Perhaps one of the hardest parts of Open Access for researcher (and library staff!) to grasp is which version of a work they should share via Open Access. It's likely that there will be several versions of an article or other work before it is even sent to a publisher and then it often has to go through peer review and revisions before it is actually accepted. Although researchers will want to share the final, shiny published version which looks like a journal article there are often restrictions on what they can share.
The graphic below describes the different versions of a manuscript that can exist as part of the publication process.
Depending on the open route selected and funder mandates researchers may have to think carefully about which version they should share. When dealing with Gold Open Access materials the choice of version is relatively simple as the final published version is made available by default but with the Green option this can be more confusing. Each publisher is likely to have restrictions on which version(s) they allow a researcher to share via a repository and this should be clearly outlined both on their website and in any publication agreement.
The version to be shared is typically something known as the author's accepted manuscript (AAM). This is the version of an output which has been through peer review, accepted by the publisher but not yet published. The text of the AAM will be identical to the finished version but will look different - it is often a PDF of the text as an author would see during preparation rather than a version which looks like a published article. However it is not always straightforward to determine which version is which, as demonstrated in the Learn More box on the left.
Given that the choice of manuscript can be complicated a common question from researchers is why does any of this matter? As long as the work is openly available why is it a problem? The answer is due to copyright restrictions. A typical publication timeline showing different versions of a paper is below:
As this graphic illustrates, different versions of an output are created at different points in the submission process. The submitted manuscript (the first draft a publisher sees) may go through several rounds of peer review before the content and format is deemed suitable for publication. At the point where a title agrees to publish a work it becomes the author's accepted manuscript. At this point a publisher will usually ask the researcher to sign something known as a publication or copyright transfer agreement. At this point the publisher will invest time and money in editing, typesetting and generally making the output look like a fully finished output. This may include putting it into a certain font and format, assigning a volume and issue and adding the logo of the title. The proofs are sent back to the researcher for checking and then the final, published version of record is produced.
As outlined above, most publishers specify that the author's accepted manuscript is the version which should be shared in an open repository. This is because:
Once a publisher starts preparing a manuscript for publication and putting it into their house style they want to protect this investment which is why they often prevent the sharing of the final stylised output. They accept (sometimes grudgingly) that sharing outputs openly is mandated by research funders and that without these funders they would not have outputs for their publications but they often place restrictions on which version can be shared.
This can all be extremely complex and researchers should always check their publication agreements carefully for full details of what they are allowed to share and where. A handy guide can be found in the SHERPA/RoMEO database which outlines publisher copyright and self-archving policies.