A recent article in the Chronicle of Higher Education sought to explore what the authors of the 10 most-downloaded Sci-Hub articles think of Sci-Hub. Aside from cherry-picking its facts (the journalist interviewed three authors out of dozens involved in the papers), the question itself is a red herring. After all, because publishers assume financial risk for scholarly and academic authors, the economic answer is obvious a priori — authors should have little problem with piracy of their material if said piracy might increase the number of people reading their work. After all, they suffer no harm from this.

It’s like asking publishers if they care that academics are having their bank accounts hacked — the question is not relevant because it’s not their money at stake.

Money has been a central and simmering issue in the access debates, including the fact that research authors are unpaid. In the Chronicle article, the perceived unfairness of unpaid research authors is referenced in two sentences regarding an author named Pober:

Mr. Pober says he doesn’t mind that many people download his paper free since he didn’t make money from its publication. In fact, like most academics, he paid to submit his article.

(Two small corrections for the journalist — Pober is an MD/PhD, but is referred to as “Mr.” per Chronicle of Higher Education editorial policies. I doubt Dr. Pober paid to submit his article as submission charges are rare. Instead, he most likely paid some fee after acceptance.)

The issue came up again recently in a Bloomberg Views overview of Elsevier’s acquisition of SSRN, as the writer tried to explain the economy of research publication:

The university-professor authors, editors and referees of the journals, meanwhile, usually receive no monetary compensation for their work.

It was also raised in a comment this week on a Kitchen post.

The implicit complaint each time someone mentions paying research authors in passing is that not paying them is unfair. To generate this effect, the idea of paying research authors is presented as if it makes perfect sense and would be normal. But I’ve never seen the idea and its potential consequences explored at length, and normal is a relative measure.

Of course, most of the readers here know it’s not unfair or abnormal. Authors have the economic relationship to publishers in which they are not paid, and in many situations pay fees prior to publication (color charges, data charges, submission fees, page charges, and/or APCs), because their financial risk is eliminated and their rewards for publication are indirect but significant — by publishing, they lay claim to their findings, making them rivalrous; by publishing, they claim priority over potential competitors; and by publishing, they show their employers and authorities in their field they are not shirking. Publish a paper in the right journal or earn a stellar publication reputation, and you can go far. Many careers have been made in this way.

Putting aside the fact that Pober and his co-authors likely benefited indirectly from a strong research effort and well-cited and popular paper in a very good journal — getting more grants, more budget, better postdocs, better facilities, more speaking invitations, more press coverage, more influence, a greater reputation — let’s pursue a scenario in which they also would make money directly from their paper’s publication.

The strawman question for today: What if publishers paid research authors to publish, on a widespread basis, and in amounts that would be meaningful?

For payments like this to make sense in the long-term, the positives would have to outweigh the negatives. The question becomes: Do they?

There are some potential positives — for instance, authors might be more willing to defend publishers against piracy sites like Sci-Hub, and might understand publishing economics more completely. Authors might be greater advocates of driving traffic to their papers (although incentives and abilities might remain mismatched here). The PR problems publishers have faced over the past 15-20 years might be blunted if academics saw publishers as a path to self-enrichment. It could also quiet the voices crying “exploitation” to some degree, as payments would appropriate authors into the financial upside of publishing.

But now we have to look for the downsides.

For the sake of this part of the discussion, I’ll avoid a standard royalties model, as it is overly complicated and unlikely to be implemented on a wide scale. There is an example of this. In 2008, Cold Spring Harbor Laboratory Press began paying authors and editors royalties for one of their journals (CSH Protocols), setting aside 10% of subscription revenues and then dividing this up based on the share of usage each author received. Authors received about $300 for papers getting the highest usage, while some received as little as $3 (publishing at the end of the year meant low usage, so low royalties for that year). If they received less than $25, Cold Spring Harbor rolled the payments over to the next year. Protocols typically have few authors, and the program apparently didn’t generate many more papers for the journal, according to one of its founders. The amounts certainly were below the $600/year threshold that would have required Cold Spring Harbor to provide the royalty recipients with 1099-MISC tax forms and file these earnings with the IRS. The program has since been discontinued. In its place, a one-time honoraria has been substituted, something not unusual for commissioned works in journals (e.g., review articles). According to those who inherited the program, the administrative burden of a royalty program was too high given the small amounts being paid out.

Since the royalty model is unrealistic, I’ll instead focus on a per-article fee to authors, as this illuminates many of the downsides that would arise in either case, while allowing us to do some straightforward math.

The first downside — paying authors in a meaningful way would probably kill Gold OA and CC-BY in their tracks and move the entire industry fully back to the subscription and licensing model, with stricter copyright enforcement and authors more directly appropriated into copyright enforcement. Piracy would not be tolerated throughout academia, and payment terms would probably require authors to control distribution of their works beyond the publisher’s own channels. Authors would be more reluctant to share reprints without payment as this could imperil their reputation for future payments, which would mean that ResearchGate and Academia would probably wither and die. You only need to think a moment, and these effects become clear — publishers would have more control over author behavior in general having paid them. (Note: Some people may view some of these effects as positives.)

But let’s continue to explore the idea of paying authors beyond this one major set of consequences.

One important baseline difference in academic and scholarly journal publishing is that the sheer number of authors we deal with far outstrips the number in trade publishing, either book or magazine. A single research paper in some fields could have more authors than exist on a book publisher’s entire author list or a magazine publisher’s entire cadre of writers. This is a key difference.

Because of the scale generated by large author lists — lists that are growing longer with each passing year — a major and clear negative is that aggregate expenses across the board would rise, as more money would be needed in the system than before in order to pay authors while leaving other aspects funded as they are now. Because the scenario of authors being paid would cancel revenues from author payments to publishers, payments would trigger two sources of price increase — the costs of paying authors, and the costs of recovering revenues no longer coming from authors. Estimates aren’t easy to make, but to sketch a model, let’s assume payments to authors would be large enough to be personally significant, to serve as an incentive.

Identifying who would get paid would not be a simple matter for many papers. There are many types of “authors” for scholarly and scientific papers. There are contributors, first authors, last authors, and research team participants. There are authorship groups that sometimes number in the thousands. Creating a payment system might lead to a great deal of negotiation around payment contracts, authorship roles, the authorship list itself, and so forth. This could slow the production of papers, the publication of results, and the advancement of early career scientists who would be negotiating from a weak position to be included on paying papers.

Let’s assume, for simplicity’s sake, that each author on a scientific research paper receives $100 for getting a paper published in a journal. For some articles, especially in astrophysics and epidemiology, there can be more than 1,000 authors per article, with an upper bound of more than 3,000 listed authors in some cases. This means that a publisher would have to pay as much as $100,000-300,000 per article and cut 1,000-3,000 checks. Let’s pick the midpoint, and assume the model’s paper has 2,000 authors. The administrative costs for one of these papers would be astronomical (pun intended), while the consequences aren’t clear. Would authorship lists shrink to incentivize publishers to prefer one group’s work over another’s (because it would cost less in author payments)? Would authorship lists grow, so more people could get on the gravy train? Would the model last long, or would it move to a flat per-paper payment model (discussed below)?

The per-author approach would make fields with large and collaborative authorship groups far more expensive for institutions and individuals paying to access content, while each individual author would receive a token amount at most. How much would prices rise in order to support these mini-payments to authors? Right now, I would speculate that an astrophysics article can be published for at most a few thousand dollars in costs to the publisher. Imagine that small amount ballooning by $200,000 or more. Imagine what carrying those costs in author-intensive fields — astrophysics, microbiology, epidemiology, geological sciences — would do to library budgets.

Let’s continue the scenario, where to the new $200,000 in author payments, we add the administrative costs to deal with checks, snafus, and disputes, which I’ll peg at a normal 30% overhead. Now we’re adding another $60,000 in processing costs. So an article that might have cost $2,500 in expenses for the publisher to publish now costs $262,500 to publish. Assuming 1,000 articles per year for a robust astrophysics journal, an expense line of $2.5 million explodes to expenses of $262,500,000, and this does not take into account the previous payments from authors now foregone, which may be another $1 million in a year. That’s at least $260 million in expenses the publisher would have to absorb — costs that would certainly be passed on to libraries, subscribers, and others. Estimating 5,000 institutions for a rough estimate, that would be a greater than $50,000 price increase for one title. Clearly, that is not going to work.

Astrophysics has an extremely robust and collaborative authorship community, so let’s move to an area where fewer authors typically work together — biomedicine. Here, let’s assume an average of 12 authors for the sake of discussion to see what problems might occur in groups this small.

Biomedicine is also collaborative, and multi-center trials are not unusual. First-author position and last-author position mean a great deal. Multi-national collaborations are fairly common in this and other fields. Ethical issues are more front and center.

For this, let’s write the story as it might occur a few years after author payments have become the norm involving a paper that’s slightly above average in size, and slightly more complex than average in composition:

This illustrates a number of potential scenarios and responses to incentives. But did this seem like a set of decisions an editor should be involved in? Does it help or hurt the evaluation of the science?

Let’s assume a milder model, in which there is a flat per-paper payment that the authors themselves have to divvy up in some manner, leaving the editor and publisher out of it. Let’s assume the payment is $1,000/article. At approximately 1.5 million published articles per year, the financial impact would be $1.5 billion on the industry. With the industry estimated to be between $10B and $25B in total revenues, we’re talking a 9-15% tax on the system to pay authors. Assuming publishers pass along that tax, institutions would face a 9-15% addition bundled into an annual price increase for a site license. This would not go down well, and would cause ruptures in many library budgets if generally applied in short order. But every publisher would have to agree to do this, which runs into anti-trust issues immediately. For one group of publishers to do it and other to not would create chaotic market dynamics — potentially better papers for some, but a raft of cancellations as the costs hit the market. We are in an era of stringent budgeting. How this stalemate would be broken — a standoff between author payments and risks involved in recovering the fees — is unclear.

There is also the very real possibility that author payments would become another mini-economy within scholarly publishing. We can see this in other industries, where each point of the transaction trail adds costs (credit card fees, agency fees, transaction fees, courtesy charges, and so forth). The $1.5 billion gross estimate could increase another 15%, easily.

In the case of the 15-author biomedical paper, each author would receive about $67. For the 2,000-author paper, each receives 50¢.

Either per-paper model may encourage more “salami slicing” of results, which occurs probably too frequently already — this is the practice of squeezing as many papers as possible out of one study. Imagine this behavior incentivized with money as well as publication.

But let’s assume the market dynamics make this an irresistible change, and it soon sweeps the industry. At $1,000 per article, the costs might be absorbed in a few years, if library budgets could grow to match (we’ll ignore the effect on tuition and fees for this exercise). But how long does the flat-fee scenario last? And what does that do for/to authors?

Soon, a natural market might emerge, one that would defeat the flat-fee model. Some papers are worth more than others, some authors are worth more than others, and so forth. Rather quickly, you’d have bidding. The bidding would have two sides — journals bidding for authors/papers, and researchers bidding for authorship. These both already occur as journals woo authors with promises of priority handling, good placement, cover positioning, and bells and whistles, while researchers ask to be added to papers (or are asked to join papers — in medicine, some authors are already paid by sponsors to do this).

While competition for papers occurs in the current reputational system, we find it distasteful when a senior author is paid to sign onto a paper (as noted above, it does occur). What would a senior author ask to receive in order to join a promising paper and lend gravitas? Would this researcher’s participation increase the bargaining position of the authors to drive up their payments from the publisher? How long do these negotiations take? Is more science published more quickly, or does the system slow to a crawl as everyone is now focused on their immediate financial position?

The bidding war is an interesting item to examine more closely. It brings to mind the story of Jack Andraka, who won a science prize for an approach to pancreatic cancer screening, which he then leveraged into grants and celebrity. No paper was published, his approach was found wanting (and non-novel) as word spread, and nothing substantial other than fame and noise came from it. But at one point, he might have benefited from a bidding system, pocketing a high fee based on unsubstantiated claims. How would a bidding system avoid these problems? How “in the blind” would the bidders be? Would they only be given the author names, the grant number, and the abstract, then asked to bid? Authors would be tempted to pimp their papers, as they would gain notoriety for being pursued. What if the winning bidder found fatal flaws in the paper? Could they get a refund? What kind of retraction would this be?

This scenario also brings up the ethical problems already extant in scientific and scholarly publishing — plagiarism, exaggerated claims, fraud, image manipulation — and seemingly amplifies them. After all, adding an incentive to an already heady pile of incentives would promise to only bring out more bad behaviors.

It’s entirely possible that new disclosure rules and limits akin to the Sunshine Act would be developed. Readers may come to implicitly trust journals that don’t pay authors to publish, an inversion of the current situation in which some today trust subscription journals more than Gold OA journals. To many, money is a barometer of ethical purity.

There is also the issue of US government researchers, who certainly could not accept payments. This likely goes for other governments’ employees.

Which publishers would be better-positioned to bid for papers? Large, multi-national publishers — they have scale, work in multiple currencies, and have deeper pockets to withstand an extended bidding war. Which publishers would have an easier time dealing with the administrative overheads of such a system? Large, multi-national publishers. There are already many forces working toward large-scale consolidation of journals and books under the auspices of large, multi-national publishers. Paying authors could add another.

Even if all these issues could be addressed, how much could we pay authors before other complications and costs emerged? Remember that in the US, any individual receiving more than $600 from an organization in any year is required to submit a tax form prior to payment and file a 1099-MISC to the IRS at tax time. Publishers would have to collect tax documents beforehand, and issue 1099-MISC forms for their authors to pass along to the IRS. This amount of paperwork and overhead, with dozens to thousands of authors per paper, would be difficult and costly to maintain. Errors would occur, forms would be lost, etc. And gathering tax forms from all the authors prior to publication, while certainly possible, seems unlikely to speed research along the path to publication.

In fact, it may be that after exploring the potential to get paid, some researchers would begin to prefer publishers who didn’t pay them. After all, they would be competing only on the quality of their research, not on distracting elements driven by the payment scheme. At the same time, publishers and editors would start to prefer authors who waived their payments. The mutual benefits would be real — nobody would have tax headaches, papers would be published sooner, and the science rather than the complexity of paying authors (available budget, processing) would be the focus again. Overall, not being paid may be preferred by both parties. Because the strong incentives around publication would remain, there would still be good reasons to get works published soon and in strong journals.

We already have seen a hint of this scenario — when Company of Biologists paid peer reviewers $25, the program was discontinued when reviewers themselves asked to stop being paid. It turns out it was more trouble to arrange to receive the payment in terms of time and effort than the $25 was worth to them. It’s easy to imagine researchers feeling similarly, especially after taking the time away from their labs or wards to publish a paper.

Indirect incentives allow authors to shift risk to publishers for publishing their papers, and allow editors and researchers to focus on core scientific and intellectual issues. Paying authors would add a great deal of expense to academic publishing, while tempting authors to game the system, play the market, push the limits, and incite bidding wars.

Perhaps the current practice of “paying it forward” is better for everyone all around.