arXiv and REF – together at last?

New draft REF2021 guidance was released for consultation on Monday morning. Buried half-way through this daunting 139 page document was an update to the REF Open Access policy.

This revised policy comes on the back of Research England’s report Monitoring sector progress towards compliance with funder open access policies which was released in June, and on which we have already commented.

From an Open Access perspective, additional flexibility for preprint servers has been added to the policy:

The funding bodies recognise that many researchers derive value from sharing early versions of papers using a pre-print service. Institutions may submit pre-prints as eligible outputs to REF 2021 (see Annex K). Only outputs which have been ‘accepted for publication’ (such as a journal article or conference contribution with an ISSN) are within scope of the REF 2021 open access policy. To take into account that the policy intent for ‘open access’ is met where a pre-print version is the same as the author accepted manuscript, we have introduced additional flexibility into the open access requirement: if the ‘accepted for publication’ text, or near final version, is available on the pre-print service, and the output upload date of the pre-print is prior to the date of output publication, this will be considered as compliant with the open access criteria (deposit, discovery, and access).

That’s a significant adjustment to previous advice and will be of considerable relief to many researchers who routinely publish their research in this way. Indeed, we have lobbied behind the scenes on this policy issue for more than three years.

But what does this actually mean and what should institutions and authors take from this?

Repositories, preprint servers – what’s the difference?

Firstly, this policy legitimises preprint servers (like arXiv, bioRxiv, SocArXiv and many more) and allows authors to use these systems without needing to worry about technical requirements.

This is in stark contrast to the way institutional and subject repositories are treated by the policy.  These repositories must meet all the requirements of the REF Open Access policy to be considered compliant, which is fine for most institutions because meeting the policy requirements is vital, but subject repositories are usually left in the lurch:

Individuals depositing their outputs in a subject repository are advised to ensure that their chosen repository meets the requirements set out at paragraphs 224 to 241 in this policy. REF 2021 guidance will not certify the repositories which fulfil policy requirements.

We’re still not sure if Europe PMC is compliant, for example.

Don’t just sit there!

However, just because preprint servers are okay, doesn’t mean that authors using preprint servers should assume they don’t need to do anything. There are two significant caveats to take note of:

  1. the manuscript deposited in the preprint server must be the “‘accepted for publication’ text”; and
  2. the manuscript must be uploaded prior to first publication.

Determining the deposit time is usually straightforward, so institutions will be able to monitor this aspect of the policy with some level of automation (especially for arXiv which is harvested by a range of publication systems).

However, the key challenge will be determining the manuscript version. We’ve previously described the work we do as manuscript detectives, so some level of checking with authors will still need to take place.

We are working internally at Cambridge on what our workflow will be to capture these outputs and we will be talking to our researchers on what they need to do or not once this is determined. We still encourage all of our researchers to upload manuscripts when accepted for publication until we indicate otherwise.


If there is one key recommendation we would make to all users of preprint repositories – annotate or label the records to clearly indicate the manuscript version (e.g. submitted, accepted, published).

It will help us, and you, in the long run.

Published 25 July 2018
Written by Dr Arthur Smith
Creative Commons License

2 thoughts on “arXiv and REF – together at last?

  1. Hi Arthur,

    I’m not sure if you’ve seen it, but there is a recent blog post regarding whether or not Europe PMC is an acceptable subject repository for REF2021 (the answer is yes–it is a very good way to comply with the REF):

    I don’t think any of the language regarding repositories has changed in the draft for REF2021 since EPMC wrote this post, so could you please expand on your uncertainty?

    1. Hi Diego

      As it happens, we met with the Europe PMC team just before that blog post was published. As I understand it there are three routes for material to be deposited in Europe PMC.

      1. Directly from publishers via PMC (which flows through to Europe PMC).
      2. Bulk uploads from publishers into Europe PMC (mostly NPG journals, if I recall correctly).
      3. Authors uploading papers to Europe PMC themselves.

      I’m happy to be corrected, but the Europe PMC team could only give deposit dates for (3). That’s also confirmed in their blog post. If the deposit date for all full-text records could be exposed in the Europe PMC metadata that would make all the difference.


Leave a Reply

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

This site uses Akismet to reduce spam. Learn how your comment data is processed.