Information wants someone else to pay for it: Laws of information economics and scholarly publishing

1. The state of scholarly communication: More and less

This year marks the sesquarcentenial anniversary of Philosophical Transactions of the Royal Society, a notable milestone upon which to reflect upon the evolution of the scholarly communication ecosystem. In the beginning, three hundred and fifty years ago, scholarly communication was a relatively straightforward proposition that entailed the collecting of individual research writings into a volume that was reproduced or published. This new method of communicating research findings to a dedicated audience was a vast improvement over the circulation of individual letters written by individual scholars on individual topics. This shift from the communicating of research findings from a one-to-one model, to one-to-many model (more accurately ‘some’ in the original instance) represented a major step forward in the evolution of what we now identify as scholarship. Centuries later, this is still the standard for authoritative scholarly communication, with nearly all developments and innovations replicating this original print model.

Based on its present state and trajectory, the future of research and of scholarly communications is “more”. Some of the components of “more” are readily apparent – there is more data being produced, disseminated and accessed than ever before with ninety percent of digital data being produced in the previous two years [35]. By some accounts, scientific publication output doubles every nine years, with one analysis stretching back to 1650 [42]. In some areas, the effect of the new data is so substantial that the entire research evidence base of fields appears to be shifting: in astrophysics which relies upon massive data sets for its collaborative global research output, CERN alone generates one petabyte of data daily; in genomics, where efforts to synthesize the human genome are reliant upon massive data sets, we know that the human body contains sixty zettabytes of data [11,47]. This doesn’t begin to address the firehose of data spilling out from commerce and industry that has become the lifeblood of disciplinary studies in business and management, in the social sciences such as sociology and economics, and law and government regularly engage with digital research objects such as big data.1 Even the more traditionally analog humanities disciplines like history and literature, which tended towards print artifacts as the only primary sources worthy of consideration, have (re)assembled under the mantle of Digital Humanities to directly engage with data-driven analysis of their respective corpuses. A more recent development is the distinction between Big Data Digital Humanities and Small Data Digital Humanities, by which the former is distinguished by one researcher as “large or dense cultural datasets, which call for new processing and interpretation methods”, in comparison to the latter, “regroup more-focused works that do not use massive data processing methods and explore other interdisciplinary dimensions linking computer science and humanities research”. However both are a departure from conventional humanities pursuits [18].

The digital impact on the scholarly communication ecosystem, which started with a spark and flash in ARPAnet, has been slowly burning and only now, decades later, is rapidly accelerating. With the increase in scholarly content produced and available, growth in the volume of publications seems inevitable, and this number has grown consistently at about three percent annually for over two centuries. A 2015 report, which identified twenty-eight thousand and one hundred active peer-reviewed journals publishing about 2.5 million articles a year, attributed such steady growth to a comparable increase in the number of researchers [43]. As of May 2014, over 114 million English-language scholarly papers were available on the Web [20] from circulating scholarly journals (and mega journals) with over ten thousand of these publications self-identifying as open access (with APC – article processing charges) in some capacity.2 And if we expand content from publications in traditional channels to a more capacious definition of scholarly communication, the impact of the digital space is much more apparent.

2. More stuff, but less competition – What’s wrong with the market?

In general, the trends in scholarly communication are more, more and more. In 2011, the value of the market in 2011 was estimated to be $23.5 billion [44]. But there is one area in which the trend is “less” and that is in market competition. Although the number of publications and journals is expanding at approximately three percent a year, and the market is expanding at four percent, the number of mergers and acquisitions over the past three decades have dramatically decreased the diversity of and competition among publishers. Today, following the recent merger of MacMillan and Springer, the market is dominated by a handful of companies: Pearson, Reed Elsevier, Springer, Taylor & Francis, Thomson/Reuters and Wolters Kluwer [15]. These companies happen to be the top four publishing companies globally as well [33]. And this is the culmination of a long-term trend: over the last three decades, there has been dramatic consolidation in the scholarly publishing industry (see for example [6,24]). Profit margins are commensurately high, with some credible estimates of Elsevier’s profit margin as high as thirty-seven percent.16

Thus far, there are no signs that the general expansion of the content, contributors, and audience for scholarly outputs has countered this decline in competition. In fact, preliminary studies of “gold” open access publishing17 indicate that the vast majority of subsidies for OA publishing are going to the dominant commercial publisher.18 There is, however, an increase in the participation levels with the variety of methods and mediums for transmitting research in real-time as it develops. This span the research lifecycle, from posting initial ideas and observations via social media and blogs, through the sharing of draft articles on preprint repositories, to the archiving of the post-print in a repository – all of this amidst continuous efforts to keep abreast of developing research via social media, listservs, and other channels of informal discourse. And this doesn’t begin to address the development of new – often digital – methods to capture research objects and results for which very few avenues exist for adjudication and publication. The current scholarly communication ecosystem has more producers, more consumers and more goods, but dramatically less competition. From an economic view, this presents a puzzle – in a well-functioning19 market we would expect to see plenty of competition whenever there are lots of goods and producers. What could explain the concurring trends of expanding scholarly content and contraction in competition?

3. Rules of practical scholarly information economics

One part of what happened seems obvious: scholarly communication became digital. But, how is digital different, and how is digital disruptive?

For the majority of its existence, scholarly publishing has been conducted, more or less, as a market-activity. One could argue that it was an accident of history that allowed scholarly publishers, specifically university presses, to operate as self-sustaining businesses through the sale of scholarship. It was possible due to the primary distribution channel for information being printed books and journals, which constituting a fixed, excludable product – a book couldn’t be on two library shelves simultaneously, nor could it be read by two readers. Publishers openly sold journals and scholarly monographs, and Universities (through their libraries) bought them. This industry was, however, tightly bound with higher education and research. In part, a market for scholarly publications existed because scholars needed to publish scholarly works in order to advance in their careers; scholars needed access to substantive collections to advance; and universities were able to support scholars through the income produced by teaching – all simultaneously.

Furthermore, scholarly publishing was never a pure market economy. From its inception, a researcher freely contributed articles to be published in scientific journals and also contributed labor for the peer review of articles and books, which was (and is) an integral, essential quality control mechanism within scholarly publication. And although a researcher was oftentimes paid modest royalties on the sale of scholarly monographs, this take is not comparable to the effort invested. However when one considers that the author is supported by the University with a regular salary, sometimes even a direct subvention to defray publication costs, as well as being given time to research and write for career advancement, it seems more reasonable.

Private philanthropists sometimes contributed to scholarship, as it was viewed as a public good. In the mid-nineteenth century, higher education, and indirectly scholarship, received a huge boost from the subsidization of “land-grant” colleges, initiated by the Morrill Acts of 1862 and 1890. And in the mid-twentieth century, governments began to substantially support Universities both through research grants, and indirectly through scholarships to students. Research funding has stimulated an increase in scholars and researchers, particularly in the hard sciences and life sciences [14,37].

Nonetheless, the historical scholarly communications quasi-market was relatively predictable: For centuries, to a first approximation, scholars contributed writing and review; publishers selected, bundled, distributed and sold content; and libraries bought. The shift to digital changes this. Advances in information and communication technology, resulting in the presence of robust and widespread digital storage and digital communications networks, radically changed the cost structure of knowledge production and communications.

Specifically, the marginal cost of taking a wide variety of actions is radically lower for digital objects than it was for their physical analogues. Digital objects are cheaper to access, cheaper to replicate, cheaper to transmit, cheaper to computer over, cheaper to modify, and cheaper to divide and cheaper to recombine. Moreover, in the case of scholarly communications, there are strong reasons, based on both history and economic thought, to believe that the market will not fix itself. We call these reasons, perhaps provocatively, the “rules” of practical scholarly information economics.

The first rule, which sums up the disruptive nature of technology [22], was proposed by historian Melvin Kranzberg:

Kranzberg explains his rule as follows “technology’s interaction with the social ecology is such that technical developments frequently have environmental, social, and human consequences that go far beyond the immediate purposes of the technical devices and practices themselves, and the same technology can have quite different results when introduced into different contexts or under different circumstances”. And although this explanation was not made with scholarly publishing or communications in mind, it could not fit better: We are witnessing changes not only in the scale, volume, inputs and forms of scholarly communication, but also to the audiences and uses as well.

Disruptions shake things up, but what about the long run? Classical economic theory addresses the ‘long run’. Given enough time, and the right conditions, markets will converge to an equilibrium. However, as Keynes pointed out, even if all the appropriate conditions are satisfied, such equilibration can take so long that it is irrelevant to forming public policy [19]. Keynes’ rule can be best summarized in his own words:

Frequent changes may keep the system out of equilibrium. And in the area of information and communication technology, the rate of change is so rapid, that frequent changes may keep the ecosystem for scholarly publication out of equilibrium for an extended period.

Furthermore, even in the long run, for markets to function properly, many conditions have to apply to buyers, sellers, institutions and goods: Francis Bator coined the term “market failure” [7] to denote the failure of a system of price-based market institutions to yield a desirable equilibrium.20 Some violations of economic assumption are minor: they lead only to marginal losses of efficiency in outcomes. Others issues however, are more serious, because they can interfere the markets reaching good outcomes – they undermine the rationale for using market mechanisms. Thus our third rule:

Since Bator’s provocative article, social scientists have investigated many potential micro-level causes of market failure in the real world (see for example [46]), including: structural information asymmetries; unstable or time-inconsistent preferences; externalities; transaction costs; bounded rationality; lack of common knowledge;21 path dependencies; increasing returns to scale; and non-consumptive goods. It could be argued that almost all of these issues have the potential to affect the current market for digital information. However, we believe that three of these potential causes are likely to be critical going forward: information contagion, network effects, and non-consumptive goods.

Information contagion, was coined by Brian Arthur and David Lane [9]. They note that when products are difficult to evaluate, buyers may rely on the experiences of other previous buyers – either through informal channels or more formal mechanisms. This feedback system can create path dependence, providing a strong and persistent advantage to the products that develop an early mass of positive feedback. This leads us to formulate a corollary to Bator’s Observation.

We conjecture that three attributes of scholarly communication makes information contagion likely and important: First, there are many choices of publications and publication outlets, which makes a complete evaluation of the options infeasible; second, evaluating the quality of scholarly publications without consuming them is hard, forcing a reliance on reviews; and third, mechanisms for review and commentary on digital publications are widespread.

We conjecture that a second critical cause of market failure in scholarly publishing are network externalities (demand-side economies of scale). According to classical microeconomic theory, economies of scale create monopolies. At the beginning of the 20th century, Theodore Vail recognized how a telephone network could lead to economies of scale, and successfully applied this insight to build the extensive and durable AT&T monopoly (see [48]).22 We state this principle in modern terms:

In scholarly publishing network externalities are created because much of the value of scholarly publication is not solely in the content of the publication itself, but in its meaning in context with a body of work, and in contributing to a larger collection of publications. Scholarly works are embedded in networks of citation, evidence and meaning that give them value, and that contribute to the value of the network. Moreover, the value of the scholarly information network to potential users has greatly increased as information becomes digitized, because it is now possible to compute over the entire network. For example, scholarly information networks are now used as subjects of text, topic, content and sentiment mining to extract new findings or concepts; to analyze the relationships among publications, funding, and evidence to evaluate theories, actors, institutions and policies; and to identify potential collaborators, and antecedents of successful collaboration.

The last critical source of market failure in scholarly communication merits its own category, because it has to do with the nature of scholarly publications as market goods. Economic theory predicts that markets will work well when the goods being sold are both “consumptive”, meaning that they are consumed with use; and “excludable”, meaning that it is possible for the owner of the good to control who uses it. These are sometimes known as “private” goods.

As Ostrom and Hess, Foray, Stiglitz [13,29,38] and have pointed out, knowledge goods are not private goods. Some knowledge goods, such as licensed databases or journal subscriptions, are non-consumptive but excludable, “toll goods”; others, such as public libraries are partially consumptive but non-excludable; and fundamental knowledge, such as that produced about the world by some basic research, is neither consumptive nor non-excludable – a “pure public good”.

Further, Ostrom’s career-long Nobel Prize-winning work focused on identifying the institutional design principles that enable the successful management of local common pool goods: She found that for local-pool commons, such as fisheries, to be successfully managed by a local community required that clear boundaries be in place, rules are matched to local context; actors participate in rulemaking; self-monitoring and graduated sanctions are available; low-cost conflict-resolution is available; external authorities respect community rules; and enterprises are nested with respect to the terms of governance, rules, etc. [27,28]. This is summarized in rule 4:

A corollary to Ostrom’s rule was articulated by the National Digital Stewardship Alliance leadership [3]. The report notes that because the digitization of information lowers the cost to access, almost every institution now relies for its business, operation, and mission on large amounts of information that go beyond the institutional boundaries. The amount of information is so great, and the risks so diverse, that no single organization can effectively ensure long-term access to all the information it needs. At the same time, for many pools of digital information, multiple institutions value it. Together, these factors imply the following:

An implication of Ostrom’s argument is that information goods, if left to the private market, will be under-provisioned – the amount of knowledge produced and used will be less than is good for everyone in aggregate. This is a one of the core rationales for government support of basic research.

4. Rebundling scholarly publications for markets and commons

It isn’t news to anyone that systems of higher education are experiencing challenges

The structures and roles within scholarly communication, including university presses, archives, repositories, academic libraries, authors, journal articles, monographs, datasets, authors, editors and reviewers are interconnected in complex ways.

In their current form and for the most part the digital scholarly monograph and journal article duplicate the form and function of their print ancestors. However, this is not a necessity – advances in digital production and communication make it possible to unbundle the content and semantics of scholarly publication, and to change the relationships among stakeholders and scholarly works. There are a handful of publishing experiments innovating on container and content in this area such as University of Minnesota Press’s Forerunners, a groundbreaking departure from the book–journal binary.24 However, none of these so far has achieved widespread adoption.

It’s not hard to think of other possible desirable features for the scholarly work of the future. One description of the ideal “article” might be as follows: a high-quality, authoritative work; expertly-selected and thoughtfully-curated; one that participates actively in the information network; connects seamlessly to primary sources; is expertly-marketed and widely-discoverable; can be accessed on and off line; offers interactivity; contains multiple pathways of “reading” and understanding; supports reader services such as annotation and institutional services like usage evaluation; is regularly-updated and systematically-versioned; is included in tenure files and wins disciplinary prizes; is durable and easy to preserve; and is produced, maintained, and disseminated through an economically-sustainable process. A contrasting model of the future scholarly “article” is a minimalistic nano-publication: an assertion – perhaps concerning evidence, theory, analysis (and possibly an annotation on another work); preferably expressed in a machine-understandable form; and is associated with a persistent identifier for the work and accompanied by the identifier of the author.

How to generate, compare and evaluate these two proposed “bundles” of features is challenging. One potential frame of reference is the literature on the scholarly publishing process: Morris Barnas and Lefrenier [25] define the primary functions of a journal as selection, preparation, collection, navigation and collection. Their guidebook further emphasizes the importance of the functions of editing, production, ex-post evaluation metrics, marketing, sales, and resolution of ethical issues through policy and process. Campbell [10] describes the core lifecycle of scholarly publishing as comprising the following phases: soliciting and managing submissions, managing peer review; editing and preparing the manuscript; producing the final form of output; publishing and disseminating; and promoting and archiving. They also provide a list of ways in which the publishers add value to scholarship: sorting and assessing research outputs; publishing primary literature, supplementary data and patents; aggregating content; distilling evidence through reference works and meta review; creating standards and seeking consensus; aggregation of content; granularization, tagging and prioritization of content; identification; application of rules; systems integration; data structure and exchange standards; content maintenance; integration of content from multiple sources; creating and monitoring behavior change; development of workflow analytics and best practice benchmarking.

We propose a complementary framework for comparing and evaluating new forms of scholarly publication. Rather than focus directly on what publishing does, the value it adds, or on the “ideal” model of publication, we propose to focus on the features relevant to how publications are, and could be used by the various actors and stakeholders in the scholarly evaluation, e.g. researchers, in their roles as readers, writers and reviewers; publishers; administrators; and scholarly intermediaries, such as libraries; funders, etc. Framing scholarly communications in terms of the actors involved and the affordances provided suggests a number of questions: What are the critical actions that stakeholders apply to scholarly publications in these use-situations? What are the use-situations that are most relevant to each group of stakeholders?

In the absence of in-depth analysis of a large sample of use case, any list of the affordances of scholarly publications should be considered preliminary – however, we can identify a subset of these: First, many of Unsworth’s list of scholarly primitives [39] may be applied directly to publications: discovery, annotation, comparison, sampling, representing (formatting), and reference.

Second, reference, identification and citation have been the object of extensive recent study in other contexts, and are clearly important use-cases for scholarly publications; e.g., stakeholders frequently cite scholarly works, and this activity is central to the process of scholarship [4,12]. A study of the uses of citation reveals affordances that apply directly to scholarly publications: attribution, identification, access, persistence and verification. In other words, scholarly publications can be cited, and through citation the work can be attributed to creator, contributors and authors; identified uniquely as the object of the reference; an instance of the work located and accessed; and the object can be verified to correspond to the citing reference through versioning, fixity or provenance.

Table 1 sums up the inventory of eighteen affordances of current scholarly publications. More can certainly be added, however our conjecture is that a publication with these affordances (perhaps with minor refinements or extensions) would satisfy all primary uses of existing scholarly publications. Further, we note that the existing affordances can be nested; for example, it is common to annotate a specific sample (selection) of a publication, rather than the entire work. More generally, these affordances may be conceivably nested to an arbitrary depth.

The shift to digital information and communication technologies offers the opportunity to unbundle the current affordances of the scholarly journal article and monograph, remix these, and extend them. Future inventors of a new form of scholarly publication could develop publications that satisfy any combination of these affordances and which could add new ones (e.g. active alerting, interaction, automatic updating, etc.).

The possibilities are beyond easy enumeration. We suggest that a productive approach will be to identify mutually-dependent groups of stakeholders (such communities of practices, actors using a specific common pool resource, vertical markets) and focus on the affordances they most rely upon. And we believe that the central challenge is not simply to invent a unique remix, or add novel new affordances, but to produce scholarly publication “products” that will not only be acceptable within a market framework, but also will be sustainable through a knowledge commons.