissemination needn’t be dependent on conventional publishers at all. Indeed, the existence of arXiv.org and the numerous open-access university presses worldwide show that there are already perfectly workable and acceptable alternatives to the conventional publisher model. The current dominance of conventional publishers does not negate the validity of the above thought experiment. One may question, then, what value is added by the publisher.

If policymakers were to reduce the publishing process to its essentials, they would find two valuable contributions that the publishing process makes to the author’s initial submitted draft article. The first of these contributions is peer review. Peer review is a service, managed by publishers, but provided by academics. The service is very valuable, no doubt, but its management has no intrinsic value. The service of academic is transferrable and, though facilitated by the publication process, it is not a direct product of it. Peer review is not unique to conventional publishing, either. So the traditional locus of peer review must be for historic reasons, and our thought experiment above might just as easily locate peer review elsewhere.

The second valuable contribution that the publication process makes to the author’s draft is that it inherits some of the brand value of the journal in which it is published. This brand value serves as an important signal to readers of the quality of the published work, and some of the brand value rubs off on the author, helping her to signal her own academic worth to others.

It would be tempting to conclude that this sort of brand signalling is an intrinsic contribution of the publication process, and the greater relative stickiness of journal brands would support such a conclusion. This understanding is incomplete, though: the article does not simply inherit the brand – it also contributes to it, in the same way that all prior articles published in all journals have contributed to all journal brands. Indeed, a journal’s brand should be more accurately thought of as a by-product of the freely given work of academics: not an additional good, but a secondary one. A journal’s brand is simply the result of an underlying process that mediates and accumulates academic esteem. Such processes are not unique to publishing, and our thought experiment might readily design new ones.

Beyond journal brand and peer review, policymakers should consider the other activities of publishers as nonessential to the research dissemination process (even though some of it may be highly desirable for academics and worth paying for, such as copy-editing, design, or even society activities subsidised by journal income). This isn’t to suggest that policymakers should attempt to do away with publishers, but rather to illustrate that it is straightforward to conceive of their contributions as being less essential and unique than they often attest. This deals with point 1 of the publisher argument above.

Turning to the assertion that short embargoes will lead to subscription cancellations, one uncovers more interesting things. While some conventional publishers are permissive enough to allow authors to upload copies of the final published version of record to the Internet, most conventional publishers restrict this to authors’ accepted manuscripts (AAMs) instead. I am sure that even those with access to the final version of record would agree that, in the vast majority of cases, the AAM contains all of the academically important aspects of the research article. Even if this were debatable, it appears that conventional publishers are on my side of this debate: if the AAM were meaningfully inferior, there would be no need for an embargo. The very existence of embargo periods should lead policymakers to conclude that AAMs, distributed to the Internet via repositories and indexed by search engines, are academically sufficient and therefore optimal.

Such a situation may be unpalatable for conventional publishers, but they can reassure themselves that, even with the AAMs available online with no embargo, readers will still subscribe to journals because:

1. citation practices, particularly in the humanities, depend at the moment on paginated print copy, and these practices require a version of record to be available for citation (though it is straightforward to think up solutions to this: numbered paragraphs, signposting through sub-headings, AAMs annotated with VOR page breaks, direct correspondence with the author(s), and so on);
2. many people want to read the journal’s editorials, book reviews and other peripheral content.

There are three further reasons why cancellations are unlikely, and all relate to the fact that libraries are the principal subscribers to academic journals. Firstly, if free online access to articles takes place at the article level, as with author uploads, then it takes place anarchically. Knowledge about which articles are freely available within a given journal is very therefore difficult for a subscriber to obtain. Libraries do not have the resources to monitor which articles are freely available in any given journal, nor do they have the appetite to make predictions about how this will change in the future. This means subscribers’ decisions to cancel a given journal on the basis of its contents being freely available online are impossible to make until (almost) all of the journal content is virtually guaranteed to be available for free. However, even at that moment, points 1 and 2 above would still apply.

Secondly, librarians say that embargo lengths are a relatively unimportant factor in whether they decide to cancel a journal. Crucially, one of the major factors is journal price. Libraries will only cancel journals if they can’t afford the journal, and the following condition did not apply.

Thirdly, most librarians find it difficult to cancel individual titles, as they are tied into ‘big deals’ for subscriptions to many titles. Librarians cancelling their big deal would immediately face higher costs, not lower costs. The irony here is that libraries are therefore more likely to cancel society/minority journals first – it is ironic because of all of the ‘additional’ value provided by the subscription publication process, it is the dependent activities of learned societies that policymakers ought to be most sympathetic to.

We now turn to the final part of the publisher argument at the top of this piece. Here I present one simple truth: policymakers find embargo periods very hard to swallow, because they fundamentally undermine the public good of research publication by putting a price on recent knowledge. If policymakers accept that an optimal strategy is to annul existential threats to journals, it is reasonable to examine the evidence that such threats exists without losing sight of this simple truth. What they will find is this: there is no evidence that short embargo periods put journals at risk, and many journals currently allow embargo periods of less than 24 months, and a substantial proportion of journals allow authors to put their article online immediately. Policymakers should therefore feel well justified in asking why they should not simply set embargo periods to zero.