[Editor’s note:  This essay first appeared in Against the Grain (November 2013).  Joe would like to thank Katina Strauch for her editorial insights in developing this.]

Everything has limits. While there is much discussion about the limitations of the traditional publishing model, where users or their proxies (e.g., libraries) pay for access, the natural limits of open access (OA) publishing are often overlooked or are discussed only in unproductive, heated online forums. What I propose to do here is to identify some of the natural limits of the Gold variety of OA publishing with the aim of focusing subsequent discussion on how to moderate or eliminate those limitations.

I said that traditional publishing has its limits, too, and that they are well known, but perhaps it would be advisable to rehearse those limits briefly.

The most significant characteristic of traditional publishing is that is designed to operate in a market economy. For some, anything that smacks of the marketplace is anathema for scholarly activity, but even more moderate souls will be prompted to ask what happens when there is literally no market? This is not an unusual situation for scholarly material. Some research is so specialized that the number of interested readers is tiny, at least today (who would want to predict the impact of research a decade or a century from now?). Such specialized work exists, if it can be made to exist at all, outside the marketplace. Other material lacks a market for the simple reason that there is no money to pay for it. This is the case for a great deal of scholarly material in the developing world, and even in the First World a library with no money to spend represents no market at all. Traditional publishing has limits and they are marketplace limits.

The marketplace limits of traditional publishing affect every aspect of the publishing process. Most importantly, it gives rise to the practice of pre-publication editorial review (because only the better works will find a market), which in turn means that a small number of editors serve as gatekeepers. For some people, editorial review is the strength of traditional publishing; for others it is an almost satanic practice that suppresses free speech. What is indisputable is that editorial review under the traditional model aims to restrict what gets published by making judgments about a work’s importance, appropriateness (for a particular publisher or journal), originality, and other subjective measures of quality. This means that some authors and works do not get published at all, which is a limit of a kind.

The traditional editorial model imposes an almost binary distinction between what is published and what is not. Are we comfortable that whatever is not published is totally worthless? Or do we believe that materials fall onto a continuum with outstanding work on one end and worthless books and articles on the other, with most works lying somewhere in between? Should an article that makes a small contribution but not a grand one be shut off from readers entirely?

Everything changes with OA, though there are many varieties of OA and it is hard to generalize about all of them. The most important distinction is between the Green and the Gold varieties. In Green OA authors deposit copies of their articles in publicly-accessible repositories. Green publications continue to participate in the traditional publishing process, but the self-archived copy represents an escape valve, as it were, providing access to the material even for those who do not have access to it through personal or institutional purchases. To some extent Green OA can be said to live outside the marketplace, as the cost of creating the material is subsidized by the purchasers of the same material through the traditional system.

Gold OA, on the other hand, is very much market-based, but it was conceived to exist in a different kind of marketplace from the one for traditional publishing. Gold OA is “author-pays,” that is, there is a fee paid by the creator of the work or the creator’s sponsor to produce the work. This is the diametrical opposite of the “user-pays” model of traditional publishing, the economic model we all participate in when we purchase a textbook in a college bookstore or a digital edition of a mystery novel on Amazon or, if we are librarians, when we subscribe on behalf of our institution to a journal or magazine. Thus for Gold OA, the customer is not the reader but the author, and the purveyors of Gold OA services work diligently to appeal to the author.

This brings us to the first natural limitation of Gold OA, namely, that it is susceptible to devolving into vanity publishing. This is a charge that advocates of traditional publishing make all the time, and it is not without merit. The flip side of the “predatory publishers” that Jeffrey Beall has brought to our attention could be said to be the “predatory” author, someone with nothing to say but who pays to say it anyway. Calling these authors “predatory,” however, would almost always be unfair. It would be more accurate to say that some authors, who find the traditional venues closed to them for whatever reason, are under enormous pressure to publish to meet departmental requirements. Such pressure can result in desperation, and there is no shortage of services that bill themselves as Gold OA publishers that are ready and willing to take their money. Beall is doing the community a good service, in my view, by beginning a process of identifying good and bad Gold OA venues. Vanity publishing exists on the borderline of Gold OA publishing, defining one of its limits and limitations.

As a practical matter, however, vanity publishing is a much less serious problem than many suppose. For predatory publishers we have to be on our guard, and we thus should all congratulate Beall for his work, but for truly predatory authors the solution is simple: we don’t read them. Thus, sophisticated readers safely ignore the paper that proposes to use household plumbing as an information technology network (the now-defunct Red Herring technology magazine published such a piece several years ago as a hoax), we disregard the essay on telekinesis, and we log out when we are instructed to study the cultural links between Celine and Celine Dion.

Knowing what to ignore is another matter, however. And here we come to another of the natural limits of Gold OA publishing, the need to assert a publisher’s brand. This may sound like hooey to the many people who argue for “article-level metrics,” but without a reliable brand, readers could be subjected to authors of doubtful merit whether they could be classed as predatory or desperate. The protection we have against this is an evolving set of best practices for peer review (coupled with the brand that sponsors the peer review). Peer review, whether of the full-bodied kind practiced by such established journals as Nature and The Lancet or the scaled-back variety championed by PLoS ONE, nudges the least promising authors out of our line of vision.

Thus Gold OA is defined not only by the network technology that enables it to facilitate communications but also and more importantly by the human network that sits atop the IT network, whose job it is to exercise human judgment. We can call this another natural limit of Gold OA, that it is not a technology business but an aspect of human affairs (and in this respect not unlike its counterparts among traditional publishers).

A more serious limit of Gold OA publishing is that it works for some fields and not for others. The reason for this is the economic model. For an author to pay for publication, the author must have the money. Researchers working in areas rich with grant money (e.g., life sciences) can put publication fees into their grant budgets, but woe to the scholar of Chaucer or Prester John. There have been many attempts to come up with low-cost ways to attract humanities scholars to OA services, but to date none has achieved the critical mass of, say, arXiv or PLoS ONE. The funding problem for Gold OA services in the humanities could be solved in one stroke if a consortium of universities or the federal government were simply to decide to underwrite the operation, but in the current fiscal climate in the U.S., this is improbable. We are living in the period of the Tea Party Academy, and that helps to set one of the limits of Gold OA: it is a publishing model for the rich disciplines.

The fact that OA has grown up around research articles is not an accident. Putting aside the hostility many librarians have toward high-priced scholarly journals, research articles are brief enough not to require large capital investments and often part of a fast-moving flow or conversation about a particular topic. Compare this to a monograph on the Roman Empire in the first century BCE or a study of the evolving reception of Boccaccio in the English-speaking world. I spend a great deal of my professional time studying how much it costs to creat an article or a book, and the cost of book-creation is far, far higher than most people suppose, even if the publisher is not paying an author a large advance. When all costs, including the appropriate allocation of overhead, are taken into account, a book could require an investment of as much as $50,000. Some people have put that number lower (you will hear figures as low as $15,000); most put it around $25,000. For my purposes here, it doesn’t matter which end in the range you determine is closest to the truth, as even $15,000–or $5,000, for that matter–is a very big number when the economic model is Gold OA.

And here we see a very important limit for Gold OA: it is very hard to implement for works that are longer than an article. This is because the author has to pay for everything, whereas in the traditional model, the costs are shared by all the customers. Some journals charge as much as $5,000 to make an article OA; PLoS ONE charges $1,350. Those figures are a fraction of what it costs to make a book, even if the book is published only in a digital edition. (As a rule of thumb, the cost of print comes to about 20% of a publisher’s net receipts. Many suppose that this figure is much higher.) For Gold OA to fully embrace long-form scholarship, it is going to have to come up with some extraordinary innovations to lower costs.

We should spend a minute on the cost structure for journals to see what limits it imposes on Gold OA. In a recent excellent article, Andrew Odlyzko noted that the average article published under the traditional system garnered revenue of about $5,000. He reached this figure by dividing the number of new articles published each year into the total revenues of the journals industry. (Interestingly, Elsevier came in just slightly above the average.) There is a lot that is squishy about that figure (using new articles leaves out the revenues and costs of managing backfiles; the average varies widely by discipline; what constitutes an article?: etc.), but it’s useful as a guideline.

With PLoS ONE charging a mere $1,350 per article, there is a big gap to close: $3,650. Where will that money come from? We know it can’t come from the authors, many of whom struggle to find the money even to pay a fee the size of PLoS ONE’s. Eliminating print won’t close the gap, and even if it were eliminated, the gap is too large. Some people would argue that much of that $5,000 is profit (hiss), but even PLoS ONE operates at a surplus. The fact is that the gap cannot be closed without tossing out other things that we associate with journal publishing.

PLoS ONE managed to lower its costs (and to operate at a profit) by changing the nature of editorial review. This is a provocative point, but for PLoS ONE and many other Gold OA services a keydecision was to review material not based on its importance or originality (the hallmark of a traditional journal) but merely on its methodological rigor. This has the practical effect of increasing the acceptance rate from the neighborhood of 30% to somewhere around 70%, which in turn more than doubles the revenue without significantly increasing the costs. Many Gold OA services also drop copy-editing as a way to lower costs even further. This is a limit of a different kind, presenting a challenge to the author who is not a native-English speaker.

Thus one of the limits of Gold OA is that it cannot sustainably practice the form of peer review and other editorial oversight associated with traditional journals. Is that a good or a bad thing? It depends. If you subscribe to the view that the authoritative model of traditional publishing is a good thing (as do most tenure and promotion committees), then it is a bad thing. If you think that this model should be challenged, it is a good thing. For my part, I think it is a different thing and that comparing Gold OA publications to traditional journals is adding apples and oranges. Why can’t we have both?

Although the benefits of OA publishing are broadcast regularly (speed to publication, free access to disadvantaged people, the establishment of community-based forms of review, the availability of texts for large-scale data-mining, etc.), the limits are less frequently identified. But Gold OA has them, and they include not being able to provide services for all disciplines, difficulties in working with longer texts, disadvantaging scholars whose primary language is not English, a need to attack the cost structure and the editorial regime that is associated with it, and, most importantly, the requirement of a human factor to resist submissions by inferior authors and the need to assert a brand to reflect the presence of that human factor. I don’t see that any of these limits are a reason not to support Gold OA publishing, but they do argue for continuing to support traditional publishing at the same time.

What we need to minimize these limitations, or at least to understand them better, is to study them and to talk about them. There is a place for an online review or multiple reviews of OA services, for which Beall’s work is only the beginning. PLoS should be put under the same scrutiny that we now see for Elsevier. This is not to denigrate Gold OA publishing but to improve it. The practices of OA publishing should be treated in the same way as the articles in OA publications–that is, openly.