This is a rather speculative thought piece.
Where we want to get to is authors doing the minimum amount of work necessary to get their papers published, but not less work than that. Two recent developments suggest that the time might be right to think again about this: the first is the announcement by PLOS that they were shutting down development of their new workflow system Aperta; the second is the Manuscript Exchange Common Approach initiative that I noted in a comment following recent post here.
The main lesson from PLOS’s Aperta decision is that each journal editorial office has different workflows, making it next to impossible for any one system to accommodate them all without becoming complicated, and therefore impossible to give much help to authors in handling those different workflows. The argument is that these different workflows support the different editorial and peer review practices of different journals. There is clearly value in such variation, to differentiate different journals with different missions, but I would be interested in any empirical evidence showing how much researchers (as readers, reviewers and authors) rate the value of this variation against the value of their time in navigating the resulting landscape. Part of that calculation is the amount of time authors devote to using manuscript submission systems, even if the task is often delegated to less senior staff. To see the order of magnitude, suppose each submission takes two hours, and each article is submitted on average 1.5 times (ie, on average, half of articles are rejected by the first journal, and submitted to and accepted by a second journal) – then in the UK, publishing around 150,000 articles per year, this amounts to 450,000 hours that could be spent doing research or teaching. More bluntly, assuming these were staff on £30 per hour (ie, not professors), that is over £13m.
We do have some evidence of the amount of time taken to deposit papers into repositories, and with some caution this might perhaps be extrapolated to cover preprint services too. In 2014 Research Consulting produced a report estimating 48 minutes per deposit, or a total of 120,000 hours annually, not all of which is that of authors. Clearly services like the Publications Router are intended to reduce this by passing at least the metadata, and usually the full text, to the repository. Nevertheless, it is perfectly possible to imagine an author putting their paper into a preprint service, submitting it to at least two different journals, and depositing it into their institutional repository, re-keying data at least four times – even before any reporting on that publication to funders. Can we do better?
Well, as I hint in the opening paragraph, yes I think we can, and part of that could be the Manuscript Exchange Common Approach (MECA) initiative. What if the first place to which an author turned – increasingly likely to be a preprint service – started the ball rolling by wrapping the fairly minimal amount of metadata, plus the preprint itself, into a MECA package? With the authors’ and journal’s permissions (and a variety of models could be imagined for granting those permissions), that package could be imported into, and part-populate, the manuscript submission system. You could imagine a protocol like SWORD being useful for this. It is certain that further information would be needed from the author at this point, but that would supplement rather than duplicating information they had already provided. Similarly, if the article needed to be cascaded, or resubmitted, or deposited in some form into a repository.
Of course, there might be all kinds of business reasons why this would be difficult but, since our business is helping researchers, perhaps those could be overcome. Eventually, of course, you might want CrossRef and other identifier services to play a role in some of this, though the exact nature of that role might be debated.
That could certainly ease things for the author. But could we go further, and help more even with that original submission to the preprint services? What if authors were able to call on a tool that pulled data from a range of local and remote systems to create a MECA package even for the initial (eg preprint) deposit?
That’s a long way off, and many things may change in the world of scholarly communication. However, some of those things, for example the vision of “next generation repositories” with their resource-centric approach, don’t seem incompatible with a MECA package and SWORD automatic submission and deposit. Having said that, the view of repositories above, at the end of the line with no direct engagement from authors, might not command wide support in universities, and so this speculation is at best only part of the picture.