The TL;DR summary: In February 2017, when Elsevier were accused of selling one paid-for hybrid open access article, at first they sowed doubt about it, then three days later admitted it to be true. In their admission they state that it is the only wrongly paywalled open access article “affected” at their websites. They have apparently checked their “system” and say there are no more wrongly paywalled articles at Elsevier websites. Their statement is however demonstrably incorrect as I show in this post. There are more paywalled open access articles at Elsevier, and by their own admission Elsevier don’t seem to know about them. This is professional incompetence. HybridOA as whole, not just at Elsevier, is not reliable. We should not use hybridOA services if we actually want reliable open access in perpetuity.
Elsevier and other legacy publishers, have an unfortunate track record of paywalling articles that research funders have paid-for to be openly accessible. Elsevier in particular are notable for their consistency in this regard: they have been caught by diligent readers, selling access to “open access” articles in 2014, 2015, 2016 and 2017. They are so notable for this practice that it is on their Wikipedia page amongst a long section of other controversies.
Responding to the March 2014 scandal, Elsevier claimed it was merely “bumps on the road to open access”. Shortly before Christmas (a great time to bury bad news) on the 22nd December 2014 Elsevier quietly disclosed that they had refunded “about $70,000” to readers across the world who had purchased access to articles that should have been free to access at Elsevier. Naturally, as this self-disclosure came just before Christmas it went largely unnoticed and unreported.
Now we fast-forward to February 2017. As an academic, fed-up with abysmally poor publishing services, I have decided to systematically run heuristic checks on articles that Wellcome Trust has paid-for to be open access. The Wellcome Trust and other research funders have started to publish detailed reports on which articles they have paid-for to be open access, with which publishers, and how much they paid publishers to make each article open access in perpetuity. I have used these public records supplied by Wellcome Trust to cross-check articles that are supposed to be open access against what is actually open access at publisher websites.
From this exercise so far I have found at least seven Wellcome Trust paid-for “open access” articles that are currently paywalled, and for sale to readers for up to $58.80 per article, at major academic publisher websites. I have publicly given details of two of these articles so far: firstly, one paywalled by Emerald Publishing Group and one paywalled at Elsevier at a journal called Mitochondrion.
The first response: sow doubt on the claim
Initially, Elsevier responded to my observation on a public mailing list:
Ross – we are looking into this, but an initial glance at the article suggests that the author first decided to publish gold OA and then decided against it. We’ll have a little look to make sure all the metadata is correct.
— Dr Alicia Wise, Director of Access and Policy, Elsevier | Feb 15th 2017 09:21:44 GMT (source)
Note that Elsevier’s initial response suggests that I might have been incorrect in my assertion that this article should be open access and should NOT be on sale to readers.
I am a scientist. I make statements based upon evidence. I am no fool. But Elsevier seem to think I am.
The second response: admit to one error but clearly deny any further extent
Three days later Elsevier admitted that I was correct – the Wellcome Trust had in fact already paid Elsevier £2,168.08 many years ago to make this article open access, it should have been open access at their website, and should NOT be on sale to anyone.
Hi Ross, this has taken a little time to bottom out but indeed the article should be OA. In 2013 you may recall we initiated an exercise to add/correct and generally clean up OA license information on articles. This article was part of that exercise, and a little unusual: we had invoiced the author, and then the institution wanted to receive the invoice but in a different currency, so the original article invoice was unpaid during the cleaning exercise.
Our normal practice is to publish articles OA, but to move them back behind the firewall if after a grace period the invoice remains unpaid. So it was picked up as an article with an unpaid APC and moved behind the firewall. We’ve gone through the system, this is the only article affected. There’s nothing malicious about what happened here, and we’ll be reimbursing the institution’s APC, and your PPV charge. We will also run a check to see if anyone else paid PPV for this article and if so reimburse them as well.
— Dr Alicia Wise, Director of Access and Policy, Elsevier | Feb 18 2017 00:03:24 GMT (source)
Within the above statement from Elsevier there is something much more astonishing. Elsevier clearly wants us all to “move along, nothing to see here” by stating that the Mitochondrion article is “the only article affected” by paywalls that shouldn’t be.
But there are more paywalled open access articles at Elsevier journals
Until now I haven’t disclosed publicly that there is at least one other paid-for “open access” Wellcome Trust article that is currently paywalled at an Elsevier journal, despite having been handsomely paid-for to be made open access in perpetuity. It’s at a rather well-known Elsevier journal too; The Lancet. According to data Wellcome Trust has made available, in 2012 they paid Elsevier £5,280.00 to make the article in the screenshot below, open access in perpetuity. In 2017 I find it is on sale at ScienceDirect for $35.95 + VAT. I bought access to it just to verify that it is actually on sale and indeed it is.
The screenshot below is my receipt documenting my purchase of the above “open access” article from The Lancet:
Now let’s return to Elsevier’s most recent statement and you’ll see why I’m shocked:
“We’ve gone through the system, this [the Mitochondrion article] is the only article affected.”
I have just provided evidence (The Lancet article) that there are more paid-for “open access” articles at Elsevier that are paywalled and have been sold to readers. As my receipt demonstrates, my purchase of this Wellcome Trust paid-for “open access” article post-dates the above statement by Elsevier. Therefore Elsevier’s statement would appear to be incorrect.
So we now have evidence of two further important things to note:
- Elsevier’s entire system for handling hybrid open access is broken
- Elsevier are evidently incapable of accurate self-assessment
Elsevier say they have gone through their “system” and that there are no other open access articles “affected” by paywalls. But I have just provided clear evidence to the contrary.
To be clear I’m not accusing Elsevier of knowingly lying about this. I am going to charitably interpret this as an unknowingly false statement – they believe it to be true from what their faulty “system” tells them, but in reality it is not true, other articles are affected. It is abundantly evident to me that Elsevier’s internal systems and processes for tracking invoices for APC payments, handling, and marking hybrid open access articles simply aren’t robust. I believe that this forms part of a body of mounting evidence that Elsevier is professionally incompetent at hybrid open access publishing and that this incompetence financially benefits them through pay-per-view payments & re-use licensing to the detriment of the scholarly poor (those outside paywalls). I don’t allege malicious intent. The point is that regardless of intent, this incompetency is demonstrably generating profits for them – that is undeniable and that it is causing real harm (financial loss to readers who have paid for articles that should not be for sale).
What is the true extent of this problem at Elsevier in 2017? Why does it matter?
I have to admit I don’t have a definite answer as to the extent, millions of articles are published every year, I can’t check them all myself. But one thing I do know and can now evidence, is that clearly Elsevier don’t know either! I think a full independent audit with forensic accounting is urgently needed. Why? As I mentioned earlier, back in 2014 Elsevier self-disclosed that they had charged readers $70,000 to access articles that should have been “open access”. Since 2014 the rate of publication of hybrid open access articles has increased so I would expect if a competent independent audit were done of purchases of ‘mistakenly’ paywalled articles between 2015 and now, the sum to be reimbursed might well be in excess of $100,000. Given Elsevier doesn’t know that some of its paid-for hybrid open access articles are on sale, can we trust their self-reported audit of such incidences prior to 2014? I think not, hence a thorough independent audit is needed.
The scale of this could be very big. I have only partially cross-checked the accessibility of hybrid open access articles from just one research funder at Elsevier. According to CrossRef data there are at least 14,155 different research funders. The charge to make articles hybrid open access at Elsevier is typically around $3000 but can be much more as we have seen from The Lancet article which was £5,280 to make open access. When found to be in error, Elsevier typically refund the APC-paid, so therefore if just one “open access” article per research funder is mistakenly paywalled, it would equate to a refund of approximately $42,465,000 USD. This amount is still chump change to Elsevier: a company that in 2013 posted an annual adjusted operating profit of £826 million on an annual turnover of £2.126 billion.
Nor in terms of scale is this scandal confined to just Elsevier. There are paywalled “open access” articles for sale at Emerald Publishing Group, Cambridge University Press and Wolters Kluwer / LWW to name but a few. I have evidence, which I have already informed Wellcome Trust of. I don’t make these claims lightly.
I stress however that this scandal is confined to hybrid open access publishing.
Only fully open access publishers are reliable
How many times have PLOS accidentally paywalled an article? 
How many times have PeerJ accidentally paywalled an article? 
Full open access publishers such as PeerJ, PLOS, MDPI, Hindawi, Copernicus, Pensoft, F1000Research, and Ubiquity Press to name but a few, have an open access compliance “error” rate of absolutely ZERO, out of hundreds of thousands of articles published so far.
In some respects these are facile statements to make about full open access publishers. But that is my point. It is known that PLOS and PeerJ cannot paywall and ‘mistakenly sell’ articles to readers because their systems are open by default and do not have paywalls. At fully open access publishers, all articles are open and thus when an APC is paid or a fee waiver awarded, there is never any failure to make articles available in a manner that is both freely accessible to readers (not paywalled) and licensed under an open access definition compatible Creative Commons licence that Wellcome Trust requires.
If research funders and institutions really want open access publication, under open access definition compliant licensing, they should use full open access publishers.
Luckily not all research funders allow their money to be used to pay for hybrid open access charges. This is sensible policy in my opinion, given the evidently incompetent open access service provided by many, not just Elsevier.
Conclusions: what can funders & APC-paying institutions do about this going forwards?
As soon as reasonably possible, stop paying for hybrid open access. It is not reliable. If funders and institutions really want open access, use preprint servers and full open access journals. Hybrid open access is demonstrably unreliable, vastly over-priced, and does not help towards long term goals of open access as it helps keep subscription-access journals going and takes good articles away from full open access journals. It is time to end the experiment.