Scientific publishing is in a state of flux.
This is an effort to collate a dossier on publishing into a single document. A lot of the information comes from https://scholarlykitchen.sspnet.org/ (the official blog for the Society for Scholarly Publishing) which, as far as I’m aware, is not aligned with a particular publishing model, appears (to me) to be agnostic, generally cites sources inline, and is composed by independent writers. So I find it useful and recommend it. Other information comes from sources that I cite and discussions with publishers, fellow editors, colleagues, librarians and heads of funding agencies.
Some of the points below could be debated and I’m happy to stand corrected. I’m thankful to my colleagues who have provided feedback and corrections to earlier drafts.
Author Electronic Preprint: Is the author’s .pdf version of the submitted paper – i.e. the unreviewed and not copy-edited manuscript submitted for publication.
Author accepted manuscript (AAM): Is the author’s .pdf that contains the revisions arising through peer review and as accepted for publication in a scientific journal. The AAM has not been copy-edited or journal formatted.
Published (journal) article: Is the paper following copy-editing and formatting by the journal and with the journal masthead. The contents of the published manuscript are otherwise identical to the AAM. The journal manuscript is typically subject to a specific copyright.
Open access preprint repository: is a repository of electronic preprints approved for posting/publication after some form of moderation (usually basic screening and a check against plagiarism). arXiv (hosted by Cornell University) and bioRxiv (hosted by Cold Spring Harbour Laboratory) are the two most relevant to neuroimaging journals. From all sources, these upload ~ 10,000 and 1,000 new papers per month respectively. bioRxiv only accepts preprints prior to acceptance, but allows authors to update versions (implying that authors can ensure the last version is the AAM). Both have restrictions on the type of article they post – e.g. bioRxiv will not post review manuscripts. They are free to authors and readers although there are some costs associated with their operation (e.g. including an annual license to assign DOI’s) which are covered pro bono by the host.
Section 2.2 covers which journals allow posting and updating of electronic preprints.
Article Processing Charge (APCs): Are those costs that the contributing author (or their host Institute or funder) is invoiced for by a publisher at the time of acceptance, so that the journal article can be deposited in PubMed Central (PMC) at the time of publication and be made available under the Creative Commons Attribution (CC-BY) License.
1.2. Academic Publishers: Publish scientific journals and monographs
Commercial publishers: Are for-profit academic publishers, dominated by five for-profit companies (Elsevier, Springer-Nature, Wiley-Blackwell, Taylor & Francis, and Sage). Note that commercial publishers typically also publish non-academic material such as novels. Profits are split between building/enhancing the publishing business and distribution to share-holders.
Scholarly and Library publishers: Is a scholarly society (such as the Society for Neuroscience and the American Association for the Advancement of Science (AAAS; publishers of Science)) which have a membership revenue stream, or a University library (such as MIT Press, OUP and CUP) that publishes academic manuscripts and monographs. Note that many societies outsource their society publications to commercial publishers (such as the Society for Biological Psychiatry). Others publish themselves, such as the Society for Neuroscience.
Most scholarly and library publishers are not commercial entities, although they may make profits from publishing journals which they shift back to their capital reserves and to their host Institute or society. For example, OUP made a profit of 74.8 million pounds for the year ending March 2016 and distributed 30% back to the University.
Not-for-profit academic publisher: Includes PLoS and eLife. Not-for-profit publishers have been typically launched via grants (PLoS via a $10 million grant and eLife via an initial US$25 million in 2012 and a further $US$38 million in 2017) but aim for long term financial viability and independence from grants or donations.
Note that, as with academic societies not-for-profit publishers generally must strive for an annual “profit” or surplus operating revenue. This is for two reasons, first to re-invest in its activities and marketing, and second, to build cash reserves for a “rainy day”. For example, PLoS has in the past made up to 10% p.a. and built cash reserves of $30million. More recently, PLoS made a $1.7 million loss in 2016.
2. Journal financial models:
2.1. Subscription (toll) journal: Charge a subscription fee to individuals or entities to access/read, either for a journal subscription or per article. Authors do not pay an APC to publish in traditional subscription journals.
When in traditional paper form, all journals were subscription-based – i.e. libraries or institutes paid a subscription to the publisher who then shipped them the (paper) journal. Those not able to physically access those libraries typically requested paper reprint requests by (snail) mail to the author. Following the transition to electronic preprints and papers over the last two decades, the subscription-based model survived by restricting electronic downloads of the journal paper to those registered to these same libraries/institutes.
2.2. Green Open Access (Green OA): Authors are free to share their author preprint “anywhere at any time” and can update their preprints upon acceptance with the AAM (i.e. the revised, as accepted author manuscript). This is true for example at Neuroimage,
Note that with some repositories (e.g. bioRxiv), the original preprint must be uploaded prior to journal publication, although the version can be updated following each revision.
Green OA papers are usually published in subscription journals which have a Green OA policy. The copyright for the paper is usually a non-commercial Creative Commons (“CC-BY-NC-ND”) license. The journal article usually remains behind the paywall for an embargo period of 6-12 months.
2.3. Gold Open Access (Gold OA): At the time of acceptance, the author or funder is invoiced for an APC (see above) so that the journal article can be deposited in PubMed Central (PMC) at the time of publication and be made available under a Creative Commons Attribution (“CC-BY”) License.
- An “OA Journal” is one that only publishes Gold OA papers, i.e. does not have a subscription component.
- A “hybrid journal” (e.g., NeuroImage, Human Brain Mapping) is typically a subscription journal with an added a gold OA option, i.e. an opt-in APC for authors whom wish to make their article gold OA.
Before we move on to “diamond open access”, a few brief notes on the APC.
The cost of “running a journal”: Journals do (or are supposed to do) quite a lot more than managing the peer review process and type-setting/copy editing an accepted manuscript: Not all of these are necessary and below I’ll outline the very minimal cost of new publishing models. In fact, here is a list of “102 things that a good journal does”:
Some of these can be debated, but in my experience, scientists are not always great editors, and when it comes to issues like plagiarism, author conflict, fraud etc. do not know much about publishing law or ethics (why should we?): I’ve handled papers at journals where such disputes rage for weeks while senior editors are switched off and there is no accountable publisher. At a very busy journal, disputes happen frequently enough and there are internal processes to swiftly and appropriately respond. Journals where I have edited thus have a journal manager, a special issue journal manager, annual meeting editors meetings, cover artists, a publisher, access to legal and ethics teams, copy-editing etc. (but see notes on this in Section 4.2).
Another insight into the cost of publishing can be seen through eLife’s blog when they started their US$2500 APC: https://elifesciences.org/inside-elife/b6365b76/setting-a-fee-for-publication
Note that eLife partition quite a lot of money toward editorial stipends. But from being a board reviewing editor (BRE) at eLife, I can underline that there is a lot of work over and above soliciting reviews.
Costs and rejection rates: If the entire cost of running the journal comes from the APC’s (i.e. Gold OA journals), then these necessarily go up (all else being the same) when the rejection rate goes up.
Red Curve = 850 + 350*(1/Acceptance Rate))
Note that this is not a comprehensive dataset but appears to be illustrative. For a thorough (but slightly dated) report see here; or for a very thorough report (but also dated), see here.
This is one of the reasons that the APC for Nature Communications (8% acceptance rate; US$5200) is much higher than Scientific Reports (59% acceptance rate : US$1760, both Springer-Nature). Additional factors include the use of professional editors at some journals (Cell, Science, Nature etc). Note that F1000 (and Wellcome Open Access, which “runs off” F1000), essentially accept all papers (they are accept-and-then-review models) and charge a US$1000 publication fee, essentially a little below the red curve. But they run off an author led accept and review process and hence essentially do not have editors.
In contrast, the editor-in-chief of Nature estimated the cost of a Nature or Science paper in 2012 to be in excess of $US10,000
It should also be noted that this is a contestable area for a number of reasons: 1. There is a difference in the cost-to-publish and the actual invoiced charge because of market forces. While PLoS, BMC and Science Advances (AAAS) are not-for-profit, Nature Communications (Springer-Nature) is for-profit and may add a greater profit margin; 2. The actual data behind these estimates are often commercial-in-confidence and therefore hidden (although see the eLife costs); and 3. The continual innovations in publishing platforms may likely drive down the actual costs per article.
Hybrid versus gold OA APCs: I think it’s reasonable to assume that the APC’s in hybrid journals are generally lower than the APC’s in Gold OA journals because they are offset by the subscription fees they still attract. As an example, the APC for Gold OA in NeuroImage is US$3000 versus US$5000 for the gold OA-only Cell Reports (both Elsevier journals, with comparable Impact factors and H-Indices). In fact, Elsevier explicitly state that they set their APC’s partly based on “other revenue streams associated with the journal”;
Similarly, PLoS use revenue from PLoS ONE to subsidise PLoS Biology and PLoS Medicine (higher APC’s but professional editors and much higher quality and rejection rates, see figure above).
One might also read:
One of the most common misconceptions I encounter is the conflation of an “Open Access Journal” as being a “Not-for-profit” journal. This is definitely not the case. Commercial publishers now host far more Gold OA journals than Library and not-for-profit publishers and are able to make profits that way (usually by keeping rejection rates, copy-editing and quality control low). In terms of mega-journals, Springer-Nature’s Scientific Reports is larger and more profitable than PLoS ONE:
You can read some of the possible reasons here:
Note also that many society and not-for-profit journals operate within a subscription model (such as AAAS/Science).
The other most common misconception about “open access journals” is the conflation with other initiatives of open science, including “open review”, review-after-acceptance, data and code sharing etc. All of these are really orthogonal to the publication financial model. I’ve compiled a (shorter!) dossier on peer-reviewing models and will publish that separately.
2.4. Diamond Open Access (Diamond OA): endeavor to have zero or near zero APCs and no subscription fees.
There are very few scientists (?any?) who are not staunch advocates of making their research papers freely available – not only is this socially desirable, but *may* increase a paper’s influence (but see https://www.mitpressjournals.org/doi/10.1162/REST_a_00437#.V865hfkrLIU).
However, some hold concerns about Gold OA: namely that it imposes publication costs on individual laboratories in those regions where funders do not pick up the APC. Even in those regions where funders do allow APC’s to be taken from funded grants (such as in Australia) these must be deducted from existing grants – i.e money for post-docs and experiments. Gold APC’s can also shift publishing costs between different sectors of national economies – e.g. from the Education sector (a substantial revenue raising sector in many regions) onto the Health (the NIH) or Research (the MRC) sectors, without necessarily reducing the overall costs or profits of publishing;
A second concern is the relationship to so-called predatory publishers, i.e. those where authors pay to publish in the absence of tight peer review controls. But others have argued that this risk is over-stated:
At the core of the concern with both subscriptions and APC’s is the argument that publishing an AAM on an archive is very cheap, that peer review is free, that editors are largely voluntary/poorly paid (except in the case of professional editors at Science, Nature etc) and that APC’s should hence be much cheaper than the red curve above; APC = 850 + 350*(1/Acceptance Rate)).
Diamond OA journals rely upon cheap manuscript hosting services and very basic cross-referencing. They also rely heavily upon a large team of volunteers to run papers through the system, act as editors and undertake copy-editing. It’s not clear that a journal can be run as a diamond OA from scratch, although it may be possible for a society journal (?OHBM’s platform, Aperture) to be Diamond OA.
An example of a diamond OA journal, Volcanica:
Another example is The Journal of Machine Learning Research:
The latter is reasonably well established, is a relatively modest throughput journal (~120 papers p.a.) and ranks 19th in its field by impact factor.
Finally, there are also “overlay journals” such as Discrete Mathematics and Theoretical Computer Sciences (https://dmtcs.episciences.org/) that are overlay journals, that is they do not produce their own content, but select from texts that are already freely available online.
There are no existing large-volume Diamond OA or overlay journals.
In case you are confused, here is a helpful schema,
3. Plan S, Plan T and all of that.
Okay, so why all the concern about various open access models?
3.1 Plan S: Proponents of OA publishing have become frustrated at the relatively slow movement to Gold OA and have launched Plan S, “By 2020 scientific publications that result from research funded by public grants provided by participating national and European research councils and funding bodies, must be published in compliant Open Access Journals or on compliant Open Access Platforms.”
Plan S lists a number of copyright and other items that are generally consistent with a gold OA paper in a hybrid paper. However, Plan S also explicitly states that “The ‘hybrid’ model of publishing is not compliant with the above principles.” Note that Plan S is much stricter than simply not covering APC’s in hybrid journals, but also states, “The Funders will monitor compliance and sanction non-compliance.” (e.g. with-hold funding).
Plan S is arguably deliberately disruptive – not just an endeavor to ensure papers are gold OA, but to challenge the subscription model of publication. Plan S now covers many major European funding agencies:
Plan S is not without its distractors – not only commercial publishers, but also society and not-for-profit publishers such as the AAAS, which said that Plan S “will not support high-quality peer-review, research publication and dissemination,” would “be a disservice to researchers” and “would also be unsustainable for the Science family of journals”
A recent open letter critiquing Plan S (currently with >1400 signatories) lists concerns about the complete ban on hybrid journals, global unevenness, the shift of publishing costs to laboratories. perceived restrictions on academic freedom and the over-looking of repositories permitted with existing green OA:
While Plan S has been embraced (as a plan, by many major funders) in Europe, other regions such as the USA/NIH, China, Canada and Australia have not signed on, but rather have existing policies that are consistent with hybrid and green OA:
Why doesn’t Plan S permit gold OA papers in hybrid journals? I cover this below in Section 3.3
3.2. Plan T: Aim to cover OA journal costs through a mixture of submission and APC fees.
As noted above, if a journal self-funds from APCs, those have to cover all other papers/costs. In the case of e.g. F1000, Wellcome Open Science, this is less of a problem because “all” papers are accepted. As rejection rates go up, the APC from accepted papers must go up in proportion and this is compounded by highly selective journals having higher quality proofing, copy-editing, editorial checks and full time professional editors. The argument for Plan T is to distribute costs amongst submitting and publishing papers. If the submission fee is sufficiently high (e.g. US$350), then the final APC can be kept approximately constant across journals (e.g. US$1200) – those with high rejection rates still get more revenue per published paper.
I’ve even heard it said, by a senior person within the open science community, that there could be three tiers – a submission fee, a fee for peer review and a final APC fee – an ultimate users-pay system.
An advantage of Plan T is that it might/should encourage people to be more circumspect about submitting papers. Most journals where I have edited, receive a reasonable fraction of papers that are clearly not in scope (mainly too clinical).
There are clear disadvantages here. For Plan T to work, the submission fee has to be non-refundable. Editors already make people pretty upset when they triage reject their papers – to charge them for the “service” would likely cause considerable push-back. A substantial proportion of our decisions are “reject with option to resubmit” and then the question of further submission fees arise.
In my experience, few people agree with decisions to reject their paper following peer review. To charge more for a peer review rejection would stoke these concerns so I can’t imagine that working seamlessly.
3.3 Publish and Read models (Project DEAL):
Hybrid journals gather revenue from a mixture of subscription and APC fees. One is supposed to offset the other (in theory, keeping APCs lower than in Gold OA journals and in theory also taking the heat off subscription fees for hybrid journals). This seems fine when contributors also come from a mix of regions – some have APCs paid for them, some are from subscription regions and hence publish and read for free. But if a region is funding all of its researchers to pay APCs, then it currently still has to pay the same subscription fee, in order for its research community to access papers published outside its jurisdiction.
Read and publish models are supposed to fill this gap – they work via a single agreement that combines a reduced subscription fee and a fixed and reduced APC for all published papers. This means that researchers within DEAL can publish all their papers gold OA and also read all papers in those same journals.
I personally think this is a reasonable interim solution as scientific publishing inevitably moves away from subscription-based funding (http://www.michaeleisen.org/blog/?p=1710), allowing all scientists to publish at reasonable rates and have full access to all papers in journals where they publish. It also takes into account the substantial international heterogeneity in the rate at which the OA transition is occurring. Moreover, this has been achieved between the Association of Universities in the Netherlands (VSNU) and several major commercial and society publishers (Wiley, Sage, and the American Chemical Society)
Other advantages include, permitting funders to cap APC’s, permitting the distribution of publication costs across Institutes, laboratories, funders, and the educational sector. Maybe this is also a challenge in some areas, as funders and universities/libraries also have to agree how to split the costs.
I believe not all commercial publishers like this model because future revenue is not locked down at the contract, but also depends on how many papers are actually published (and hence how much revenue comes from the APC’s).
When I started editorial work, I barely too notice of the financial model of publishing. More recently its become a foreground issue. As in science generally, I hope a solution can be found through constructive and collegial exchange.
All the best, Michael Breakspear