I first wrote this for oral presentation on January 3, 2004, at the Annual Meeting of the American Philological Association in San Francisco. Thereafter I worked up the second section, on obstacles, into Open access in the humanities for the February 2, 2004, issue of my newsletter. The version here builds on the original APA presentation and the revisions made for my newsletter. It has since been published in the proceedings of the APA meeting, Syllecta Classica, 16 (2005) 231-246. It has also been translated into Italian by Francesca Di Donato, Promuovere l'"open access" nelle scienze umane, Bollettino Telematico di Filosofia Politica, September 15, 2005.

Promoting Open Access in the Humanities
Peter Suber, Philosophy Department, Earlham College

What is open access?

The short definition of "open access" is free online access. But there are several important nuances to elucidate. First, while the access is free of charge to those who already have an internet connection, it is compatible with priced access to enhanced or print editions of the same texts. Second, making the works accessible without charge removes price barriers to access, the most important barrier for most people. But open access also requires removing the permission barriers to access and use, for example, copyright and licensing restrictions that require permission before one may copy, download, store, redistribute, crawl, or link to the texts. Third, one way to remove permission barriers is to put the work into the public domain. But another way that's just as effective, and somewhat more attractive to authors, is for the copyright holder waive some rights and retain others, consenting to open access while reserving the right to block the distribution of mangled or misattributed copies.

Compared to print, the internet lets us achieve wider distribution and lower costs at the same time. That's a very good reason to use it. But the internet doesn't itself remove the price and permission barriers that restrict access today. These are removed by the will of the author or copyright holder. We scholars are likely to consent to open access for journal articles because we are not paid for them. We write them in order to make a contribution to our fields, and thereby to make a contribution to our careers. We are paid by our employers to make this kind of contribution to knowledge, which compensates us even when only a handful of people around the world care to read what we've written. When we consent to open access, we increase the size of our audience and the impact of our work. But we don't lose any revenue. So we have everything to gain and nothing to lose.

By contrast, musicians earn money from their work. They will either lose revenue by giving it away online or they fear that they will. Their consent to open access will be much harder to obtain. We can already see some of the attempts to obtain it: the argument, for example, that free online access to music will allow consumer sampling, which in turn will increase consumer purchasing. There's some evidence that this works, and some artists who swear it works for them. There's also some evidence on the other side and many artists —and many record labels— refusing to consider it.

This should make us pause to appreciate how special, or how peculiar, the scholarly journal article is. It is just about the only creative work whose creator willingly gives it away. That makes it anomalous, exceptional. This is the same property that makes it the low-hanging fruit for open access.

We can put it this way. Phase 1 of the open-access movement is to achieve open access for works by authors who give them away. Phase 2 is to achieve open access for works by authors who earn money from them. This will require persuading authors to take a risk, which was not necessary in Phase 1. Phase 3 is to achieve open access through copyright reform, an even more difficult job than persuading musicians to consent to open access. Intelligent copyright reform would enlarge and preserve the public domain and thereby enlarge the domain of works for which there are no permission barriers to access. We could do this through establishing the first-sale doctrine for digital content, restoring fair-use rights denied by digital rights management (DRM) and repealed by the Digital Millennium Copyright Act (DMCA), shortening the term of copyright, and letting federal copyright law preempt the state contract law that governs licensing agreements.

There are two primary vehicles for delivering open access: archives and journals.

An open-access archive does not perform peer-review. It merely provides open access to its contents. It may be sponsored by an institution (like a university or funding agency) or by a discipline. It will typically contain both preprints and postprints. There are half a dozen open-source software packages to create and maintain these archives. So setting one up is virtually costless. The difficulty is to encourage scholars to deposit their papers in them.

When archives comply with the protocol of the Open Archives Initiative (OAI), then they are "interoperable". This just means that programmers can create search engines and other data services that deal with all the separate, compliant archives as if they formed one grand virtual archive. So users needn't know that you deposited your paper in a certain archive; they needn't visit that archive and run a separate search on it, and they needn't even know that the archive exists.

Open access journals do perform peer review. Even if all the editors and referees are donating their labor, like the authors, peer review costs money. But this is just the cost of facilitating the process in which the labor is donated —sending a copy of the file to a referee, monitoring how long the referee is holding it, nagging referees who dawdle, keeping authors informed, keeping track of which version is the most recent, keeping track of the journal's acceptance rate, and so on. As you can imagine, software to manage this process is getting better all the time. And since November 2002, one package for doing this has been open source (Open Journal Systems).

But while open-access journals have lower expenses than print journals, they still have expenses. The model that is catching on in the natural sciences is to charge the author an upfront processing fee to cover the journal's expenses in vetting the article, preparing the manuscript, and disseminating it online. In practice the fee is typically paid by the author's research grant or employer, not from the author's pocket. Many of the world's largest private foundations that fund scientific research have declared their willingness to pay these fees. The fee at the open-access journals published by BioMed Central is $500. At the Public Library of Science the fee is $1500. These open-access publishers only charge for accepted papers, in order to protect the peer review process from contamination and suspicion, and they are willing to waive the fees entirely in cases of economic hardship.

Diagnosis:  why is open access moving so slowly in the humanities?

Open-access archiving took off fastest in physics and open-access journals took off fastest in biomedicine. There are fascinating cultural and economic reasons why these disciplines opened first. But let's focus on the other end of the pack where open access is moving the slowest. Why is it moving so slowly in the humanities?

Here are nine differences between the humanities and the sciences that help explain their different rates of progress.

  1. Journal prices are much higher in science, technology, and medicine (the STM fields) than in the humanities. The pricing crisis is not the only reason to consider open access, and not the primary reason for researchers themselves, but it's a major one for libraries and universities. In the humanities affordable journals defuse the urgency of reducing prices or turning to open access as part of the solution. Researchers have the same motivation to consider open access in the humanities and the sciences —to enlarge their audience and increase their impact. But the sciences see a convergence of motives, and hence a partnership of stakeholders, missing from the humanities.

    According to the 2002 Library Journal pricing survey, the average subscription prices for journals in STM fields were 10-20 times higher than the average prices in the humanities. For example, compare biology ($1,097.01), chemistry ($2,143.22), and physics ($2,218.82) with history ($126.35), literature ($110.51), and philosophy ($146.60).

  2. Much more STM research is funded than humanities research. Hence, in the STM fields there is much more money to pay the processing fees charged by open access journals. In the humanities, there are fewer open access journals, and nearly all of them operate without processing fees.

  3. At least in the U.S., the government funds far more STM research than humanities research. Hence the taxpayer argument for open access (that taxpayers shouldn't have to pay a second fee for access to the results of taxpayer-funded research) is stronger in the STM fields than the humanities. The taxpayer argument isn't the only argument for open access, but it's one of the strongest and certainly one of the first to appeal to policy-makers and the public. It may only apply to a fraction of STM research, but that fraction dwarfs the comparable fraction of humanities research.

    Total U.S. federal funding for university research in fiscal 2001, in all fields, was about $19 billion, which constituted about 60% of all funding for university research. Of this, eight federal agencies, all in STM fields, provided 97% of this funding, and two of the agencies alone, NIH and NSF, provided $14.2 billion or 75%. By contrast, the National Endowment for the Humanities (NEH) budget for 2002 was $124 million, less than 1% of the STM funding from the 8 leading federal agencies alone.

    (Sources, GAO report of November 14, 2003 and NEH budget request for 2004.)

    If we don't limit ourselves to university-based research, then the total research budget of the U.S. is much greater, $110 billion in 2003. All the funding beyond the subset for university-based research was for the STM fields, including defense, none for the humanities.

    (Source, Rand Corporation, Federal Investment in R&D, 2002, Chapter One.)

  4. On average, humanities journals have higher rejection rates (70-90%) than STM journals (20-40%). This means that the cost of peer review per accepted article is higher in the humanities, lower in the STM fields. Hence, for open access journals that cover their expenses through processing fees on accepted articles, the fees would have to be higher at the average humanities journal than at the average STM journal. This combines badly with the fact that the humanities receive much less government and foundation funding than the sciences.

    H.A. Zuckerman and R.K. Merton first showed this disparity in rejection rates in the 1970's, and ALPSP confirmed it in the 1990's. See the discussion thread on this question from January 2001 in the American Scientist Open Access Forum. See H.A. Zuckerman and R.K Merton, "Patterns of evaluation in science: Institutionalization, structure and functions of the referee system," Minerva, 9 (1971) 66-100. Also see a brief online summary of the 2001 data collected by ALPSP.

  5. There is more public demand for open access to research on (say) genomics than Greek grammar, which is one reason why genomics has more federal funding than Greek grammar. Of course research can be worth funding and making openly accessible even in the absence of public demand, but public demand tends to create funding, policy, incentives, and lobbyists. Note that public demand ranks some scientific research topics above others (medicine above field biology, for example), just as it does in the humanities (American history above Roman history, for example).

    I can acknowledge this even though my own field (philosophy) is in the humanities. But a more contentious way to make a similar point is to say that STM research is more socially useful than humanities research, at least in the way that attracts funding. This is what makes foundations and governments want to pay for it, and what makes them receptive to the argument that the subsidy for open-access dissemination is worth paying too since it makes a useful research project even more useful.

    Before fellow humanists write me angry letters, I'm not saying that humanities research isn't socially useful, or is less useful than the sciences, merely that this is the perception of most funding agencies. There are two kinds of usefulness, which is why the sciences and humanities coexist wherever civilization takes root. But each kind of usefulness tends to be dismissed or misunderstood by champions of the other. The most succinct wisdom on the usefulness and fundability of humanities research was uttered by Aristippus, a Greek philosopher who sought patronage from one rich Athenian after another. Dionysius once asked him, "Why do I always see you philosophers knocking on the doors of the rich, but I never see the rich knocking on the doors of philosophers?" Aristippus replied, "Because philosophers know what they need and the rich don't."

  6. Preprint exchanges meet more needs in the STM fields than in the humanities. STM researchers need to know quickly what is happening in their microspecialization, partly to build on it in their own work and partly to avoid being scooped. Moreover, they need to deposit their own preprints quickly, partly in order to influence fast-moving research and partly to establish priority over others who might be working on the same problem. Preprint archives are very common in the natural sciences, very rare in the humanities.

    Of course humanists build on one another's work too and worry about scooping and being scooped. But there's no doubt that the urgency of timely notification of other work is greater in the STM fields than in the humanities. The explanation may lie deep, for example, in their different ways of being socially useful and their different ways of recognizing and rewarding the solution of problems.

  7. Demand for journal articles in the humanities drops off more slowly after publication than demand for articles in the STM fields. This means that humanities journals will worry more than STM journals that offering open access to articles after some embargo period, such as six months after publication, will jeopardize their revenue and survival.

    There are three differences here: objective rate of decline in demand after publication, objective risk of lost revenue from delayed open access, and subjective fear of lost revenue from delayed open access. But none of these means that delayed open access will really jeopardize revenue and survival, either in the sciences and humanities. The revenue from selling access to old issues is miniscule, and losing that revenue will not harm a healthy journal, especially when offset by enhanced access, visibility, and impact that can be translated into increased submissions, advertising, and (if they still exist) subscriptions.

  8. Humanities journals often want to reprint poems or illustrations that require permission from a copyright holder. It's much harder to get reprint permission for open-access distribution than for a limited-circulation, priced and printed journal. And when permission is granted, for either kind of distribution, it usually costs money. This is why open access will come last to art history.

  9. Journal articles are the primary literature in the STM fields. But in the humanities, journal articles tend to report on the history and interpretation of the primary literature, which is in books. STM faculty typically need to publish journal articles to earn tenure, while humanities faculty need to publish books. But the logic of open access applies better to articles, which authors give away, than to books, which have the potential to earn royalties.

Summary:  open access isn't undesirable or unattainable in the humanities. But it is less urgent and harder to subsidize than in the sciences. Progress is taking place, and as more humanists come to understand the issues, and the strategies that work, we should expect to see progress continue and accelerate. For example, humanists may have fewer reasons for preprint archiving than STM researchers, but most of the advantages of preprint archiving still apply in the humanities and they are starting to have an effect. Humanists may feel less urgency to launch peer-reviewed, open-access journals, and find it harder to do so without funding for processing fees. But there are still reasons to launch such journals and other funding models to sustain them. Humanists may be more skittish about offering open access to their books than to their journal articles, but there are reasons why informed authors will choose to try the experiment. In the next section I sketch some of the strategies to facilitate these advances.

Recommendations:  how can we advance open access in the humanities?

Most of these are recommendations for keeping costs down without sacrificing quality. One goes beyond journals to archives, and one goes beyond Phase 1 solutions to Phase 2.

  1. Use journal management software to reduce the costs of peer review. Use open-source journal management software to reduce the costs about as far as possible. Currently, Open Journal Systems (OJS) from the University of British Columbia Public Knowledge Project (PKP) is the only open-source journal management software. (I have no financial ties to OJS or PKP.)

  2. Do without copy editors. At most journals with both copy editors (who improve language) and disciplinary editors (who supervise peer review), the copy editors are paid salaries and the disciplinary editors donate their labor. Hence, dispensing with copy editors will reduce costs without interfering with peer review.

    Someone might object that much writing in the SSH fields is atrocious and needs scrupulous copy editing. I couldn't agree more. (Don't get me started.) But copy editing wouldn't help much with this problem because journals that publish atrocious writing seem to think that it's professional or sophisticated and wouldn't employ copy editors who favored mere clarity. I feel free to say this because the average writing in my field —philosophy— is much more atrocious than the average writing in classics.

    Many STM journals dispense with copy editors in order to save money. But when they do use copy editors, it's often because they receive many submissions from non-native speakers. This is much less of a problem at humanities journals.

    Just so that I am not misunderstood:  I regard copy editing as a valuable service. I don't want to discard it unless there is no other way to lower costs enough to provide open access. Dropping copy editing is not a way to improve a journal, just a way to cut costs without reducing a journal's essential service of providing peer-reviewed research to a scholarly community.

    The best solution —as long as we can't teach scholars to write clearly— is to find enough revenue to pay for copy editing alongside essential services like peer review. This is happening at the Public Library of Science journals, for example. However, we must admit that the shortage of funds makes this unlikely in the humanities. If we have to choose, then I would definitely prefer open access without copy editing to copy editing without open access.

  3. Get universities to pay processing fees

    Major public and private funding agencies in the sciences have declared their willingness to pay the processing fees charged by open-access journals. But I don't know of any foundations supporting humanities research that have followed suit. One reason, clearly, is that there are virtually no such journals today and no constituency seeking foundation support for them. This might change slowly, if open-access journals in the humanities can overcome the chicken-and-egg problem and survive long enough to ask for continued support. But the National Endowment for the Humanities (NEH), for example, is unlikely to help here. It conceives its grants more to subvene a scholar's salary than to pay for a specific piece of research. Nor could it give a grant to a journal to redistribute to scholars to cover the processing fees on accepted papers, since the NEH does not "regrant" or give grants for others to distribute.

    Is it hopeless to expect universities to pay the fees for their humanist faculty? Perhaps today it is. Most universities are not only strapped, but perpetually strapped —or at least that's how it seems from the hallways of the humanities. But as open access spreads, university library budgets will realize large savings from the conversion, concellation, or demise of subscription-based journals. The first priority for this savings should be to pay the processing fees to support the superior publishing model that made the savings possible.

    No doubt, however, this kind of solution is more long-term than short-term. But while I admit that, I also want to suggest that visionary leadership in universities will understand that investing in the open-access alternative today, before subscription savings make it easy, will not only share knowledge and accelerate research, but also save the university serious money in the long run.

  4. Even if universities can often pay the upfront processing fees, open-access journals in the humanities will have to explore other ways to cover their expenses.

    For example, Philosopher's Imprint is an open-access philosophy journal from the University of Michigan. It's motto is, "Edited by philosophers, published by librarians." Because the philosophers and librarians are already on the university payroll, the journal charges no processing fees. The result is that authors don't pay for dissemination and readers don't pay for access. The same model works for STM journals in fields with low funding. The Journal of Insect Science is published by the library of the University of Arizona at Tucson. For more on libraries as journal publishers, see recent reflections by Harry Hagedorn and Eulalia Roel.

    Another model that works where funding is minimal is the overlay journal, which is essentially just an open-access archive plus an editorial board. The board takes submissions through the archive, makes its judgments, and deposits approved (perhaps revised) articles back in the archive. By using open-access archives as the publishing infrastructure, overlay journals reduce expenses about as far as possible without skimping on peer review. The University of California recently launched a series of overlay journals in all disciplines, based on the UC eScholarship Repository.

  5. Experiment with retroactive peer review. In retroactive peer review, a journal starts with unrefereed submissions on an open-access web site, subjects them to its own favored method of vetting or review, and then either copies the approved articles to a different site or marks them somehow as approved. The reason for experimenting with retroactive peer review is not that open access depends on peer-review reform or vice versa; they are independent. The reason is simply that retroactive peer review costs less than traditional, prospective peer review, and this fact matters in fields that are not well-funded. There are many kinds of retroactive peer review, some more rigorous and adequate than others. This universe has barely been noticed, let alone explored. The importance of experimentation is to find the more rigorous and adequate forms of it, not just ways to save money. If we can find these rigorous and adequate forms of retroactive peer review, then there is no academic loss, and much financial gain. There might even be academic gain, especially if you think that peer review as currently practiced leaves considerable room for improvement.

  6. Work for price reductions and open access (two different things) for STM journals. High STM journal prices harm the humanities as well as the sciences, and the savings from lowering these prices can be the salvation of the humanities. One expensive STM journal can cost more than $20,000/year, more than 100 middle-tier humanities journals. Moreover, we know that the rising prices of STM journals cause libraries to cut into their book budgets, which hurts the humanities much more than the sciences. When libraries buy fewer books, university presses accept fewer manuscripts. Open access in the STM fields would produce savings that could be spent on both (1) monographs and (2) processing fees for open-access journals.

    Quoting Kenneth Frazier in Library Journal Academic Newswire for November 20, 2003: "A lot of library directors won't admit this, but often the STM increases [in journal subscription prices] come out of the hide of the humanities." [Not online.]

  7. Notice that all the recommendations so far have concerned open-access journals. If their costs can be kept sufficiently low, or if savings elsewhere in university budgets can subsidize them, then they can thrive. But journals are only one of two major avenues to open access. The other is open-access archiving. Every university in the world can have an institutional archive built from open-source software, and should. It can be installed in a weekend at essentially no cost, and even if very successful would use fewer resources than the MP3 traffic generated today by students. If there were no open-access journals in the humanities at all, scholars could still deposit their preprints in their institutional archives, and in most cases their postprints as well. This would not only provide open access to those articles, but also give the authors freedom to publish in any journal of their choice, open access or not.

    This model works just as well with disciplinary archives (dedicated to all the eprints in a field) rather than institutional archives (dedicated to all the eprints produced by the institution). Here are are some examples of disciplinary open-access archives in the social sciences and humanities.

  8. Finally, all the recommendations so far have been "Phase 1" recommendations —ways to provide open access to the texts that scholars already consent to give away. But we should consider the obvious "Phase 2" recommendation: humanists could consent to open access for their monographs, not just their journal articles.

    There are two reasons for an author to consider this possibility.

    1. Free online full-text might increase net sales. This is the experience of the National Academies Press (for research monographs) and the Baen Free Library (for science fiction novels). For some of the explanation why, see reflections by Michael Jensen, director of the National Academies Press, and Eric Flint, co-founder of the Baen Free Library.
    2. Even if open access doesn't increase net sales, the benefits of open access are significant (greatly enlarged audience and increased impact) and the royalties on an average humanities monograph between zero and meager.

    Book authors who are still nervous could consent to open access after they think the majority of purchases has already occurred. During the time that the monograph is toll-access only, the author could still provide open-access excerpts and metadata online to help scholars find the book and learn whether it is relevant to their own research projects.

Translated into Italian by Francesca Di Donato, Promuovere l'"open access" nelle scienze umane, Bollettino Telematico di Filosofia Politica, September 15, 2005.

For more on open access, see my blog or newsletter.

Peter Suber
Research Professor of Philosophy, Earlham College
Open Access Project Director, Public Knowledge
Senior Researcher, SPARC
peters@earlham.edu

Copyright © 2004, Peter Suber. This is an open-access document.