Researchers have for many years had access to new platforms and channels for networking and sharing resources – the ‘A-Z of social media for academia’ was already a substantial list at its launch in 2013, and now links to almost 150 ‘items recommended by academics for use in their professional lives’.1 It is widely assumed that researchers’ use of these new communications channels has accelerated in recent years, and that in many cases it comprises sharing copies of the full text of their published articles and books or book chapters. In light of a growing level of concern among publishers – and to some extent among librarians, particularly those with copyright expertise – about this ‘off-grid’ sharing of copyrighted content, we sought to test the underlying assumptions and determine what level of realignment might be necessary to bring the interests of publishers, libraries and researchers more closely together.
Exploring researchers’ sharing behaviours
The ‘A-Z of social media for academia’ list also retains links to defunct services, including iAMscientist, MyOpenArchive and Posterous, reminding us that many services aimed at academics are ultimately unsuccessful. Where services launched by scholarly publishers and libraries have failed, it is often because they necessitated a change in researcher behaviour – for example, how and where they discover, share or discuss work, or how and where they connect and collaborate.2 The success of any new initiative is clearly dependent on the seamlessness of its fit within current researcher workflows. In that context, we3 set up the FairShare Network4 in 2016, to explore researchers’ choices, preferences and expectations around sharing, and flesh out the framework into which related services and solutions must fit. The FairShare Network undertook a survey in spring 2017, which was distributed via e-mail and social media, reaching an estimated 200,000 researchers. We had 7,538 responses, of which:
- 64% self-classified in STEM subjects and 36% in humanities and social sciences
- 40% came from Europe, 18% North America, 14% Asia Pacific, 8% South Asia, 6% Middle East, 6% Africa, 5% Central & South America, 3% Australasia
- 18% were still completing their PhD, 39% had completed their PhD within the last 10 years, 16% had completed their PhD 11–20 years ago, 7% had completed their PhD 21–30 years ago, 5% had completed their PhD 31–40 years ago, 3% had completed their PhD more than 40 years ago, 2% preferred not to answer and 10% did not have a PhD.
Which scholarly collaboration networks (SCNs) do researchers use?
We asked people about which networks they were familiar with, inviting them to choose from a drop-down list and also enabling them to give a free-text response indicating any others. We then asked them how often they used those networks with which they had indicated they were familiar (Figure 1).
ResearchGate, Academia and Mendeley were the most commonly used networks, in terms of the number of people that were familiar with them (the n numbers). However, usage of ResearchGate is substantially more frequent, with over a quarter (26%) of ResearchGate users in our survey indicating that they use the site at least daily, another 41% indicating that they use the site at least weekly, and 18% at least monthly. In total, 85% indicated that they are using the site at least monthly. A previous study by the Nature Publishing Group5 saw a similar figure (88% of ResearchGate users visiting the site at least monthly) but the difference between the two results sets would indicate that the breakdown is changing in favour of more frequent usage of SCNs. In Nature’s 2014 study, only 8% of ResearchGate users visited the site daily, compared to 26% in our 2017 study. This serves as a reminder that SCNs are not a last resort when accessing scholarly publications – they are a daily resource.
What do researchers do with SCNs?
The nature of SCNs as a daily resource is also borne out by a shift in what researchers say they are doing with SCNs. It was previously the perception that researchers were using them as ‘digital CVs’ – Nature’s 2014 study6 had 1,589 users of ResearchGate, of whom 68% said they used it ‘in case contacted’, which was the most popular reason given for using the site; only 35% said they were using it to post their work or discover other work.
However, in our 2017 study, we found that the most popular purpose for which SCNs were being used was to access otherwise inaccessible content – 66% of researchers were using these sites for access, while 57% were uploading their own work. We are unable to draw a longitudinal comparison between these two separate studies, but the difference between 35% sharing/accessing work via ResearchGate in 2014 and around 60% sharing/accessing in 2017 is indicative of a trend that has grown extremely fast in just three years.
How does researchers’ usage of SCNs square with their views on copyright?
As well as asking researchers about their sharing behaviours, we also asked them about copyright. Perhaps surprisingly – particularly given the sharing behaviours they had just acknowledged – 83% agreed or strongly agreed that copyright should be respected. On the other hand, 60% agreed or strongly agreed that they should be entitled to upload their work to SCNs, regardless. Our finding that most respondents (79%) say they do check copyright before uploading publications to SCNs is also at odds with another recent study7 which found that, in a sample of 500 full-text articles on ResearchGate, 40% contravened copyright.
Our interpretation is that researchers do value copyright, but they also (primarily perhaps?) value convenience. Networks provide an easy way to promote themselves without being seen to be too self-promotional. Uploading work is also convenient because it helps to reduce the flow of requests via e-mail. SCNs put a lot of ticks in the convenience box and – given that it is only possible to list your work in such sites by uploading a PDF – convenience ends up trumping conscience, and authors end up sharing the full text even if they feel a bit uncomfortable about the copyright implications.
How else do researchers share and access work?
It is also important to remember that SCNs are not the only ways that researchers share content (Figure 2). Approximately half of the survey’s respondents – about the same number as were using Mendeley – simply get content they need by e-mailing the author, or by Googling. Other ways of obtaining content include the Twitter hashtag ‘#icanhazPDF’ (an academic spin on the ‘I Can Has Cheezburger?’ cat meme);8 in 2011, academics started using the #icanhazPDF hashtag to ask each other for PDFs, via Twitter – the person requesting the PDF posts a DOI, their e-mail address and the tag. Someone with access to the article then sends it by e-mail, and the original tweet is deleted.
Researchers also post copies of their work on their own or their institution’s websites, in their institution’s repository, or in subject repositories. With the rise of services like the Open Access Button, Unpaywall and Kopernio, which automatically find a free copy of anything for which a user has hit a paywall, those shared copies are increasingly visible.
How are publishers responding to authors’ sharing?
Author sharing – when it contravenes copyright agreements – places publishers in an awkward position. Most wish to ensure that the concept of copyright is not undermined by publishers being seen to turn a blind eye to infringement, but at the same time, most wish to protect their relationships with authors and be supportive of efforts to maximize readership. Legal challenges to the hosts of illegally shared PDFs are ‘blunt tools’ and not ‘sustainable’,9 and have damaged the relationship between publishers and authors.10, 11 They have led to minor changes12 to ResearchGate’s terms and conditions, and to some of its functionality, but on a practical level, these changes have not actually helped publishers to better count ‘off-site’ usage. Meanwhile, individual publishers,13, 14 providers15 and collaborations16 are working on or have launched new approaches to link sharing. However, authors cannot post a link, by itself, to PDF-based sharing systems (such as repositories, or SCNs). Clever new linking solutions are therefore solving a different problem, not directly tackling the challenge of PDF-oriented sharing systems.
A win:win:win solution
The FairShare publishers, in contrast, have recognized this and supported the development of a PDF-based solution. Building on the FairShare Network’s research, Kudos has worked with development partners the American Thoracic Society, Emerald Publishing, FASEB, The IET and MIT Press to prototype and test a ‘shareable PDF’. Given that the only way for authors to list their work in some systems is to upload a PDF, we have created a summary PDF that links through to the full text on the publisher website. This enables publishers to leverage researchers’ continued use of SCNs – harnessing them as discovery channels, rather than trying to compete with or displace them – and also results in the creation of added-value content that can be used by authors to promote their work in a range of other channels.
Shareable PDF has been quick and cost-effective to prototype, as it builds on existing pre-publication integrations between Kudos and manuscript submission services such as ScholarOne and Editorial Manager, and on Kudos’ existing reporting suite for researchers, publishers and institutions. The PDFs (see example, Figure 3) contain a plain language summary of the article – making it easier for a wider audience to find and understand – and include ‘value-added’ content such as an author ‘perspective’, along with trackable links back to the full text on the publisher website. Shareable PDFs can be created before, at, or after publication, and the workflow is consistent across all participating publishers, making it easy for authors to upload them to online profiles. The creation, sharing and response to Shareable PDFs is tracked, and reported in the context of other sharing efforts (e.g. e-mail, social media) and metrics (e.g. usage, citations and Altmetrics) to give authors unique intelligence into which approaches are most effective in building readership and impact. These data also help publishers and institutions to surface, measure and compare sharing and usage via ‘closed channels’ (SCNs and other PDF-oriented systems, as well as e-mail, and personal Facebook/LinkedIn postings), none of which are visible to other tracking systems.
Shareable PDF can also act as a ‘container’ for the linking systems referred to above, meaning that it can ensure that users of SCNs and author/institutional websites can still gain access to the full text via those sites. The benefit of this approach is that it brings usage via those channels back into the publisher–library sphere, enabling it to augment (rather than depress) institutional usage statistics, and opening it up for measurement and comparison with other sources of usage. Everyone wins: authors benefit from improved metrics for their publications (because usage will not be fragmented across multiple platforms); publishers and libraries gain new abilities to track and understand researchers’ sharing via otherwise closed channels; and SCNs and other websites continue to be destination sites for discovery and continue to attract the users and site visits on which their business models are in some cases built.
What are the results to date?
At the time of writing – in February 2018 – Shareable PDF has been soft-launched for approximately ten weeks, initially for four publishers, with phased roll-out to a further six publishers during that period. The nature of a soft launch means that while it has been available as part of the existing Kudos workflow for those publishers’ articles, no promotion to or education of authors has taken place. Nonetheless, it has immediately resulted in authors and readers transitioning to copyright-compliant sharing, increased traffic to publisher websites, and only positive responses from authors. Over 500 authors – 58% of those for whom the feature has been visible – have already used the new ‘share as PDF’ feature. Results to date show the number of publications being shared via Kudos has increased by 36%, with PDF immediately becoming the most popular format for sharing (58% again), with the average click-through rate for PDFs that have been shared being 27%.
Authors like Shareable PDF because it retains the value of the sites in which they are sharing – metrics there continue to build – but it also improves the publisher download metrics that are reported to their institutions and funders. They also value the simplicity and consistency that Kudos and Shareable PDF bring to the complex sharing landscape. With so many options for how they disseminate their work, and a range of different publisher policies and technologies, it is helpful to be able to manage the narrative around and performance of their work in a single place.
Researchers are exploring a range of new channels for disseminating their publications, many of which are ‘closed’, making it near impossible to fully understand the readership, or to understand the best ways to build readership and impact. The successful launch of Shareable PDF indicates that authors and readers are willing to adopt new approaches to sharing in those channels, enabling publishers and institutions to track, report on and benefit directly from content usage in networks and channels outside the traditional information ecosystem. This has the potential to change publishers’ dialogue around SCNs (in particular), from trying to compete with or displace them, to leveraging them as highly popular discovery channels.