The Embargoes Don’t Work: The British Academy provides the best evidence yet

The debates today on implementing Open Access pivot around two key points. The first is the perceived cost of a transition to a fully funded Open Access publishing environment. The second is the question of delaying access to copies of research outputs made available through repositories – how long should embargoes be? In both cases the question of what data and evidence is reliable has been contentious but information is accumulating.

Given the obsession with embargoes from everyone interested in Open Access it would be easy to think that they logically sit at the core of the debate. Traditional publishers and their lobbyists spend a lot of time and money lobbying for policy statements and legislation that include longer embargoes. And in the interest of full disclosure PLOS and others spend quite a bit of time (although not so much money) advocating for shorter, preferably zero, embargoes. They must surely be important!

But in actual fact they are the residue of a messy compromise. The assumption that lies at the heart of the embargo argument was that if some version of a peer reviewed article is made available through a repository then a subscription publisher needs some period of exclusivity to recover their costs. The evidence that this is the best way to protect publishers while widening access is thin to say the least.

The basic assumption is that having an author manuscript freely available online poses a risk to the subscription business of the publisher. This is a matter of  controversy and I am on the record (and oral comments) as saying that I see substantial evidence that no damage is done and no credible evidence of any damage – a conclusion endorsed by a UK Parliamentary report. But it is the flip side of the argument that is perhaps more important. That is the idea that embargoes help traditional publishers maintain their business while sustainably allowing wider access through repositories.

When the NIH Public Access Policy was implemented the idea of a 12 month embargo on access to the content in Pubmed Central was a compromise. The figure of 12 months wasn’t really evidence based but was seen as ‘safe’. But it was always a messy compromise. For advocates of public access it was a high water mark, a figure to be reduced over time, eventually to zero. For traditional publishers the narrative was built of “well its ok for biomedicine – it’s a fast moving field”. You can draw a straight line from that narrative to the bizarre situation we have today in which many access policies allow embargoes to differ from discipline to discipline.

The latest salvo in this debate comes from a report commissioned by the UK’s HEFCE from the British Academy, the UK’s National Academy body for the Humanities and Social Sciences scholars. The first thing to say is that the report brings a valuable set of information on the landscape of scholarly publishing in H&SS to the table – in particular the data reported in Chapters 2 and 3 are going to be very helpful. In this post I want to focus on the conclusions the report comes to on embargoes and the work reported in Chapters 4 and 5.

The report sought to provide new data on potential risks to the sustainability of the H&SS journals business. This data comes in two forms, the first (Chapter 4) is an analysis of the online usage patterns of a set of H&SS journals – which is intended to inform a discussion of the business risks to  journals of their articles appearing in repositories. The second (Chapter 5) is survey data on UK collections librarians asking what the most important factors are in cancelling journal subscriptions.

The usage data is problematic for two reasons. Firstly while much data has been presented on the way article usage declines over time, with claims that it is relevant to discussions of embargoes, no link has ever been established between usage patterns, free online access and damage to the business of a journal. Second, there are some serious issues with the maths at the heart of the analysis. For example while the mean half lives (putting aside a curve shape for which the term ‘half life’ is meaningless) differ between disciplines the 95% confidence intervals are huge (Table 5). No testing of whether these differences are statistically significant is reported but a line in the report suggests using this data to support the report’s conclusion is untenable: ‘…the main differences are within disciplines, not between them’ [p59]. I will dissect those issues in detail at a later date but for the purpose of this discussion I’m going to take the usage conclusions of the report at face value. On the basis of reported mean half lives the report states that H&SS, along with physics and maths, form one cluster with biomedical sciences being different.

The survey data from Chapter 5 is also useful.  While all surveys have their limitations the results confirm the anecdotal evidence that librarians regard availability of non-version of record article copies and the timing of availability as amongst the least important information in deciding on journal cancellations. Usage by their institution, the requirements of research staff to access the version of record and cost are rated as more important. As an aside there is an oft-referred to ALPSP survey that is cited as coming to the opposite conclusion to the BA results. That survey was deeply flawed (see my Commons testimony for details) but even putting aside those issues it asked a different question: essentially ‘all other things being equal…would shorter embargos lead to cancellations’. The BA survey shows that in the real world, where things are never equal, that this is an unimportant factor.

The conclusion I draw from these two sets of data is that there is no value in longer embargoes for H&SS – indeed that there is no need for embargoes at all. H&SS cluster with physics and maths, disciplines where substantial, and importantly concentrated, portions of the literature have been available prior to publication for over 20 years and where there is no evidence of a systemic failure in the running of sustainable publishing businesses. A recent report from the London Mathematical Society (LMS) stated that they saw no risk to their business from the availability of author manuscripts online and that online availability of author manuscripts had no significant effect on traffic to their journal sites (they note that the situation might be different if the version of record is online but no public access policy requires this).

This runs counter to the standard narrative to justify embargoes. Why do institutions continue to subscribe to journals when the ‘same’ content is available online for free? This would only be the case if factors others than online availability of manuscripts drove subscription decisions. This might be the case if other factors, such as overall cost, scholar demand or access to the version of record were more important factors. This is exactly what the survey data in the BA report shows supporting the view that short embargoes are not a risk to the sustainability of subscription journals in H&SS.

The report itself however comes to the opposite conclusion. It does this by creating a narrative of increasing risk based on potential loss, that things might change in the future, particularly if the degree of access rises, that the survey can only ask about the current environment and hypothetical decisions. We can, and no doubt will, continue to argue about the degree of this risk and what evidence can be brought to bear. For me decades of the ArXiv and Astronomy Data Service and seven years of mandated deposit to Pubmed Central with no evidence of linked subscription cancellations seem like strong evidence. Remember that the report states that physics and maths are similar to H&SS. But increasingly I’m feeling this whole argument is rather sterile.

The more productive approach is to ask question from a positive perspective. Do embargoes help a traditional publisher or journal to navigate the rapidly changing environment? If so what embargo is optimal? Many publishers are choosing to offer an APC based Open Access publishing offering. We know this works in many (well funded) STM disciplines. However we also know that in H&SS there is less free cash and a transition to a directly funded system will be more challenging. Different approaches may well be required. A real alternative for a publisher is to take a strong stance on the additional value they offer in the version of record by supporting author archiving.

If the final published product is easier to read, easier to find, better formatted or better integrated with a scholar’s workflow and they find the value add sufficient then they or the institution will be willing to pay for it. The survey data, the LMS report and in a different way the SCOAP3 program all show that where customers see added value created by a publisher over the manuscript they are willing to pay for it. Any publisher, if they want to survive, has to compete with the free circulation of scholarly documents – both ‘grey’ literature and the circulation of copies of the version of record. That competition just got a bit tougher with the advent of the internet.

Embargoes are an artificial monopoly created to make the competition a bit less fierce. But truly, if a publisher believes that they add value and wants to be competitive then why should they fear a Word doc sitting on the web? Indeed if they do it suggests a lack of confidence in the additional value that they offer in the version of record. The best way to give yourself that confidence is to be tough on yourself and take a good look at how and where you add value. And the best way to do that is to compete successfully with “free”.

The best way to stay competitive is to be prepared and ready to adapt when the competition gets tougher, not to collude on retaining monopolies that make that competition less intense. (As an aside I wonder from time to time whether the push from large publishers for longer embargoes is in part a strategy to make smaller publishers more complacent on this issue and easier to pick off as margins shrink and scale comes to matters even more than it already does)

No doubt we will continue to argue for some time yet whether short embargoes cause harm. But for those traditional publishers with a subscription business seeking to navigate our rapidly changing world I think it’s precisely the wrong question. Ask whether longer embargoes actually help your business over the short, medium and longer term. Really ask that question, because if the answer is yes, you have a problem with your own confidence in your own value creation.

To my mind the British Academy report shows pretty strongly that embargoes don’t help. They offer essentially no business benefit while reducing the focus on the one thing that matters – creating the value that should differentiate scholarly communications from just slapping something up on the web. Clearly the authors of the report differ.

But rather than just take my summary, or that of the report, at face value my suggestion would be to do what any scholar should do: read the report in full, critique the evidence and come to your own conclusion. What this report does bring is new data and in particular data focussed on the questions of managing a transition to wider access for the H&SS literature. We need more of this data and we need to focus all of our critical faculties, and all of our various disciplinary approaches, to understanding how to use that data to effectively plan and manage a transition to the Open Access future.

This entry was posted in Open Access, Open Access policy. Bookmark the permalink.

One Response to The Embargoes Don’t Work: The British Academy provides the best evidence yet

  1. Pingback: Week in Links – May 16 | Open Access Button blog

Leave a Reply

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <s> <strike> <strong>