All posts by Arthur Smith

A Fast-Track Route to Open Access

In the last two years, since the REF 2021 open access policy came into force, the Open Access Team has received an ever increasing number of manuscript submissions for archiving in Apollo, Cambridge’s institutional open access repository.

We have been thinking long and hard about ways to cope with the workload, by scrutinising existing practices and streamlining workflows, because we want to provide the best possible service to our researchers, commensurate with the University’s world leading research.

This blog introduces what is perhaps the greatest overhaul of our workflows since the service began: a new ‘Fast Track’ deposit system.

Work it harder

Before the start of the REF OA policy (2014-2016), the Open Access Team would process and manually curate every manuscript submission we received. Authors could expect an initial response within 1-2 working days, after which (usually within a month) we would archive their manuscript in Apollo.

A simplified workflow for a typical manuscript was:

  1. Manuscript uploaded by submitter in Symplectic Elements.
  2. Item created in Apollo (DSpace) workflow
  3. Helpdesk ticket created (Zendesk).
  4. Open Access Team reviews manuscript, advises submitter and makes a decision.
  5. Open Access Team archives the manuscript in Apollo and informs submitter.

Both the decision (4) and archive (5) steps take time. For each manuscript we would need to decide whether the files we received could be archived, what funder open access policies were at play and the open access options available from the publisher. We could then advise authors about their open access choices.

To archive a manuscript the process was broadly the following:

  1. Review the helpdesk ticket (Zendesk) for the open access decision.
  2. Enter as many publication details as possible in Symplectic Elements.
  3. Retrieve the submission from the Apollo (DSpace) deposit workflow.
  4. Add licence and metadata to the record.
  5. Review the submission and approve for archiving.
  6. Move the item to the relevant departmental collection and apply an appropriate embargo (if required).
  7. Finally, update the helpdesk ticket and send the original submitter a link to their Apollo record.

Each manuscript took on average 18 minutes to archive, which, besides being manually tedious and prone to error, was extremely time-consuming. Add to this the time required to make the initial decision and each manuscript submission could easily take 30 minutes for the Open Access Team to fully process from start to finish, especially if an open access fee had to be paid.

Fast-forward two years and with the rate of new manuscript submissions now peaking at over 1,300 per month, simply processing manuscripts for the REF would require more than four full-time staff members. Whilst these manual processes were viable for a handful of submissions a day, they became unwieldy at scale.

Make it better

Our first attempt at speeding up our open access system began in August 2017. To start we made a number of operational changes to reduce the time spent processing manuscript submissions:

  • We would rely entirely on the metadata present in Symplectic Elements to populate the Apollo records (i.e. we would not curate manual records).
  • The Open Access Team would no longer update the helpdesk records, instead internal record keeping would be automated as much as possible.

Unfortunately, the number of steps in the Apollo workflow was still roughly the same as the previous process, but with one key difference: a new field to record what we call the ‘Fast Track’ decision. There were seven Fast Track options:

  • Submitted
  • Proof
  • Published (not open access)
  • Published (open access)
  • Accepted (published)
  • Accepted (not published)
  • Other

The first six options represent the vast bulk of all manuscripts received by the Open Access Team, and ‘Other’ option simply acts as a catch-all for anything else. By simply knowing what sort of manuscript has been uploaded much of the decision and archiving process can be automated. However, the agent still needed to retrieve the item from the Apollo workflow, check the version of the file and publication status of the paper, add some metadata fields, approve the item, and move it to an appropriate collection.

Figure 1. The Apollo workflow page of a typical manuscript submission, with the addition of the new ‘Fast Track’ field.

The choice of Fast Track decision leads to four possible outcomes which would ‘trigger’ actions in our Zendesk helpdesk:

  • Submitted, proof, published (not open access)
    • Email submitter, ask for accepted manuscript
  • Published (open access)
    • Archive in Apollo (no embargo) ⇒ Email submitter Apollo link
  • Accepted (published), accepted (not published)
    • Archive in Apollo (embargoed) ⇒ Email submitter Apollo link
  • Other
    • Refer to Open Access Team

Despite being a much faster process, it was still manually tedious. It could also require up to 33 actions from agents (29 mouse clicks) and 14 web pages to be loaded, still not very user friendly. However, the time to archive had decreased from 18 to 9 minutes – a 50% reduction from the previous fully manual system.

Do it faster

So what if all the steps involved in processing a manuscript submission could be reduced to the absolute minimum, and be actionable within a single webpage? After a short development sprint, the Open Access Team launched the ‘Fast Track Deposits’ interface last September. A snapshot of the user interface is shown below.

Figure 2. The Fast Track interface. Choosing one of the options in blue is enough to fully archive a manuscript, or process it for further action by the submitter or the Open Access Team.

At the top of the page, the agent can see a ‘publication summary’ including the item title, the journal title, and publisher DOI if available. Both the item title and publisher DOI are hyperlinked, so that the agent can Google-search the item or land on the publisher’s webpage with a single mouse click.

The agent must first inspect the file and check that it is a suitable version (i.e. either the accepted version or the open access published version). If wrongly labelled, they must relabel the file via a dropdown menu, and add/delete files as appropriate. The agent then ‘describes’ the manuscript (i.e. decides whether it is the accepted, published, submitted or proof version) and submits their decision. The decision determines the trigger behaviour in the automatically populated helpdesk ticket. The agent is then free to move on to the next item.

If the decision is ‘accepted’ or ‘published open access’, the item is deposited and the submitter is automatically notified via email. For submitted, proof, and non-OA published versions, the author receives an automatic email asking for the accepted manuscript. Items are archived in the repository under a generic collection, and any forthcoming publication details are added to the record via external source information in Elements.

To see just how efficient Fast Track is we’ve prepared a short demonstration video which captures some of the key features:

Video 1. Real-time demonstration of the Fast Track system.

Makes us stronger

Agents therefore need only make one decision: identify the file version. But the real ingenuity of the Fast Track system is that embargoes can be set automatically by:

  1. Taking into account the decision made by the agent (e.g. no embargo if published open access);
  2. Detecting publication status and publication dates from Elements; and
  3. Retrieving journals’ embargo policies via Orpheus (you can learn more about Orpheus in our previous blog post).

In some cases, usually because we don’t know the publication date, we can’t determine the embargo length of an accepted manuscript. In such cases we apply a 36 month embargo from the date of the Fast Track decision. We know that this embargo won’t always be correct, however, we routinely check manuscripts in Apollo and update embargoes accordingly.

Figure 3. Simplified overview of the Fast Track process. The key decision is to determine the type of manuscript that has been submitted. Everything else is handled automatically.

Since launching Fast Track the average time to process a manuscript is 1-2 minutes. More than 8,000 items have been processed since launching the phase two Fast-Track interface. If items processed under the phase one effort are included, the number goes up to just over 14,000. And since a picture speaks a thousand words, Figure 4 below shows the effect produced by the new interface launched in September on our backlog of unprocessed submissions.

Figure 4. Historical change in the number of unprocessed open access manuscript submissions. The total number of outstanding manuscript submissions peaked at nearly 2,400 in September 2018. Immediately after launching the Fast Track website the backlog dropped dramatically and was completely eliminated by March 2019.

We will continue to develop Fast Track to further streamline our processing of manuscripts. We have already started to partner with librarians and administrators across the University to leverage the collective knowledge about open access which now exists within the University’s professional academic services.

Get in contact: If you are running a DSpace repository and would like to implement Fast Track to work alongside your existing workflows email us at support@repository.cam.ac.uk

Published 23 April 2019
Written by Dr Mélodie Garnier and Dr Arthur Smith
Creative Commons License

Blood: in short supply?

Two years ago (almost to the day) we called out Blood for their misleading open access options that they offered to Research Council and Charity Open Access Fund (COAF) authors. Unfortunately, little has changed since then:

Neither of these routes is sufficient to comply with either Research Councils’ or COAF’s open access policies which require that the accepted text be made available in PMC within 6 months of publication, or that the published paper is available immediately under a CC BY licence.

At the time, we called on Blood to change their offerings or we would advise Research Councils and COAF funded authors to publish elsewhere. And that’s exactly what’s happened:

Figure 1. All articles published in Blood since 2007 which acknowledge MRC, Wellcome, CRUK or BHF funding. Data obtained from Web of Science.

Over the last two years we’ve seen a dramatic decline in the number of papers being published in Blood by Medical Research Council (MRC), Wellcome Trust, Cancer Research UK (CRUK) and British Heart Foundation (BHF) researchers. The number of papers published in Blood that acknowledge these funders in now at its lowest point in over a decade.

It’s important to remember that the 23 papers published in Blood in 2017 are all non-compliant with the open access policies of Research Councils and COAF, and if these papers acknowledge Wellcome Trust funding then those researchers may also be at risk of losing 10% of their total grant. If you are funded by Research Councils or one of the COAF members, please consider publishing elsewhere. SHERPA/FACT confirms our assessment:

Sign the open letter

We’re still collecting signatures for our open letter to the editor of Blood in the hope that they’ll reconsider their open access options. Please join us by adding your name.

Cambridge Open Access spend 2013-2018

Since 2013, the Open Access Team has been helping Cambridge researchers, funded by Research Councils UK (RCUK) and the consortium of biomedical funders which make up the Charity Open Access Fund (COAF), to meet their Open Access obligations. Both RCUK (now part of UKRI) and COAF have Open Access policies which have a preference for ‘gold’, i.e. the published work should be Open Access immediately at the time of publication. Implementing these policies has come at a significant cost. In this time, Cambridge has been awarded just over £10 million from RCUK and COAF to implement their Open Access policies, and the Open Access Team has diligently used this funding to maximum effect.

Figure 1. Comparison of combined RCUK/COAF grant spend and available funds, April 2013 – March 2018.

Initially, expenditure was slow which allowed the Open Access Team to maintain a healthy balance that could guarantee funding for almost any paper which met a few basic requirements. However, since January 2016 expenditure has gradually been catching up on the available funds which has made funding decisions more difficult (specifically Open Access deals tied to multi-year publisher subscriptions). In the first three months of 2018 average monthly expenditure on the RCUK block grant alone exceeded £160,000. We are quickly reaching the point where expenditure will outstrip the available grants.

One technical change which has particularly affected our management of the block grants was RCUK’s decision last year to move away from a direct cash award (which could be rolled over year to year) to a more tightly managed research grant. In the past, carrying over underspend has given us some flexibility in the management of the RCUK funds, whereas the more restrictive style of research grant will mean that any underspend will need to be returned at the end of the grant period, while any overspend cannot be deferred into the next grant period. As we are now dealing with a fixed budget, the Open Access Team will need to ensure that expenditure is kept within the limits of the grant. This is difficult when we have no control over where or when our researchers publish.

Funding from COAF (which is also managed as though it is a research grant) has generally matched our total annual spend quite closely, but the strict grant management rules have caused some problems, especially in the transition period between one grant and another. However, unlike RCUK, the Wellcome Trust will provide supplementary funding in addition to the main COAF award if it is exhausted, and the other COAF partners have similar procedures in place to manage Open Access payments beyond the end of the grant.

Where does it all go?

Most of our expenditure (91%) goes on article processing charges (APCs), as perhaps one might expect, but the block grants are also used to support the staff of the Open Access Team (3%), helpdesk and repository systems (2%), page and colour charges (2%), and publisher memberships (1%) (where this results in a reduced APC). The majority of APCs we’ve paid go towards hybrid journals, which represent approximately 80% of total APC spend.

So let’s take a look at which publishers have received the most funds. We’ve tried to match as much of our raw financial information we have to specific papers, although some of our data is either incomplete or we can’t easily link a payment back to a specific article, particularly if we look back to 2013-2015 when our processes were still developing. Nonetheless, the average APC paid over the last 5 years was £2,291 (inc. 20% VAT), but as can be seen from Table 1, average APCs have been rising year on year at a rate of 7% p.a., significantly higher than inflation. Price increases at this rate are not sustainable in the long term – by 2022 we could be paying on average £3000 per article.

Table 1. Average APC by publication year of article (where known).

Year of publication Average APC paid (£)
2013  £1,794
2014  £1,935
2015  £2,044
2017  £2,187
2018  £2,336

Elsevier has been by far the largest recipient of block grant funds, receiving 29.4% of all APC expenditure from the RCUK and COAF awards (over £2.5 million), though only accounting for 25.5% of articles. In the same time SpringerNature also received in excess of £1 million (which as we’ll see below has mostly been spent on two titles). With such a substantial set of data we can now begin to explore the relative value that each publisher offers. Take for example Taylor & Francis (£107,778 for 120 articles) compared to Wolters Kluwer (£119,551 for 35 articles). Both publishers operate mostly hybrid OA journals and yet the relative value is significantly different. What is so fundamentally different between publishers that such extreme examples as this should exist?

Table 2. Top 20 publishers by combined total RCUK/COAF APC spend 2013-2018.

Value of APCs paid Number of APCs paid Avg. APC paid
Publisher £ % N % £
Elsevier £2,559,736 29.4% 971 25.5% £2,636
SpringerNature £1,050,774 12.1% 402 10.6% £2,614
Wiley £808,847 9.3% 279 7.3% £2,899
American Chemical Society £411,027 4.7% 251 6.6% £1,638
Oxford University Press £379,647 4.4% 169 4.4% £2,246
PLOS £267,940 3.1% 168 4.4% £1,595
BioMed Central £245,006 2.8% 153 4.0% £1,601
Institute of Physics £189,434 2.2% 98 2.6% £1,933
Royal Society of Chemistry £156,018 1.8% 106 2.8% £1,472
BMJ Publishing £144,001 1.7% 68 1.8% £2,118
Company of Biologists £140,609 1.6% 50 1.3% £2,812
Wolters Kluwer £119,551 1.4% 35 0.9% £3,416
Taylor & Francis £107,778 1.2% 120 3.2% £898
Frontiers £103,011 1.2% 61 1.6% £1,689
Cambridge University Press £77,139 0.9% 38 1.0% £2,030
Royal Society £73,890 0.8% 52 1.4% £1,421
Society for Neuroscience £69,943 0.8% 26 0.7% £2,690
American Society for Microbiology £63,056 0.7% 36 0.9% £1,752
American Heart Association £53,696 0.6% 14 0.4% £3,835
Optical Society of America £39,463 0.5% 17 0.4% £2,321
All other articles £1,654,228 19.0% 690 18.1% £2,397
Grand Total £8,714,794 100.0% 3,804 100.0% £2,291

Next, journal level metrics. The most popular journal that we pay APCs for is Nature Communications, followed closely by Scientific Reports. Both of these are SpringerNature titles, and indeed these two titles make up the bulk of our total APC spend with SpringerNature. Yet these two journals represent significantly different approaches to Open Access. Nature Communications, along with Cell and Cell Reports, are some of the most expensive routes to making research publications Open Access, whereas Scientific Reports and PLOS One sit at the lower end of the spectrum. It is interesting that we haven’t seen a particularly popular Open Access journal fill the niche between Nature Communications and Scientific Reports.

Figure 2. APC number and total spend by journal. In the last five years, nearly £450,000 has been spent on articles published in Nature Communications.


Managing the future

While the OA block grants have kept pace with overall expenditure so far, continuing monthly expenditure of £160,000 would risk overspending on the RCUK grant for 2018/19. To counter this possible outcome the University has agreed a set of funding guidelines to manage the RCUK (from now on known as Research Councils) and COAF awards. For Research Councils’ funded papers the new guidelines place an emphasis on fully Open Access journals and hybrid journals where the publisher is taking a sustainable approach to managing the transition to Open Access. We’ve spent a lot of money over the last five years, yet it’s not clear that the influx of cash from RCUK and COAF has had any meaningful impact on the overall publishing landscape. Many publishers continue to reap huge windfalls via hybrid APCs, yet they are not serious about their commitment to Open Access.

In the future, we’ll be demanding better deals from publishers before we support payments to hybrid journals so that we can effect a faster transition to a fully Open Access world.

Published 22 October 2018
Written by Dr Arthur Smith
Creative Commons License