Open access – the systems journey

Updating large subscription-based content systems to accommodate new license and payment models is complex, but we’re overcoming the challenges

Over the past year, and particularly last week, the integration of our ScienceDirect platform and an external service has raised questions about whether we are charging readers to download open access articles.

The articles in question were always freely accessible to readers using our platforms; however a button linking to Rightslink, a service from CCC for ordering reprints and other reuse permissions, was appearing on every article on ScienceDirect. Rather than offering appropriate permissions for different types of articles, the same rights and permissions were offered. In some cases, these permissions were already permitted under the applicable OA licenses.


The following update was posted on December 22, 2014:

While at the original time of writing we anticipated all changes would be complete in the summer, this is now on course to be completed by the end of 2014. By this time, we will have regenerated about 100,000 article PDFs on ScienceDirect to ensure that the PDF version is aligned with the HTML version in terms of labelling and licenses. It's not that this number of articles had incorrect labeling, but rather that regenerating one article requires all articles in the issue to be regenerated. We also needed to ensure the articles display correctly across all our platforms, including those of Cell, The Lancet and HealthAdvance, which has added additional complexity.

We also discovered during the course of 2014 that some open access articles were also incorrectly sold through our pay-per-view services, and have taken steps to reimburse or credit customers for a total value of about $70,000. We also refunded the APCs for two articles that had not been properly opened on ScienceDirect at the time of publication. The OA systems journey does not end here but continues. In 2015 we will focus on titles owned by our society publishing partners, for example.[divider]

In summary

Every article on ScienceDirect had a link offering readers the opportunity to purchase additional rights which in some cases were not needed.

We never intended to charge for material or rights that should be free. This problem should not have arisen, and now that it has, we are taking all possible steps to correct it and reimburse those affected. We are immediately refunding all those we are aware of, as well as others we learn of.

About 50 customers have paid unnecessarily. Eleven of the articles were published under a CC-BY license (totaling less than $1,200) and the rest under other CC licenses or the Elsevier User License. The total value of all these orders is about $4,000.

We are stopping the means that led to inadvertent requests for payment and expect the issue to be largely resolved as a result. We readily invite users to contact us via immediately if they find remaining loopholes.

In this article we provide insight into the many systems changes that Elsevier and other publishers are making to support growing interest in open access.

It has never been our intention to charge for material or rights that should be free. Since last summer, we've been working to enhance our open access metadata and also working with CCC to enable them to ingest and distribute our open access metadata to handle these situations correctly.

We've come to learn, however, that around 50 customers have paid incorrectly to reuse content published under one of the open access licenses we offer. Eleven of the articles were published under a Creative Commons CC-BY license (totaling less than $1,200) and the rest under other CC licenses or the Elsevier User License. The total value of all these orders is under $4,000.

We are happy to make amends as it has never been our intention to charge for material or rights that should be free. This is a problem that should not have arisen, and now that it has, we are taking all possible steps to correct it and recompense those affected. We have begun the process of issuing refunds to these customers, and have also turned these links off for all open access articles on ScienceDirect to ensure that no one else is inadvertently charged for these permissions. We are also working to ensure that none of our other platforms include Rightslink buttons for OA content until we've fully integrated our OA metadata with CCC. Overall, less than 0.15 percent of the rights and permissions orders placed through CCC for our OA content were handled incorrectly.

The systems journey

Elsevier and other established publishers are making a wide array of systems changes to support the growth in open access publishing in the international research community which we serve.  We would like to share some insight into the nature and scale of these investments.

As demand for open access grew slowly at first, Elsevier published open access content on a relatively small scale between 2006 and 2012. Meanwhile, we were engaged in many stakeholder discussions and pilot initiatives to define how open access publishing can scale and support a broad array of funder and institutional policies. The good news is that since April 2013, the pace of change has accelerated in a positive way with publication of the UK national open access policy. The Finch Group gave all stakeholders a shared understanding of direction and helped choreograph changes that each need to make on our shared journey. Read about some of the latest changes at Elsevier here.

Prior to the clarity and direction provided by the Finch Group, most of our open access publishing relied on manual workarounds in the online systems that had been built over the last 15 years to support subscription publishing. This includes ScienceDirect, but also other platforms such as Cell Press, the Lancet and our society publishing partner platforms too. While these manual workarounds enabled us to start offering OA options quickly, they were never designed to scale. So when the new UK open access policies became clear, our first priority was to get our infrastructure organized to accommodate future articles.

During 2013 we invested significantly in upgrading our core metadata management systems to ensure that we have an automated system which contains a central, consistent overview of all of our open access content.  In September 2013 we integrated our flagship ScienceDirect platform with this metadata to provide discovery, search, and browsing services for open access content. You can read more about those changes here.

In February 2014, internal systems were launched to enable our production department to automatically capture and syndicate this open access metadata to a wider array of systems. Our current focus is to integrate properly with third-party platforms (e.g., Rightslink) and to ensure our open access content surfaces correctly on the other Elsevier systems.

We are also working to ensure that the metadata is corrected for all of the free and OA content published prior to April 2013 under the old manual workarounds. This includes articles where an author or funding body has paid an Article Publishing Charge, content that Elsevier has decided to make openly available such as from our Open Archive journals, and miscellaneous content such as editorials in selected journals. We are now sorting this properly into a set of distinct classes, which can then be properly tagged and labeled so that each class of content can be handled appropriately in internal and external services.

So when will this all be reliable?

As anyone who has ever administered a database knows, reconstructing one with incomplete metadata  is a time-consuming and largely manual process. Currently, there are still some inconsistencies in license statements and copyright lines for content due to the article's metadata even though the articles in question are freely accessible. We are updating the copyright statements on existing PDFs and in HTML and XML versions of articles too and ensuring all of our OA content includes an explicit link to the relevant license stating re-use terms. We are working through these tasks as quickly as we can with the expectation to have it completed by the summer of 2014.

We have already solved many classes of problems, but there will continue to be some bumps on this journey toward open access. We take operational excellence very seriously here, and so we are working very hard to rectify all problems.

If you spot something that is wrong, please don't hesitate to let us know. We are systematically investigating whether there are cases where customers have been charged incorrectly, and if so we will proactively reach out to anyone impacted. Again, if you have any questions or concerns, please let us know. You can contact us at any time via

Elsevier Connect Contributors

Chris ShillumAs VP of Product Management, Platform and Content for Elsevier, Chris Shillum (@cshillum) is responsible for Elsevier's shared product platform. His team looks after the content management systems, access management systems and APIs that power our flagship products including ScienceDirect and Scopus. His current work includes looking into the application of text analytics and big-data processing capabilities to our products, and he is helping to define our text-mining strategy. He also represents Elsevier on the boards of key industry nonprofit organizations, including CrossRef and the International DOI Foundation. Most recently, he has been leading Elsevier's participation in the ORCID initiative and helping to ensure its success.

Alicia Wise, PhDAs Director of Access and Policy, Dr. Alicia Wise (@wisealic) leads in driving Elsevier's vision for universal access to high-quality scientific publications. She leads strategy and policy in areas such as open access, philanthropic access programs, content accessibility, and access technologies. Based in Oxford, she has a PhD in anthropology from the University of North Carolina at Chapel Hill.

comments powered by Disqus

6 Archived Comments

Charles Oppenheim March 21, 2014 at 12:49 pm

Whilst this initiative is welcome, I still don't understand why Elsevier started offering gold OA options in the first place without having the technology in place. The history of hand-crafted activities is laughable. Whatever way you look at it - either Elsevier has done this deliberately, or it is stunningly incompetent (I believe the latter, though others will disagree) - the company comes out of this sorry saga appallingly. I hope the Elsevier staff responsible for this fiasco will be disciplined.

Alicia Wise March 21, 2014 at 2:41 pm

Hi Charles,

Thanks for welcoming this initiative, and for the question. You ask why we began offering gold OA options without perfect systems in place. Would the UK electronic library initiatives ( have got off the ground were this the bar? Of course not. (As we worked closely together for some time in JISC contexts, I hope you will permit me to draw on our common shared experiences here.) Large, complex, systemic change takes times and routinely involves the sorts of pilots we engaged in and describe in our article.

You ask if we have ‘done this deliberately’. If you mean systematically piloted and explored, and then committed publicly and whole-heartedly – as part of the UK journey toward Open Access envisioned by the Finch Group – to work with other stakeholders on a managed transition to scale up open access publishing, then yes, we did. Will there be bumps on this road? Yes, there have been and there will be. Will we celebrate the very capable and dedicated staff who are making this transition possible? Yes we will: deliberately and without blaming them for failing to do the impossible.

With kind wishes,


Charles Oppenheim March 21, 2014 at 3:58 pm

Alicia, with respect, the elib initiatives were always meant to be experiments, and it was expected that many would fall by the wayside but a few would succeed, and either way lessons would be learned. In contrast, I would expect a large commercial publisher starting a new service to have thoroughly tested it before launch. You say the technical staff are dedicated and capable; then the fault lies with the manager who authorised the launch. I worked for 12 years in the electronic publishing industry and know just how heavy the alpha and beta testing should be before a launch. Someone has let Elsevier down very badly.

Alicia Wise March 21, 2014 at 8:26 pm

Hi Charles,

We haven’t launched ‘a product’ in the way you perhaps assume. Rather we are embedding open access very broadly across the organization, in an array of products and services, and through a broad range of strategic initiatives.

You are right that the eLib programme was a portfolio of projects, however they were also part of a broader strategic effort to galvanize the creation of digital libraries. Your post led me to search and re-discover this piece by Reg Carr about the Follett Review ( In it eLib is positioned in this broader strategic context.

With very kind wishes,


Chris Shillum March 24, 2014 at 4:35 pm

Some of you have asked us to provide additional information about steps underway to correct open access articles that bear "All rights reserved" in their copyright lines, or which do not bear an open access license, and to explain why this will take some time. We are happy to do so.

The quick answer is that we have plans underway to correct the PDF and HTML versions of open access articles across all of our platforms. The reason this will take some time is that we have a series of steps to complete, starting with adjustments to our data model and implementation of those changes in our core systems, followed by data migration and integration with each of our end-user platforms.

The way content in presented in our online platforms reflects decisions taken in the design of our original SGML DTD in 1992, later updated in XML in 2005. 2005. ( Back then a decision was taken, for various reasons, not to capture in XML the explicit copyright line as it appears in the printed article, but rather to tag the copyright holder, along with a couple of attributes indicating the copyright type and the copyright year. Each online platform was then to use this information, along with "base data" indicating for example, the Elsevier division that owned the journal and the article type, to reconstruct the full copyright line for online display.

As our publishing policies have evolved, and the variety of copyright transfer and license agreements we support has increased, the logic that has to be implemented in each online platform to generate the correct copyright line has become more and more complex. In turn, ensuring that all of our publishing platforms, along with those of our partners, implement these rules correctly and keep up to date with changes, has become increasingly difficult.

We are now fundamentally changing our approach to managing copyright metadata: we will generate verbatim article copyright lines at a single point of our production process and distribute them for display in all of our online platforms. We will also move copyright information out of our main XML files and into the new OA metadata store described in the main body of this article. The new architecture includes a messaging service which allows us to push out changes in OA and copyright status to all our platforms without reprocessing the content itself. This is really important, for example, to support our Open Archive program and also to allow us to correct mistakes quickly if they occur.

We are now working through a multi-stage process to implement these changes. First, we have to design and extend the metadata schema of the OA metadata store to fully represent our fairly complex article level copyright data; second we have to gather and migrate the copyright data from our existing articles into the new store; and finally we have to integrate all of our publishing platforms with the metadata store and change the rendering rules to use the new source of metadata to display the copyright information rather than the old.

This process requires careful coordination among multiple development and content management teams and will unfortunately take several months to complete. Separately, we also need to update the copyright statements in our PDF files which require having our typesetting partners reprocess and redeliver them.

We will aim to make incremental improvements and corrections as we move through these implementation steps, but this does mean that different versions of articles may be out of alignment during the transition. Our aim is for all steps to be complete by the summer, and this continues to appear an achievable date. While this work is ongoing, the articles remain freely accessible to all.

Egon Willighagen March 25, 2014 at 7:35 am

Dear Chris Shillum,

thank you for taking the time to explain some of the things where Elsevier stands at this moment. It surely highlights why things are the way they are. It does surprise me, though. Changing a XML Schema is not difficult, making tools to test workflows (unit testing) is not difficult either. It is common to make bad decisions in IT design, because it is just hard to predict the future. That is known, but that if you are a multi-billion company, these really are minor problems, unless you, at an executive level, decide to ignore them. The first thing that would have been done, under my supervision, is to expand the metadata as soon as you know there is going to be variance. If you decided to misrepresent that in the past, so that you no longer expand to the correct information, than some seriously bad decision was made in the past. Fine. That happens, and Elsevier has had almost 10 years to fix that. Why hasn't that been fixed yet? That cannot be due to lack of funds, it can only because lack of interest. That is what the community is disappointed about: with so much revenue, with so much profit, you cannot hide behind these fairly common IT issues. IT is hard, but you took on the job, and then you do it properly, and you are transparent in what goes wrong. In this latter, you have just taken steps.

Here are some comments on some of the statement. You write that representing copyright and license information has become more and more complex. It has not. It became complex because Elsevier made a mistake in the past. Open was a topic in 2005 already; Elsevier could have done it right from the start; it should and must have done so.

You write that "[you] are now fundamentally changing our approach..." and that "[you] have to design and extend the metadata schema". I am happy to hear that, but it this was started too late, and this claim was made before too. I find it hard to imagine that since "The Cost of Knowledge" this project is still in the designing phase! That is shocking!

If you really like to establish some trust with (some of) the community, I invite Elsevier to report weekly about the progress during these months, the exact changes made to the XML Schema, publicly ask for input on the design. The community has a lot of experience in this area, and Elsevier can make use of that.

Additionally, I strongly recommend Elsevier to hire an Open Knowledge expert. Someone that has been in this field for more than 10 years (there are plenty), that can help Elsevier do this transition in a smooth way.

All explanation that I have seen so far suggest that Elsevier has an outdated, expensive workflow, expensive IT systems they do not understand, and unable to take publishing to this century, like newer publishers are doing. Defending an outdated business model (referring to your outdated production platform) must not be defended with IP. That is counterproductive and you know you are loosing support from the scholarly community every day.

You cannot continue selling steam engines and then try to ask patience of your customers that steam engines are complex. If you failed to innovate into gasoline engines, that was a business mistake. Elsevier made this business mistake. This gasoline engine has been around for more than 10 years now, and Elsevier has been ignoring it (by insignificant investment). Online platforms around steam engines may sounds attractive, but only has made it worse for you. As now is apparent.

Good luck with the transition of Elsevier into the electronic age. It goes way too slow, and I do not envy you in having to do the PR of this fail of the executive management.

(On behalf of myself and not my employer.)

Mike Taylor March 25, 2014 at 10:04 am

All of this makes sense as far as it goes. The question that remains is, how is it possible that the pages in question have the gold "Open Access" label at the top? Whatever metadata is used to generating that can surely also be used to prevent the generation of "All rights reserved" and the other incorrect information. See for example

Chris Shillum March 25, 2014 at 6:12 pm

Egon – thanks for taking the time to read my comment and respond. As mentioned in the original post, our journey to upgrade our systems really started just under a year ago with clarity about how the recommendations in the Finch report would be implemented in the UK. Could we have initiated changes sooner – certainly, and if we had we would still be making systems changes now because only hindsight is perfect. We are now making significant investments in an infrastructure to support OA publishing, but – as I am sure you and other readers will appreciate – while none of the individual steps are particularly complex, they have to be taken one at a time and coordinated carefully.

I think we will have to agree to disagree on your steam engine analogy – Elsevier was at the forefront of investment to move scholarly publications from print to online, and we continue to develop highly innovative services and solutions. The current investments in our OA systems are designed to get our OA publishing services to this same high standard.

Chris Shillum March 25, 2014 at 6:21 pm

Hi Mike – thanks for the comment. The metadata which drives the Open Access label and the copyright line are currently stored in different systems. In the long term we need to move these to the same place. I mentioned in my earlier comment that we are also looking at short term fixes, what you’ve suggested may be one of them depending how long it takes to implement vs the long-term fix.

Charles Oppenheim March 25, 2014 at 7:43 am

Alicia, I see at the bottom of this web page the words "Copyright © 2014 Elsevier B.V. All rights reserved." Yet your own Terms and Conditions for contributors to this discussion states

"We do not claim ownership of any content, application or other material that you or third parties provide to us (including feedback and suggestions) or post, upload, input or submit on or through the Site, including our blog pages, message boards, chat rooms and forums, for review by the general public, registered users of the Site or by the members of any public or private community ("Submission") and we are not responsible for their content, accuracy or compliance with relevant laws or regulations. However, by posting, uploading, inputting, providing or submitting ("Posting") your Submission you grant us and sub-licensees a royalty-free, perpetual, irrevocable, worldwide, non-exclusive right and license to display, publish and otherwise use your Submission in any format in connection with the operation of our respective businesses (including, without limitation, the Site).

" The fact that you claim copyright in text I have submitted to Elsevier under a licence is, alas, typical of the approach Elsevier takes to copyright law and the law of contract. I now require you to amend the notice at the bottom to read "© 2014 Elsevier b.v. and contributors" (note the removal of "all rights reserved", because I choose NOT to reserve such rights).

Alicia Wise March 25, 2014 at 6:04 pm

Hi Charles,

You are right, there is a conflict between the general copyright notice which appears on all our web pages and the specific terms and conditions for Elsevier Connect itself. We can improve on this, and will do so. You will notice the same problem on ScienceDirect where the generic terms and conditions are actually at odds with the copyright notices and licenses on full-text articles. Another example of the journey that we are on.



David Roberts March 27, 2014 at 1:13 am

(Replying to Alicia Wise)

It is not just ScienceDirect: I went through all versions of an article affected by this bungle that a blogger pointed out, by using the doi url, and then using all four options for access presented, including a site designed for use by the nursing community. The copy of the article on the latter had no copyright statement, merely the page as a whole---in which the html-coded article was embedded---had a standard "Copyright © 2014 Elsevier B.V. All rights reserved.".

Even ScienceDirect html version had a second copyright line that clearly applied to the article, even if the whole page was claimed copyright by Elsevier at the bottom.

This sort of correction needs to applied to _all_ Elsevier's platforms, I would wager.

© David Roberts, released under a Creative Commons Attribution (CC-BY) license.

Bernhard Mittermaier March 25, 2014 at 8:36 am

Dear Alicia and Chris,

I have two questions:

You say "Overall, less than 0.15 percent of the rights and permissions orders placed through CCC for our OA content were handled incorrectly."

Does that mean that 0.15 percent of the orders dealt with OA content (and 99.85 percent with Toll Access content) or does that mean that 99.85 percent of the orders regarding OA content were handled correctly and 0.15 percent incorrectly?

In my opinion there are two sorts of clients which have been mistreated: One sort (about 50) paid for something where there was nothing to pay for, and you say you will refund. Good.

The other sort (far more than 50) paid for something that wasn't deliverd as it should have been. These are the authors along with their institutions and funders who have paid for their "sponsored article". Can you comment on their situation?



Alicia Wise March 25, 2014 at 6:57 pm

Hi Bernhard,

On the first question, it’s the latter – more than 99.85% of orders for OA content were handled correctly. On the second question – all of the articles have been freely available on ScienceDirect.

With kind wishes,


Bernhard Mittermaier March 26, 2014 at 10:28 pm

Dear Alicia,

with regard to the first question: So that means that you have a total of about 33,000 transactions were you charged customers correctly for reusing OA content (if 50 are 0.15%, 100% are 33,3333)? What kinds of reuse are these? Under a CC license, a third party isn't allowed to do everything, but in my understanding it isn't Elsevier who could grant such a right, but only the authors. Or do I misunderstand anything? Or is that the case under the "Elsevier license", and all encounters fall under that license?

With regard to my second question: Having the articles freely on ScienceDirect is surely the most important thing when "sponsoring" an article. But it's not everything because otherwise there was no need for different licenses and authors would be fine with any license. Imagine you were selling cars in different colours, and regardless of the chosen colour each customer received a red one. What will the buyers of blue and black cars say if you tell them: "Never mind, you did get a car."?

Best regrads,


Alicia Wise March 28, 2014 at 12:58 pm

Hi Bernhard

Thanks for the comment. The numbers are indeed right. The vast majority of these are for people who check whether they can proceed without additional permissions and who are told ‘yes, you can’. Where an OA article is made available under one of the non-commercial licenses (e.g. CC-BY-NC-ND) then there are sometimes charges for specific commercial re-use permissions. Examples might include a magazine publisher that wishes to re-publish 10 illustrations, or a pharmaceutical company wanting to use the abstract, an excerpt, and a table in marketing materials. Secondary rights revenue of this kind is very important for some journals, particularly in health and life sciences, and can help to keep both APCs and subscription prices lower than they would otherwise be. Our right to administer these types of permissions flows from the license to publish granted to us by authors.

We absolutely agree with you that having open articles freely available on ScienceDirect is the most important thing, and also that it is not the only important thing: licenses and labels are important too. We have needed to make some difficult decisions to phase and prioritize our system developments.

With kind wishes,


Charles Oppenheim March 26, 2014 at 11:22 am

Regarding the reply to my comment, I expect Elsevier Connect to change its copyright notice right away. I have proposed the wording though minor variations of what I propose would be fine. Unless this trivial change is carried out right away, I will be contacting my MP to make a formal complaint about Elsevier's business practices and will ensure maximum publicity for the issue. I understand why Elsevier's complex journal article publishing systems need time to change, but I am talking about the simple wording that appears at the bottom of a web page.