Overlay text
Image courtesy of Dora Mitsonia

We spend much time these days wondering when the academic journal as we know it will cease to exist. Opinions on the future of journals vary widely. There are those who say it will live forever and others who see the journal as an ugly reminder of the sins of big publishers – exploitative vehicles for dragging a profit from those who can’t afford to pay.

Perhaps it is worth taking a step back and asking ourselves why the journal exists in the first place.

What are the hard-to-replicate functions of a scholarly journal in our connected age? Here I refer back to Michael Clarke’s excellent 2010 post in the Scholarly Kitchen entitled “Why Hasn’t Scholarly Publishing Been Disrupted Already?” Let’s start with “validation”. For a journal article to be considered seriously – and thus the journal be taken seriously – there must be an acceptable process of quality control. This means peer review. Then there is “designation”. Clarke says that “…perhaps the hardest (function) to replicate through other means, is that of designation. By this I mean that many academic institutions (and other research organizations) rely, to a not insignificant degree, on a scientists’ publication record in career advancement decisions. Moreover, a scientists’ publication record factors into award decisions by research funding organizations. Career advancement and funding prospects are directly related to the prestige of the journals in which a scientist publishes.”

“I know all this”, you might say, so why bring it up now?

I was thinking this through in the face of Tim Gowers’ recent announcement of a new type of journal – the overlay journal. What is an overlay journal? It is really the word “overlay” that matters. Whether we are talking about a journal is arguable, and I discuss this a little later. The basic idea is that an overlay journal does not produce its own content. Rather, it links to already available content, most likely residing in a preprint server. One such server is arXiv, a preprint server in mathematics, physics, astronomy, some computer science, and other related fields. Anyone may submit a preprint of their articles to arXiv, though there are moderators who ensure some level of relevance. In addition, for a preprint to be listed in a particular category it must be endorsed by another arXiv user with a preprint already on the server. The whole thing is funded and run by Cornell University with further funding from the Simons Foundation and fees paid from member institutions. It is worth mentioning that we, at the American Mathematical Society, encourage our authors to post their articles in arXiv.

The new overlay journal, Discrete Analysis, is not intended to be “publishing” in the sense of publishing original articles in a journal. Instead, for those articles deemed to be worthy, preprints will be peer reviewed (it is not clear what model of peer review will be used – but to some extent this is irrelevant), and then links will be provided to arXiv. There will also be summaries/abstracts that one may use to point to the preprints themselves. The result will be a publication that is in some parts niche and some parts generalist, focusing on articles giving “…a good description of many parts of mathematics where the structures being studied are discrete, but the tools are analytical in character…” Gowers, in collaboration with Terry Tao, will ensure that this is a quality venture. The people involved are stellar – and they make up the editorial board under the leadership of Gowers, as he puts it “for the time being”. Discrete Analysis is a very interesting experiment which deserves the attention of mathematicians and publishing and librarian communities.

So, as I say, I am mulling over whether this interesting venture may be called a journal, or not. As of writing this post I have not been able to pin Tim Gowers down for a conversation – it would be good to hear from him directly of course on his views for the future of Discrete Analysis.

My view is that while this is a fascinating way to draw out from arXiv links to good preprints in relevant fields, this is not journal publishing. In Gower’s blog he moves on from talking about the idea of the overlay journal to a more polemical discussion of how his venture is in essence the future of the journal, reducing costs and supplying quality content in a way that may be used in the same way journal articles are used now. While Gowers has every right to his views on this, I would argue that, while his is certainly an exciting way to make use of preprints in arXiv, what it does is quite distinct from a journal. As discussed above, the journal is a matter of record, and like it or not, journals form a part of the academic and recognition workflow that allows for career progress, grant making, more research and more articles to be published.

Moreover, the form of publication is significant. In fact, while there is no question that many publishers are facing up to a new age where journals may not represent the kind of profits once taken for granted, this does not mean that there are no costs associated with producing a journal of record. In fact, if one assumes that the journal contains the final published paper that has not just gone through peer review, but been worked on significantly by the publisher themselves, then one realizes how many resources are plowed into a journal. A publisher adds all manner of value in the creation of a journal, from peer review management, production and perhaps most intensive these days, the features that allow for electronic consumption by libraries and end users, and the intensive work that is put in to ensure survival of the journal in perpetuity. Academics take for granted the features that allow them to, for example, view certain types of supplementary material, or find articles that cite the one they are reading. The assembly of these components is also not a linear value chain. In some ways a journal is a magical product in which the whole is greater than the sum of the parts. What a journal represents to the community is more than just the pages themselves. How else over the span of many years can you keep up with the corpus of human endeavor?

Another interesting set of data points to consider may be seen in the work of Larivière et al in their 2013 paper, arXiv e-prints and the journal of record: An analysis of roles and relationships1, which notes that only 45% of preprints on arXiv appear in Web of Science indexed journals. In other words, only 30% of Web of Science publisher papers appear in arXiv. The take-away from this is that there is plenty of quality mathematics research that does not appear in arXiv that will need to find a home in a published journal. Interestingly, a version of this paper with no DOI attached sits in arXiv at http://arxiv.org/abs/1306.3261.

Perhaps here I should provide an extra disclaimer. I am deeply interested in the future of math publishing in my role leading the publishing enterprise at The American Mathematical Society. Indeed we are experimenting with new technologies, new business models and new publishing approaches. The job as I see it is to serve the mission of the society in a way that allows us to provide operating income for an organization that in turn benefits the global community of mathematicians through its programs, prizes, meetings and other services. I am both mission driven and business focused. This means that I will look to adopt new ways of doing publishing, but it has to make sense in context of the overall mission. In my view, journals are still very much a key piece of this business.

  1. Larivière, V., Sugimoto, C. R., Macaluso, B., Milojević, S., Cronin, B. and Thelwall, M. (2014), arXiv E-prints and the journal of record: An analysis of roles and relationships. Journal of the Association for Information Science and Technology, 65: 1157–1169. doi: 10.1002/asi.23044
Robert Harington

Robert Harington

Robert Harington is Chief Publishing Officer at the American Mathematical Society (AMS). Robert has the overall responsibility for publishing at the AMS, including books, journals and electronic products.


28 Thoughts on "What is an Academic Journal?"

I am curious as to why if peer-reviewed, and articles have DOIs (so permanency) why this journal os in some manner ‘distinct from a journal’?

1. There’s no editor for the “overlay.”
2. DOIs aren’t that permanent.

There is an editorial board – and it looks to me as if Tim Gowers is editor.
True, but the journal is using scholastica and there will be back-ups, so as permanent as any other journal (and fear of publishers failing is after all, why we subscribe to services such as LOCKSS and Portico…)

I would tend to agree with you. Authors submit their articles to arXiv first and then submit them to the publication for review. I don’t immediately see the distinction between what Discrete Mathematics is doing and what a traditional academic publisher does when it comes to being a steward for scholarship.

Am I convinced this is the future of academic publishing, though? No.

On the subject of DOIs – the two links to the article cited in the post – “arXiv e-prints and the journal of record: An analysis of roles and relationships1” should actually use the DOI as the link -http://dx.doi.org/10.1002/asi.23044 – that’s the point!

While this overlay journal seems to require authors submit their preprints for review, it does raise some tricky questions about overlay journals and “previous publication”. What happens if someone starts a similar journal but instead doesn’t ask for author consent, and just picks the best articles off of arXiv and the like, performs peer review and “publishes” them in their new journal? For those preprints posted with a CC BY license, the new journal could go so far as to copyedit and typeset those articles and present them in issues (and sell subscriptions). What happens to the authors of those preprints when they try to submit the article to a journal that is best suited to reach their community audience, or that best serves their career advancement needs? Will the traditional journal declare the article previously published and reject it?

We are definitely into the conceptual confusion gray area with this one. If there is author submission then that is arguably closer to journal publication than if there is not. If a publication simply picks good preprints then that is arguably more like a preprint aggregator.

Then there is the question of the peer review. If it is limited to selecting preprints for publication, then that too looks like an aggregator. But if there are substantive reviews, sent to the authors, followed by revisions then that really looks like a journal.

Conspicuously absent from your list of what value is added by publishers is copyediting. Is that because less copyediting is done now than earlier, or not done at all? I also wonder if you consider a journal properly peer reviewed if only the editor (or a member of the editorial board) does the review and no external experts are consulted? I know of several such journals.

If there was ever a time when copyediting added value, that time is long gone. Journal copyediting (in my experience, and when I discuss it with colleagues), runs from benign to harmful. Any journal that could reduce pages costs, by so much as a farthing, by no longer copyediting should do it. Journals offer two services to scientists: arranging peer review, and “certifying” something as legit. Perhaps in some fields they still publicise work, though in my field that’s entirely replaced by the arXiv. An arXiv overlay journal that organised peer review, and was able to apply a “stamp of legitimateness” to a paper would offer me everything I want from a journal.

Anything else they’re doing is either because it benefits them, or because their customers (librarians and the like?) request it.

I might suggest that “me and my friends” is a poor sample for analysis. For many medical journals, care must be taken to ensure that things like dosage levels for suggested treatments are accurately stated. Increasingly, statistical editing is being performed to improve accuracy and reproducibility.

What does arXiv offer in terms of plagiarism detection and image manipulation analysis tools?

As an author, of course, I have no interest in plagarism detection nor image manipulation analysis. The arXiv does have an automated plagarism/overlap detector, though I don’t know how thorough it is. Those might be things the customers want from journals, but not authors.

I might suppose that in order to achieve credibility, any overlay journal would need to do an okay job of weeding out fakery to attain credibility – though any reader of retractionwatch.com would know journals don’t need to do a perfect job to be rather prestigious. Or perhaps even any at all? I’m not specifically aware of anything journals I publish/cite do to deter fakery, and I can’t find any statements by them on their websites. Beyond reviewers finding hijinks, if they’re doing it, they’re not advertising it.

I don’t have wide data, just a sample of perhaps fifty scientists I’ve discussed publishing from the author’s perspective with. I’d be interested to see data, if there is any (I searched briefly, couldn’t find any). But otherwise, it should be obvious publishers do things because authors want them, but also because it increases their profitability or makes their job easier, or because whoever’s buying journals wants them.

As an author…

This is really the key statement. Is the journal’s customer the author or the reader? For an open access journal, the answer is the former, for a subscription journal, the answer is the latter. I can assure you that readers value not having to read gibberish or plagiarized/fake research. The use of these tools varies greatly (Rockefeller University Press, for example has routinely examined every single image submitted to every single paper for anomalies, while other journals don’t bother unless attention is called to the article).

But on the subject of what authors want, there’s an interesting new survey from Nature that asked over 21,000 authors how they select journals to publish in, and what they value from publishers. Worth a read:

Yes, of course it’s a key statement. For a subscription journal, the readers are the customer, but the authors, not the publishers, are dictating where the customers are going.

It’s certainly true, looking over the author insights, I had said authors are looking for Peer Review and Journal Prestige, while the survey results certainly indicate authors are looking for journals to publicise their results/be read by people in their discipline (and to a lesser extent, not take forever on turnaround). I would suppose an arXiv overlay journal would automagically solve these problems (at least, I totally overlooked reaching your audience as a service of a journal as that’s no longer done by journals in my field, only by the arXiv and conferences/talks/what, but that’s field-specific)

Publishers also have a role in disseminating articles. However thoroughly peer-reviewed your article is, if no-one reads it there was not much point in publishing it. All publishers – subscription and open access – need to take into account the needs of readers and work with a range of indexing databases to make sure published papers are seen by the correct audience. There is also a huge overlap between readers and authors, and I think the majority of authors would look at a journal from both perspectives before publishing in it.

So, are you saying that scientists all write perfectly clear prose that needs no editing, or that all copyeditors are essentially incompetent? Since science journals are publishing more and more by non-English speakers, are you happy to accept their sometimes mangled language if the journal costs a farthing less?

Peer reviewers often force authors to clarify their meanings – certainly, I’ve had many manuscripts where the reviewers dramatically improved the clarity (and I’d like to think I’ve done so as a reviewer from time to time).

But the copyeditors at the journal? No, I’ve never seen them do anything helpful. Usually they’re totally benign, respelling words to their preferred variety of English, but because they’re not subject matter experts, when they change sentences around they occasionally muck up the meaning. That’s fairly rare, but a pain because you have to go through everything they’ve done to ensure they didn’t screw anything up, but they don’t do anything helpful. I’d much rather deal with a journal that didn’t have copyeditors. Layout editors probably mattered when physical journals still got read by people who didn’t have “emeritus” in their title. But now, there’s no journals should have to be mucking around with that sort of thing. Provide a LaTeX (& BibTeX) template, and stand back.

Thanks, really nice summary

” the journal is a matter of record, and like it or not, journals form a part of the academic and recognition workflow that allows for career progress, grant making, more research and more articles to be published.”

I’m not sure that by selecting work already on arXiv makes it less the version of record?

” A publisher adds all manner of value in the creation of a journal, from peer review management, production and perhaps most intensive these days, the features that allow for electronic consumption by libraries and end users, and the intensive work that is put in to ensure survival of the journal in perpetuity.”
Agree, though it seems in some cases arXiv is providing these, for others they are using scholastica, and perhaps for some they have decided they are not needed (though I will note arXiv has strands of work around preservation).

To me, and I am not expert, the key service of a journal is disseminating peer-reviewed academic research outputs, this seems to achieve this without ‘color printing charges’, APCs of journal subscripts and provide access to all. For that i think it is worthy of being call a journal, and most certainly a worth model to explore.

Incidentally, reminds me I attended an event about arXiv overlay journals back in 2008

The problem is that if this is a journal and if they publish your arXiv preprint, then you cannot then publish your polished article in another journal. Many relevant journals allow arXiv preprint publication and some actively encourage it, as Robert indicates. This concept threatens that arrangement. As I suggest above, maybe it depends on the nature of the peer review.

I’m not sure how this is a problem. If you publish in one journal now, other journals will reject your attempts to publish a second time. This wouldn’t be any different.

There is indeed no problem if this is a journal, but that is the question. They are publishing preprints. If these are published unmodified then that should not preclude submitting them to another journal, because preprints are allowed. That is what “preprint” means. The ambiguity is deep at this point.

As is outlined in the linked announcement, all of the articles published in Discrete Analysis will be refereed. The articles would not be unmodified in a substantive sense, so I’m not sure what distinction you are trying to draw here.

Does it actually say that the peer reviews will be returned to the authors and the articles changed accordingly? The alternative is that peer review merely addresses the suitability of the preprint. In the former case it is a journal, but in the latter it is prepublication. My impression is that this crucial distinction is unresolved, but I have not read the announcement. Can you quote it?

The announcement only says that peer review will be done in the “traditional way”, which I would certainly take to mean you’d add v2 (and probably v3, v4) in response to the referee reports. It’s not specific, but I suspect that’s only because it’s very unlikely to occur to them that someone might take them to mean anything else.

It’s pretty unambiguous that they intend a journal in largely the traditional sense, but with the services only needed for print journals stripped away.

The question then is how the time of the editor (who has to sift through multiple articles to select ones that are worthy of attention) is paid for? Ditto the time of the person who organizes and administers the reviews? If the editor is not doing this in a systematic way, isn’t there a danger that s/he favors articles by known individuals, thus introducing bias into the selection process?

It seems to me that the “overlay” model includes the only valuable function that journals still serve today, namely, signaling to other people in and out of a field whether a piece of scholarship is a worthy contribution, so that they don’t have to estimate its worth entirely on their own. Other tasks — providing a webpage, providing links, even copyediting — are trivial in comparison.

It seems to me that this type of overlay whatchimacallit is only relevant in those few field where pre-publication is likely, which rules out its relevance in any number of fields — fields where there is significant competition, fields where people are loath to publish early results. For example, in cases where findings may have policy/political consequences, it’s really unethical to put out results that aren’t 100% solid. I can think of several cases where government has leapt on early results and rolled out a program on that basis — with predictably disappointing longer-term results. In the humanities, I suspect academics would far rather polish up their ideas before presenting them to a wider public — what’s more, I can’t imagine anyone wanting to read things that haven’t been completely thought through. So how widespread are the implications of such a “journal”?

“In some ways a journal is a magical product in which the whole is greater than the sum of the parts.”

Actually, the journal is the one thing I can think of where the parts (the articles) are greater and more meaningful than the whole (the journal brand). And there’s no magic. Sometimes it really is just simple:

Journals made a lot of sense before the internet, when we printed and published scientific findings in paper articles bounded up in “journals”. Now, these journals do not make much sense, but we still have them as a matter of tradition. That’s what a journal is.

According to the author, these are the values that a journal publisher adds:
1. “peer review management” = finding other people to do the work of reading and reviewing
2. “production” = formatting an article other people wrote
3. “features that allow for electronic consumption” = saving a word doc as PDF or HTML
4. “intensive work that is put in to ensure survival of the journal” = Well, no. The survival of the journal only “adds value” if the journal itself “adds value”.

Clearly, none of these services are very valuable compared to the content. I can do all of them in a single day, whereas actual publication time takes months to years. Almost all the real work during that time is done by volunteers, not by the publishers.

The main value a journal adds is article “branding”. The only actual reason people publish their paper in “Nature” rather than journal X is so they can say they have “a Nature paper” on their CV. It is exactly why people wear gold chains rather than iron chains. The “Nature” brand suggests quality in the same way that gold chains suggest wealth. The problem is that we have much better ways of measuring these things directly. So to extend the metaphor, let’s just replace gold chains with online bank accounts.

The journal system is primitive form of publishing. Issues? A table of contents? Pay the cost of printing each additional page? Figures need to be black and white or pay a fee? Are you kidding me? The almost-irreversible yes/no decision of journal acceptance is made by one editor and two reviewers, who are often biased (like the score of a single movie critic), rather than an aggregate based on a large group of people over a large span of time (like a rotten tomatoes score).

The aggregate score is simply better. The other system based on fewer evaluations, i.e. the journal system, is simply worse.

I am very surprised by the opinions expressed in this article. Harington’s main arguments against overlay journals such as _Discrete Analysis_ are that (i) articles from such journals cannot go into the academic and recognition work-flow (e.g., career progress and grant applications) and (ii) publishers add all manner of extra value to the article as well as ensure the survival of the journal in perpetuity.

I completely fail to see how (i) is the case. All one needs for an article to go into the academic work-flow is that it has been reviewed, is citable, and accessible to others. As far as I’m aware, articles in overlay journals are peer reviewed just like in regular journals. As long as an article is assigned a DOI or an arXiv identifier, it is citable and (in the case of arXiv) accessible to everyone.

Regarding (ii), I don’t remember a single instance where copy editing or similar activities on journal’s part would have _significantly_ improved my article in any way. However, I do remember many occasions where I had to waste lots of time to format my article according to journal’s requirements as well as the hours spent pouring over the proofs. On several occasions sloppy copy editing on journal’s part has in fact resulted in more errors being introduced in my paper! In this sense I feel like journals often actually add “negative value”.

I also don’t see how including a link to supplementary material (one of the ways of “adding value” mentioned in the article) is a feature that takes significant effort on part of the journal’s editors. Regarding indexing and citations, it is completely unreasonable (and extremely inefficient!) for each journal to attempt to index all other journals and then count citations to their articles. Google Scholar and similar platforms do a very good job with this, and it does not require any extra effort from journals to be part of it (in fact, it suffices for the article to be available on author’s website in a PDF format to be indexed). Finally, I certainly trust arXiv more that commercial journals to ensure perpetuity of my articles (what happens to my papers if the journal goes bankrupt?).

To summarize, I completely agree with Gerry Carter’s comment above that the main value journals add is “branding”. However, I don’t see how this is an issue with overlay journals. Once such journals are around for some time and have respectable people publishing and serving on their editorial boards, these journals will become established in community and thus create their own “brand” (which can then be used in grant and hiring decisions if you wish). In this respect _Discrete Analysis_ is already a respectable journal even though there are no papers published in it yet!

Comments are closed.