Thursday, July 9, 2015

Open source journals?

Adam Liter send me this interesting piece on the Dutch university boycott of Elsevier publications. The complaint is one we are all familiar with: too expensive. It seems that academic publishers make lots of money on journals. Moreover, they do this on the back of lots of free labor. Reviewers are free, editor salaries are paltry, and buyers have little bargaining power. Not surprisingly this leads to large returns.  With the internet, the thought that there must be a better way always comes bubbling to mind.  So is there?

The obvious reply is that open source journals should be able to step into this breach and provide the same product at a better price. However, these journals have not taken over. Why not? Well, one reasons is that even such journals need cash and if it is not to be extracts from university libraries then it will come from this who want to publish. Some of this can be offset with grants. But this then means that scholars without such money will have to pony up (and the costs discussed can be large, running into 4 figures).  So, what to do?

I really don't know, but I think that this is an important question, especially for linguists. IMO, we don't have enough good publishing venues and it takes forever for research to get into print. Maybe this is less important than it used to be given things like the Linguist List.  Maybe publishing in LI or NLLT or NLS is not that important intellectually and that the only serious service journals provide is vetting of candidates for tenure. I don't know. What do you think? Is there a dark side to open source journals? Can they replace the large academic publishers? Should they? Your turn.


  1. There are open access journals (e.g. Biolinguistics, Semantics & Pragmatics) and a pretty cool, full-fledged open access publishing platform ( One issue concerning the question why these and more such journals have not yet "taken over" is the widespread infatuation with "impact factors" -- be it universities, administrators, funding agencies, or who else. Especially junior academics are very reluctant to go with non-IF journals. And guess who "awards" that famous IF? Yes, one of those greedy, commercial publishers. My guess is that it'll be tough for independent, open-access journals to ever get into that club... We need more than a boycott of Elsevier -- all these major publishers with their overpriced books & journals as well as the IF-driven rationale behind the wonderful world of "publish or perish".

  2. [Part #1]

    We’re looking in the wrong place when we bemoan the high cost of publishing.

    To be clear, I am not an apologist for Elsevier et al. I can’t get too worked up about their profits (Apple makes pretty big profits too), and I find their exploitation of their monopoly infuriating. But I think that in these discussions we too easily get hung up on the wrong things.

    Open access publishing is interesting. It has a lot to recommend it. But substantially reducing the cost of publishing is not its main virtue. (Norbert: “open source” = free-to-tinker software; “open access” = free-to-read journals.)

    The enormous cost of scientific publishing is us. Anything that takes the time of expert employees in cushy western societies is hugely expensive. So the biggest costs are things that consume our time. And somebody is paying for that time. Aspects of the publishing process that use our time ineffectively are the big money drains. Costs of journals are rounding errors by comparison.

    It’s common to combine observations about Elsevier’s profits with complaints about unpaid reviewers etc. I’d suggest that we set that aside. Cost out the time spent on the review process (your time, your benefits, etc.; multiplied by a half-dozen reviews), and it becomes clear that it’s vastly expensive, and it dwarfs any publisher’s profits. Your institution is already paying for that, as part of a compact with other institutions (“we’ll pay our folks to review your research, and you’ll pay your folks to review our research”). Imagine the effects on publishing if these costs were shifted to publishers. The costs would be passed back to authors/institutions, and things would just get worse.

    Academic Karma is an interesting new concept for creating greater transparency in the review compact.

  3. This comment has been removed by the author.

  4. [Part #2]

    Yes, scientific journals are expensive. They seem appallingly expensive compared to that novel that you’re reading. But that’s because publishing your research is simply very expensive, however you cut it. Yes, the costs of putting a PDF online are fairly low, and you might be able to outsource the type-setting to cheap labor in a developing country. And you can save a little by not printing on paper. But that’s not where the costs pile up. Almost all of the costs lie in the super-expensive time of expert humans (editors, assistants, publishers, IT people) in expensive western countries. Costs of $1000 per published paper are unusually low, whether you’re looking at commercial publishers, or non-profit publishers, or open-access journals. For example, Semantics & Pragmatics is a fine example of a successful open access linguistics journal. People love the fact that it’s free for authors and for readers alike. But it’s only possible because of the investments being made by MIT and UT/Austin (editors’ institutions) and the LSA. That journal is costing somebody well over $1000 per published paper. I’m not sure why people are so troubled by the power that Elsevier exerts, while expressing no concern about the power of wealthy institutions in other publishing models. Open access is good, if it can be done well. But somebody’s always paying for the expert work.

    The biggest gains are to be had from improving the efficiency of the authoring and peer review process. The amount of time that we spend on a cumbersome review process, the amount of time that we spend on revisions, the amount of time that we spend on inefficient authoring: those are the places where the costs really mount up.

    Open Access is a Good Thing because it removes barriers to consuming scientific information. It shifts the costs. That’s different than the question of how great the costs are, and who’s paying them.

    Open Journal Systems is an interesting project. Free software for journal management. It allows people to create new OA journals at little cost. I think it’s what Kleanthes uses for Biolinguistics, and it’s what S&P uses. But note that it doesn’t address the main costs in the publishing process: the editing and peer review process. An OA journal can be run well, or much worse than a traditional journal.

    My own OA experience is with editing for Frontiers, together with Matt Wagers and Claudia Felser. It has been surprisingly positive, and we have published lots of interesting work. Frontiers takes a different approach to OA, and it draws a lot of criticism, especially for the author pays model and for its unusual review model: they vet for soundnesss only, and not for impact. So if you have something sound-but-unimportant, and can pay the author fee, then you’re in. But their approach has a couple of very interesting properties. Their review model drastically streamlines the review process. It’s insanely fast compared to what we’re used to, even with revisions and suchlike. Even reviewers praise the effects. And they invest a lot in having an exceptionally effective back-end: their management software is very good, and their back room support for editors and authors is the best I’ve seen. That makes it fairly easy for me to be an editor, with low time outlay, and no local assistant etc. The full cost of publishing with them is probably quite low, once all costs are tallied. And that’s mostly by creating a more efficient review process.

    So discussing improvements in scientific publishing is very important. But getting fixated on Elsevier’s profits is missing the elephant in the room.

  5. @Colin: I think Marx made a pretty good case for profit being inherently exploitative. So in that sense, I think there's reason to object to the profits that commercial publishers make (as well as the profits that any company makes). The basic Marxist point is that profit necessarily involves paying somebody for some of their work an amount that is less than the value of that work (i.e., no exploitation, no profit).

    Nonetheless, I think your point is a very good one. Profit (i.e., exploitation) is ubiquitous and (probably) not going away anytime soon, so we should focus on places where we can make gains in the short term. As you rightly point out, shifting the costs of open access has a number of great benefits. You mentioned that it removes barriers to the consumption of scientific research. These barriers can actually be particularly pernicious, too. Not so much in the case of linguistics, but imagine the case of somebody trying to make an informed decision about which type of hormonal birth control to use (if any) but being unable to do so because they do not have access to research that has been done on the various forms of hormonal birth control and their various side effects. That is, given the predatory copyright practices of commercial publishers and paywalls, making an informed decision about this is largely impossible unless you are somehow affiliated with a university.

    Another reason that it is good is that it allows people to contribute to research even if they are not necessarily in a student or faculty position. (It's hard to contribute to current research if you cannot consume it.) Given the dearth of jobs, it seems that the field will inevitably lose people to jobs outside of academia. However, it would be a shame to lose the potential scholarship that might come from those folks, so I think we would be doing ourselves a service if we made it possible for those folks to be able continue to contribute to the field from the "outside" if they wanted to.

    And this is why I actually think this article is really exciting. Norbert focused on the Elsevier boycott in his post, but the reason for the Elsevier boycott is what's most interesting (in my opinion). It seems that the Dutch universities have been negotiating with major publishers to do precisely what you suggest: shift the costs. That is, they have been trying to negotiate deals with publishers where, rather than paying subscription fees, they just pay fees that would make the published research open access. Elsevier, it seems, is refusing to agree to this, and so the Dutch universities are boycotting Elsevier.

    I agree with you that strategies like this would seem to be the most promising, particularly for short term gains in terms of making things better. You pointed out that deriding profits can detract from focusing on this, and I agree (though I'm all for deriding the idea of profit! :) ).

    I think another side of that coin, too, is that focusing too much on the fact that there are costs can also detract from coming up with solutions. (People seem to really enjoy pointing out that there is no such thing as a free lunch, and this often seems to preclude brainstorming ways to make lunch available to a wider number of folks, even if it isn't free.) I was at a publication workshop at the Institute last night, and I tried to ask a question about this. One of the editors leading the workshop did bring up the exemplary case of what the Dutch universities are doing, but some of the other editors leading the workshop were content to simply point out that publications cost money and leave it at that. (Though, to be charitable to them, I should mention that time and discussion were short, so perhaps they would have said more given more time.)

    We should be taking the Dutch case as an exemplary model (as well as what Kai von Fintel and David Beaver have done with S&P) of how to shift the costs and try to do the same or similar.

    1. My point is that we need to focus on the *whole* cost, and that we are to blame for much of it. Sticking it to the man up at Elsevier is fun, and probably a good thing, but it doesn't solve the larger problem.

      Kai and David run a good journal, and they are indeed creating a model for the field of how to do many things well. But they also take pride in the fact that they reject the vast majority of submissions (as much as 90%, though I may be off on the exact number), and the current system rewards that. That kind of rejection rate implies prestige, and that's a mark of success. But just think of the amount of highly qualified work, paid for mostly by students and taxpayers, that goes into all that unpublished research. That is costly. And we have a system that incentivizes the creation of such inefficient processes. If we're bemoaning the cost of publishing, then I think we should take seriously the full cost.

    2. @Colin: I don't think we're in any disagreement here. I don't think we should be bemoaning the cost.

      Though I do think it's fine (and good) to bemoan the profits that publishers make as well as their practice of predatory copyright (which ensures their profits).

      But, I do completely agree with you that we shouldn't let any bemoaning get in the way of thinking about how to shift the cost. It's unfortunate, I think, that Norbert focused the content of this blog post on the Elsevier boycott. As I said above, the really exciting thing about this piece of news is what the Dutch universities are doing that Elsevier refused to agree to (which is what led to the boycott)—namely, having the money they pay count toward publishing journal articles as open access, rather than as mere subscription fees. This ensures that publishers still have money for the costs associated with publishing while also making the content open access.

      As for the rejection rate of S&P, that's a good point, and I suppose universities would be hesitant to fund something that wouldn't follow that model, but even if some new journals start popping up that follow that model, they do still have the potential to crowd out the equivalent high-prestige journals that are run by commercial publishers and that engage in predatory copyright and paywall their content. And this seems like a good thing to me, even if it doesn't address your very valid point.

      Anyway, doing what the Dutch are doing would certainly take a great amount of coordination amongst universities, but perhaps the fact that there is now a precedent in conjunction with the fact that some of the prestigious universities seem to be worried about the issue might be impetus enough for universities on this continent to try to band together to do something similar.

  6. @Colin: I agree with your basic point that the current system has a lot of inefficiencies and requires a lot more work than necessary, and that this is overall the most pressing issue. That said, I feel rather uneasy about the equation of time investment and cost. Putting a price on volunteer work --- and most editing and reviewing is volunteer work --- is like saying a small open source project like Xmonad cost thousands of dollars because that's the combined per hour salary of the developers working on it. Except that Xmonad is a passion project of those developers that they do in their spare time, so its development cost for society is exactly 0.

    And this distinction is important because it's the main reason why publishers are increasingly considered the bad guys: they're exploiting volunteer work for their own gain and hurt science and society at large in the process. The open source analog would be if github's business model were to resell all the software that they host (*cough*app stores*cough*).

    As for what can be done, I think the main problem is i) authors using inadequate tools like word processors that save them some learning effort in the short term but create a lot of additional work further down the pipeline, and ii) editors holding themselves to unnecessarily high standards. Here's an example of what I have in mind: Computational linguistics conferences usually publish proceedings, and the process works in a simple, straightforward way.

    1. Author downloads the Latex class file which does all the typesetting. The author then writes a paper and submits a PDF through Easy Chair.
    2. Editors pick reviewers from the program committee.
    3. Reviewers write short reviews of the paper, usually within a month after the submission deadline.
    4. If the paper is accepted, the author has 2 to 4 weeks for revisions. It is mostly up to the author how much they want to revise (as in, I've never had an editor complain about insufficient revisions).
    5. The paper is submitted with the original Latex code, which must be guaranteed to compile under specific parameters (big conferences like the ACL have an automatic compilation server that the source code is uploaded to; if that doesn't produce an output, your paper won't be in the proceedings).
    6. Papers are arranged in alphabetical order, then page numbers and footer are inserted. Proceedings done.

    This system is incredibly fast and easy, and the results are no worse than what you get in a journal like Computational Linguistics with more extensive peer review and meticulous typesetting guidelines. All that extra effort has few benefits for the reader. Sometimes the prose is more polished, the figures look a little nicer, spelling is consistently AE or BE, commas are meticulously placed according to a specific style, but who cares? Science is about content, not presentation.

    The only extra cost is the initial effort on the author's part when it comes to learning Latex --- but that also pays off in the long run since it simplifies the writing process for them, too, and enables a more dynamic writing process with version control through git, powerful macros that can be reused across projects, etc.

    So that's a good system that is already in place, well tested, and very productive, though it still doesn't go far enough imho. If it were up to me, readers would actually download articles in a layout-agnostic format like markdown and use that to produce a paper with whatever typsetting, capitalization and citation conventions they prefer. Separating content from presentation is one of the most important principles of information management, and it's time we adopted that in our publishing practices.

    And then we should drop the whole journal business altogether and just have a big repository of papers, with aggregators endorsing whatever they deem the best and most relevant. So you'd no longer have an LI publication, you'd have a paper that is recommended by prestigious linguistics aggregators. Alas, a pipedream.

    1. I wouldn't put our editing and reviewing work on a par with volunteer coding projects. You and I are both paid by public universities, and they regard this work as part of our job at a research university. They primarily reward us for our contributions to the publishing process, and they understand that our involvement in peer review is an integral part of that. Yes, we have choice over which review assignments to take on, and we may work on them at weird hours, but everybody understands that this is part of our job.

      Your CL example is interesting. I don't want to be drawn into a Latex fight, and I don't think that document formatting is a huge cost factor (though it's a real one). But I think that the comparison of successful models from related fields is instructive. You're right that that CLers don't eat up large chunks of their lives in endless review back-and-forth like GGers often do (if they haven't simply checked out of the peer review process). Part of this is because the review process is a bit more streamlined, and conference focused. Part if is also because the work is easier to evaluate. The papers don't construct complex chains of argumentation, and more people are qualified to assess a given paper. And people obsess about the process less, because everybody's putting out more (and smaller) units. CLers have a relatively good idea of what counts as good (in part due to the evaluation culture), whereas GGers are less certain. So, some aspects of their model might be importable, but there may be other challenges that GGers face that make them more similar to historians.

      I wonder why a post-publication grading system needs to be a "pipe dream". It could speed the release of results. It could streamline the vetting process (taking less time than we do already). And it could ensure that people put their best effort in to their first version, rather than saying "It's not worth fixing this now; let's just wait to see what the reviewers say". Of course, one of the key things to solve -- which is already evident in the fixation on citation metrics -- is how to ensure that work on novel topics or understudied languages isn't automatically marginalized. That's an especially important issue for linguistics.

    2. @Colin: You say that "I don't think that document formatting is a huge cost factor". Do you know what the main expenses of publishers are, then? I would have thought that the main expenses of a commercial publisher are (i) typesetting and (ii) hosting/distributing. Perhaps also copy editing, but I was under the impression that this was becoming less common of a thing for publishers to do (at least for journal articles?).

      It would be interesting to compare the expenses for a journal of a commercial publisher to a journal like S&P. I've tried looking for this information in the past, but I've never really found anything remotely informative, neither for commercial publishers nor S&P.

    3. There is some interesting discussion here on Stuart Shieber's blog about the finances of JMLR, a well-known machine learning journal. JMLR is open access and has no publication fees.

      It is slightly different from the CL model which is often indirectly supported by the ACL which has some financial resources of its own.
      Anyway an interesting and informative read and some entertaining action in the comments too.

    4. @Alex: Thanks! That was really interesting and informative to read. Karthik Durvasula also just pointed me in the direction of this article in Nature about costs, which was also useful and informative.

      So it seems that the Journal of Machine Learning Research has been publishing articles at about a price of $6.50 per article. Their operating budget has been $3,500 over the course of 12 years. I suspect we would see something similar with S&P's finances. And this is nowhere near what it costs commercial publishers.

      This is what I meant above, Colin, when I said that "I think another side of that coin, too, is that focusing too much on the fact that there are costs can also detract from coming up with solutions". So it's true that editors and reviewers do cost money, but universities pay this money, not publishers. And this is probably not going to change anytime soon. So factoring that cost into what it would cost to run an open access journal doesn't make sense unless one foresees universities ceasing to pay their faculty with the expectation that they review and edit for journals.

      (I actually think Thomas's analogy to open source software is rather apt, though perhaps not in the way that he initially meant it. It is my understanding that many of the bigger software companies actually give their employees free time to work on whatever they want to, and this work often ends up being contributions to open source projects. So I think the analogy is actually sort of apt.)

      According to the article in Nature that Karthik pointed out to me, it seems that it costs commercial publishers about $3,000 - $4,000 per article, based on some guesswork and reverse engineering. (It's pretty telling, and also not surprising, that commercial publishers apparently make universities sign non-disclosure agreements prohibiting them from saying how much they pay in subscription fees ... -_- ) This contrasts starkly with the $6.50 per article of the Journal of Machine Learning Research.

      Now, it's true that that journal is a computer science journal, so the people who run it have the requisite technological knowledge to run a journal and can do the typesetting with the free and open source typesetting system, LaTeX. So perhaps the operating costs would be a bit higher for a linguistics journal, but I do also get the sense that there are a non-trivial number of linguists who do have the requisite technological knowledge to do this (or at least some parts of it).

      Perhaps this is a bit optimistic/idealistic, but I think it would be great to see editors of major journals resigning to start open access journals instead. Even if they don't necessarily have the technological knowledge themselves, I would bet that there is somebody who works in the same subfield that does have that technological knowledge and that they could partner with to start the journal.

      In conjunction with senior members in the field trying to push their universities to band together and do what the Dutch universities are doing right now, I think these two things could be very viable strategies for ending the problematic practices of predatory copyright and knowledge hoarding.

      And I know this doesn't address your very valid point of ensuring that good crosslinguistic descriptive work sees the light of day, Colin, but I see that as a tangential issue. That being said, I honestly think that it might be easier for a journal that publishes good crosslinguistic descriptive work to be started as an open access journal run by linguists rather than as a journal run by a commercial publisher since commercial publishers are so infatuated with impact factors.

    5. @Adam: I'm perfectly fine with your extension of the open source analogy, though I think even the more narrow example I gave is apt. I'm pretty sure nobody has ever been told that their salary will be cut if they don't get started on editing a journal, so this kind of work is *not* enforced by the administration, nor does it factor much into tenure evaluation or merit raises. Many people doen't even list their reviewing activities in their CV. Basically, faculty salaries are the same no matter if they act as editors or reviewers, so that work is for free at an institutional level. And I'm also sure that the editors of S&P started their journal out of a sense of duty to the field rather than their academic job --- I would classify that as a passion project.

      Anyways, that's just nitpicking, I completely agree with everything you said, and I'm happy to see Shieber's blog post confirm my hunch that a combination of i) efficient tools and ii) less micro-managing on the editorial side cuts down the costs to almost nothing.

      As for whether linguists have the requisite technical expertise: when I was still at UCLA, pretty much all the students used Latex, and me and some of my colleagues here at Stony Brook require term papers to be written in Latex. That's just a basic skill students have to acquire, just like (depending on their subfield) the ability to use praat, R, graphviz, gnuplot, matlab, NLTK, a scripting language, or Linux. And the more specialized work on the hosting side (configuring apache or nginx, setting up a dynamic webinterface with AJAX or PHP, etc.) aren't particularly hard either and also have ready-made appliances (e.g. via docker). In the years to come, we'll be able to simplify things even more through P2P-hosting and P2P-DNS, which will remove the need for dedicated server hardware.

    6. It’s often very easy to get away with not meeting one’s professional obligations, but the lack of any effective enforcement does not entail the absence of the obligation. I would think that reviewing comes under the “professional obligation” heading for all of us (although starting a journal may be another matter).

    7. Thanks for the link to the Shieber post. It’s interesting. But it is delusional to take away from this that it really costs $6.50/article to publish a quality journal. And Shieber should know better.

      When politicians complain that university budgets should be slashed because academics are lazy idiots who hardly ever work, everybody complains that the politicians have no clue about all the important stuff that we do. We hear how academics work inordinately long hours because of their many responsibilities. Now in this discussion we’re told, “Nobody’s forcing us to do such-and-such, therefore it’s free!”. You can’t have it both ways.

      The part of the Shieber piece that is correct is that almost all of the costs of publishing a modern peer-reviewed journal are the costs of highly skilled people in expensive western countries (who expect not only salaries, but health and retirement benefits, and other extravagances). The editorial process takes expert time in many different ways: corresponding with authors, triaging papers, corresponding with reviewers, turning reviews into reasonable action letters, resolving disputes, etc. etc. There are some things that have in the past been handled by editorial assistants that can nowadays be shifted to software, but that carries costs too, especially if you want to do anything that’s not cookie cutter. And whatever you do, authors expect good "customer service”. Even if you automate as much as possible and make instructions as clear as possible, there will be questions, problems, etc.

      So when it comes to the “cost” of the journal, the action lies in who is being directly paid for their efforts (programmer, assistant, editorial stipend), and whose efforts are hidden by the fact that the work is rolled into their general professional activities (faculty salaries, administrator salaries, grad student stipends & tuition, etc.). In the case of Shieber’s example, the trick is that most of the costs are hidden in the second category, because the editorial team have the skills and the professional freedom to take care of most things themselves.

      There’s another important cost that’s buried in Shieber’s example, and that’s very hard to quantify: prestige. Aside from real money, this is the currency that academic institutions want. And they spend a lot of their efforts trying to turn real money into prestige. And academics are expected to spend much of their effort seeking to attain it, in various ways. So if you or your journal have a good amount of prestige, then a lot of people will be willing to contribute their own resources (mostly time/expertise) at no direct cost to you. The journal in Shieber’s example appears to have a lot of prestige. And the money that it took to create and sustain the academics that make up the prestigious editorial team was surely immense.

      How (dis)similar are the CS and Linguistics examples? As already pointed out, CSers are more likely to have skills that allow costs to be hidden. Also, the linguistics review/editorial process is more time consuming, for many reasons (that would take a post of its own). Finally, it’s especially ripe for CS professors at wealthy elite institutions to be encouraging the fiction that their efforts are free. They are well paid (compared to linguists), and in such institutions often have very low teaching responsibilities and excellent infrastructure support (admin and tech staff). A linguist who takes on a demanding editorial role will routinely ask for a course release or editorial assistance or a stipend, or some combination of those. They have more reason to make those requests than do CSers.

      By the way, since somebody asked about S&P’s costs: they are way higher than the $300/year in Shieber’s example. The LSA’s contribution is public information, and there are also contributions from MIT and UT/Austin. Billed costs amount to more than $1000/paper. As in Shieber’s example, the biggest costs are in the ‘hidden’ category.

    8. @Colin: I appreciate the realism about numbers and costs, but I don't think Thomas or I are delusionally suggesting that the actual cost of publishing an article is $6.50. If you want to quantify what you've called 'hidden' costs (what economists usually call 'externalized' costs, I think), then yes, it's going to be much greater than $6.50. As far as I can tell, nobody is disputing that (Shieber included).

      I think the only thing that Thomas and I are suggesting is that there are ways of externalizing ('hiding') the costs in such a way so that the non-externalized costs (i.e., the operating budget) of a journal would be very minimal. The Journal of Machine Learning Research is a case in point.

      So I appreciate the realism, but I don't understand why the more nuanced position that Thomas and I are painting isn't getting any uptake. One way that I suggested one might externalize these costs is by finding people to run the journal who both have expertise in the particular subfield as well as the technological know-how to implement the software that is needed for running a journal. I'd like to reiterate the suggestion that editors of major journals resign, partner with someone in their subfield who does have that requisite technological know-how (if they don't already have it themselves), and start an open access journal as an alternative to the journal they were previously editing.

      Your points about prestige and course releases/salaries are good points. Perhaps retired professors who no longer have teaching duties could be encouraged to help start open access journals. As for prestige, this is why I think it would be useful for editors of major journals to resign and start alternatives. Presumably editors of major journals already carry around a bit of prestige in their knapsacks.

      And as for the difference between the operating budgets of the Journal of Machine Learning Research and S&P, that is an interesting point. In my opinion, this suggests two things: (i) it would be further interesting to see the actual expenditures (not just the overall figure) of S&P to see if there are places where costs might be further externalized, and (ii) this is still significantly less costly than what commercial publishers spend per article, so why don't people in other subfields (e.g., syntax, phonology, etc.) do what Kai and David have done, especially since there are other universities with prestigious linguistics departments that could presumably convince their university to foot the bill.

      Let me end by saying that I really do appreciate the realism, and I hope I'm not coming across as particularly argumentative or stubborn (or 'delusional' :p ). I just think we should be focusing on trying to come up with solutions. The realism is useful for identifying problems with possible solutions and figuring out how to refine those solutions, which is what the suggestions I am making are trying to do. So this is useful, and I really appreciate the points and discussion you've raised.

      I think we have obligations to try to come up with solutions, since the practices of predatory copyright and knowledge hoarding can be pernicious. Again, not so much in the case of linguistics, but the birth control hypothetical I gave above is a good example. And presumably the more fields that start doing this, the more fields will follow. And then perhaps commercial publishers will also be further pressured into making the sorts of agreements that they are making with the Dutch universities right now, at least if they want to save themselves.

    9. Thanks, Adam. I'm pushing on this because I don't think we'll find good solutions unless we have a realistic understanding of the problem. Running an excellent journal is not like falling off a log. It requires a lot of expertise and effort from a lot of people, and it's harder than it used to be. Not because old farts are bad at technology, but because consumers' expectations are so high. I can assure you that (i) it is very hard indeed to find somebody who combines the skills needed to run a journal very well these days; (ii) when you do find them, you'll need to get in line, because somebody else will be making use of their skills, to be a chair, or a dean, or to run an organization; (iii) if you want to get to the front of the line, you'll need to offer them some good incentive: money, prestige, time, or something that they are really passionate about, or all of the above. Few people are interested in taking on such roles, and those that are willing and that do it well are making sacrifices of some kind. People who do it well put in a lot of work, they're highly prized, and there's no way that they could do it without the tacit approval of their employer.

      An alternative is to professionalize as much of the work as possible. Have it be somebody's job to make things run well. And preferably make it somebody who is a real professional, e.g., somebody who does editorial management for many journals, and so benefits from economies of scale; somebody who is not juggling editorial work with many other roles. The downside of this is that it makes it clear just how expensive the job is. The upside is that the real cost is lower, and the work does get done better. A professional editorial manager really can do things better and more cheaply than a handful of muddling professors.

      And to return to a theme that I emphasized early in this thread, we should not forget that a couple of hundred bucks here and there in APCs are a distraction from the real inefficiencies in the cost of publishing research. Calculate how much time goes into preparing an article, and reviewing it, and revising it, and re-reviewing it, and then doing the same with a second or third journal before it appears. And think of the amount that somebody is paying for that time. If you could come up with a more efficient way of "grading"/"validating" the research, then the savings would dwarf the numbers that we talk about in APCs.

    10. @Colin: I have to say that pretty much none of your points strike close to home with me.

      We've already discussed the point of hidden costs of time investment by academics, which you simply won't convince me off because under your calculations the cost of this blog increases by several hundred bucks every time I write a lengthy post. The fact is that many academics are very passionate and idealistic people that deeply care about their field, so they do a lot of unpaid work for the greater good --- often with better results than a paid professional. That doesn't mean one should add the equivalent cost of hiring a professional or paying academics for their passion work to the cost of the product, just like a pirated movie does not equal a lost ticket sale.

      I also don't understand what you mean by high consumer expectations. Considering that places like arxiv and lingbuzz are very popular, where you get manuscript versions without professional typesetting, layout can't be the issue. Nor do I know of anybody who particularly cares about what journal websites look like or how the files are hosted; as long as they can find a PDF via google scholar, people are happy. So the expectations must be about the paper selection process. But it would be news to me that standards have increased in that respect, rather it seems that recently there have been a lot of complaints about the shortcomings of the current peer-review format. That's not really about editing workload, though, it's about rethinking the process how we evaluate papers.

      I also think that the complaint about lack of expertise is misguided because it assumes that there must be a centralized place that bundles a lot of expertise, i.e. the journal. But that's because we're still stuck with the old idea that the journal should be responsible for typesetting, publication, file hosting, catalog indexation, and maintaining the technical backend of the reviewing system.

      [cont below]

    11. - Typesetting can easily be made the author's duty --- it's your paper, so you take responsibility for every aspect of it; and the CL example and the widespread use of LaTeX among linguistics students shows that this is feasible. Maintaining style sheets/class files does not need to be done by the journal as there's already a huge number of those floating around, just use one. A non-commercial journal doesn't need its own layout because it's not the product; the papers are. And even if you want your own style sheet, that's not a particularly laborious process --- writing and debugging the Latex class file for the UCLA working papers from scratch took me a weekend, with another day for writing documentation and about one day per quarter for follow-up maintenance.

      - Centralized file hosting is completely unnecessary. All you need is that each university runs a P2P seedbox that shares all papers produced at this institution. That's fairly simple and distributes hosting costs across the community; most universities are already moving towards such repositories anyways, though they do not use any P2P protocols at this point. The journal only has to provide magnet links (or something equivalent), ideally signed with a Web of Trust mechanism to allow for decentralized certification.

      - Publication involves distributing the file and advertizing it. The former is no longer necessary given the previous point, the latter is a matter of writing a script that pushes the published paper with summary and tags to a blog, the RSS feed, and a monthly digest to LingList. You also need a script for cataloging on Google Scholar and other indexers. This script has to be created only once, with source code publicly hosted on a service on a github so that it can be reused, modified and maintained by other members of the community. In principle, you could leave indexation to the author since it's in their own interest and many authors have a Google scholar account anyways where they list all their publications.

      - The technical backend for managing the review process should also be a community process. Basically a community-maintained, open-source EasyChair+ for journals. These things usually run on a LAMP stack, possibly with some Python, Perl, or Lua scripting in the background, so you can prepackage the whole thing as a VM appliance or a Docker container, which can be downloaded and installed in a matter of minutes.

      There's probably some other procedural issues I forgot about, but I'm sure that they, too, can be tackled along the lines outlined above. So I just don't see the big hurdles, the required expertise can easily be distributed across the whole community, and the required technology already exists.

      Maybe that was your point all along, but to me it read more like "there's no way you can do cheap Open Access publications because of X, Y, Z (and if Shieber says he's doing that, he's fooling you), so quit whining".

    12. "Higher consumer expectations" are primarily associated with efficient review, rapid online publications, and especially with metrics (such as IF). In linguistics in the US we're relatively well shielded from the tyranny of metrics, but in many countries it is increasingly oppressive. From what I have seen of editors in action, this pushes them to worry about things that they didn't need to pay attention to in the past. (And the fact that linguistics review processes are so slow contributes to the fact that all of our journals have IFs that are in the toilet.)

      I agree with you that if you took out the review/grading/publicizing function of journals, then things would be much easier. But that's about the only function that journals still serve. (And the publicizing is more helpful to some than others. The rich and wealthy don't need that so much.) It's managing a fair and effective review process that eats up most of the resources and expertise. And it's one thing to have people who *can* do what it takes, and another to have people who will reliably do it, year in year out. It's hard work.

      Scott Walker agrees with you on the hidden costs issue. That's why he thinks that you should be teaching twice as many classes (or doing something else that counts as useful work). I disagree with him. [Sorry, the SW comparison is a low blow.]

  7. Ubiquity press can publish OA articles for $400 in Author Processing Charges, no other costs incurred. Over $1000 APCs/ paper no longer reflects a fair market price.

    1. In line with this, I understand that Open Linguistics (a relatively new De Gruyter journal) will be asking < $400, though in 2014-2015 there is no APC fee.

    2. If I understand it correctly, de Gruyter has a policy with a sliding scale for APCs: new journals are free, the more prestigious the journal becomes, the more you start paying... Not quite the spirit of Gold Open Access...

  8. We've made a list of different venues for open access in linguistics, and also several posts explaining a bit more.

    I also suggest reading post by Haspelmath on the Language Science Press enterprise and how they work, much inspiring.

  9. As Editor-in-Chief of NLLT I feel like I should weigh in here, but I have only a few comments.
    It is important for the field to have identifiable prestigious journals, however those journals are published.
    The review process in linguistics yields papers that are significantly better in both comprehensibility and in content than the original submissions. (Thank you to all!) Hence the process is crucial for the scientific progress of the discipline. (And so I disagree with Thomas Graf here.)
    The length of time the review process takes is entirely under our control. Myself and my associate editors are endeavoring to do our part by coming to decisions on papers as soon as we receive the reviews. Reviewers can do their part by making the time for the review as soon as possible after receiving the request, rather than waiting until the review is actually due. This review process, though, indeed falls into the (largely) unpaid labour category. (And labour of love does not seem the right characterization!)
    On the paid labour side, I can say only that the person who manages the website is indispensable, and I have little involvement in or knowledge of the actual publication process. I disagree with the idea of making authors responsible for their own type setting; this seems a very poor use of authors’ time and expertise.

    1. Thanks for chiming in. I'm not sure what gave you the impression that I consider peer-review dispensable, pretty much all my suggestions where about the part of the publishing process that does not involve the author-reviewer interaction. And the very first example I gave --- i.e. how things are commonly done in computational linguistics and computer science --- still involves peer review. The most radical thing I mentioned in that respect is my pipedream of reducing journals to aggregators and that still does not remove peer review because i) those aggregators can have their own set of reviewers to decide which papers to promote, ii) readers can discuss papers publicly and anonymously, and iii) the author can (and should!) upload revised versions based on these comments (ideally with a version control system so that older revisions are still readily available).

      As for authors doing their own typesetting, the current system where authors have to manually replicate typesetting conventions for manuscript submission is often more wasteful than anything I suggested. A good typesetting system does these things automatically via a style sheet, to be provided by the journal. Even manually setting your word processor to the right line spacing takes longer than loading this style sheet. And I remember submission processes that had incredibly tedious requirements (in particular font size and face for sections, subsections, paragraphs headers, citation style, etc).

      The other big part of typesetting concerns figures and graphs, e.g. trees, plotting data and so on. Authors are already pretty much on their own in that area and use a specific set of tools, whatever works best for them. The annoying part usually isn't using those tools but when the publisher insists that these things are submitted in a way that is not compatible with that tool chain (and often produces inferior results). For instance, there have been times where I had to convert my vector-based trees from PDF to JPG just so I can include them in the doc file the publisher/editor insists on (at least not docx). Extra work for me that yields a file that is much bigger while looking much worse at high zoom levels.

      Is the approach I champion here the perfect solution to all our publishing problems? No, at least not at this point. It comes with a steep initial learning curve for people who only know word processors (the number of which is fortunately decreasing), and crowd-sourcing peer review is bound to fail unless the majority of the community is comfortable with online discussion. But at the same time nobody here has come forth with any substantial arguments against what I said about file hosting, distribution, open source management tools, and prepackaging publishing platforms as VMs. So why don't we start with those purely technical points and then phase in the more ambitious changes step by step?

  10. An interesting new initiative that's relevant to this thread is Collabra (, which was launched very recently. It's put out by the U of California, which of course has some muscle, and it includes a section on Social and Behavioral Sciences. The most notable feature of their model is their revenue distribution scheme. They collect Article Processing Charges (APCs), like many others, and they have a scheme for waiving the charge. What is unusual is that they say upfront that $250 from the APC is dedicated to supporting waivers, and that they will periodically invite editors and reviewers to choose what to do with any surplus funds, either (i) taking them as a (likely very small) payment, or (ii) designating them for institutional APC-support funds, or (iii) returning them to Collabra to support future waivers etc. They seem proud of their "low" APC, but it's set at $875/article. Interestingly, they claim that their "OA mega-journal" approach is needed in order to make it feasible to adopt their financial model. It's very early days yet, but it will be interesting to see if it attracts interest.

  11. This Facebook post from Johan Rooryck is relevant to this discussion.

    I find this particularly exciting. I think this is exactly what should be happening. Editors should be making these demands of publishers, and, if they aren't met, then the editors should resign and start a new journal. This at least partially solves the issue of new journals not having the same prestige as already established journals.

    I see that Colin has already expressed skepticism about the funding model on the Facebook thread. :p

    Anyway, just thought that I would share this here.

    1. You can call it skepticism, or you an attempt at clarification. It was informative. This is part of a broad effort to implement a cost-sharing model for OA humanities publishing. It's interesting, and it's worth a look. The folks behind the initiative seem to agree that it costs around $500 per published article to stay afloat, but the model involves having those fees covered by a consortium of willing-to-pay institutions, rather than by the producer of the researcher, as in the common APC model. The traditional model charges you to read. The APC model charges you to publish. The OLH model charges you for neither, but relies on you volunteering to pay anyway. Assuming a fixed set of costs, each model allocates the payment, and the level of control over what you pay for, a bit differently. I'd be curious to hear arguments for why one allocation is preferable to others.

    2. Does anyone actually know what fair subscription/reading access costs would be? There are a lot of articles that I would probably pay $5 to get from journals that my library doesn't subscribe to, but not $40. Could a reader-supported archive stay afloat at $5 per article? Could a monograph publisher stay afloat at $15 per booklength pdf?

    3. I am deeply in love with every single piece of information you post here. Will be back often to read more updates!

      Online Billing Software

  12. This comment has been removed by a blog administrator.

  13. Very informative site and article, i must bookmark it and keep posting interesting articles.
    Open source crm hosting