AGAINST AUTHORSHIP

Every couple of years, the institution of scientific authorship is rocked by a fresh crisis. Collaborations in high-energy physics have subverted (or perhaps rejected wholesale) the conventional notion of authorship, listing every member of a collaboration as an author on experimental papers. This has led to a string of papers with thousands of authors (the record is 5154). Some researchers are addicted to authorship: a recent study found hundreds of hyper-prolific authors (even excluding high-energy physics) who publish more than seventy-two papers a year. It would be a substantial achievement for these researchers to read all of their own papers.

If high-energy physics has a surfeit of authors, medicine faces the opposite problem. As many as a fifth of papers in medical sciences are haunted by ‘ghost’ authors—researchers who are left off the by-line in order to hide their connections to pharmaceutical companies. You don’t have to disclose funding if you’re not an author. Others want to avoid authorship for different reasons: part of the rationale for the establishment of the Journal of Controversial Ideas is that authoring papers on some topics is unreasonably dangerous. Whether or not this is true, pseudonymous authorship raises important questions about the relationship between authorship and intellectual responsibility.

The typographical uniformity of by-lines that one finds in an academic journal masks a pretty diverse set of authorship practices. When I started asking researchers about how their (sub-)disciplines handled authorship at the beginning of this project, I was surprised by the diversity of authorship practices. I realized that some disciplines list authors alphabetically, others by degree of contribution, and that first, second, and last authors have special significance (although the meaning of these positions differs by discipline). Some of the contributions that qualified for authorship were surprising: I heard about a researcher getting an author credit by allowing a team to use their supercomputer, about numerous cases of junior researchers navigating social minefields to ensure recognition, and about parts of a paper being cut to avoid the awkwardness of assigning authorship to a contributor who was sceptical about the headline claim. I heard about numerous examples of interdisciplinary collaborations in which researchers had to negotiate about whose authorship conventions would govern their presentation on the by-line. I realized that there were numerous proposals to reform authorship including the ICJME guidelines, and the CRediT taxonomy.

The crisis of academic authorship is particularly worrying because of the central role that authorship plays in academic culture and in scientific inquiry. Acquiring the status of author is a passport to a scientific reputation, to credit for discoveries, and to employment prospects in academia. Not for nothing is authorship called the coin of academic research.

One way to try to resolve this crisis is to clarify what authorship is and develop a uniform set of necessary and sufficient conditions for authorship (or perhaps a set of separate sufficient conditions). The ICJME guidelines take the first strategy and the CRediT taxonomy the second. While this strategy has a long history—in the thirteenth century, the scholastic Bonaventure distinguished between scribes, compilers, commentators, and authors—I don’t think that either strategy is likely to be successful. If we go for fine-grained descriptions, we run into the problem that there is no shared vocabulary that describes the massively heterogeneous activities (from articulating indigenous knowledge and operating machinery to conceptual clarifications and statistical analysis) that make up contemporary science. If we go for thinner descriptions, the conditions will fail to be very informative. For example, many author guidelines appeal to ‘significant’ contributions. I think this move is suspicious: in most applications, the meaning of ‘significant’ is shifted by context to serve the interests of senior researchers.

The fundamental problem with the clarification strategy is that problems with authorship emerge not because of unclarity in our understanding of authorship, but because of inconsistency in the norms governing authorship attributions. What happens when a researcher contributes an important idea or piece of analysis to a paper, but disagrees with its conclusion? How can their contributions be recognized without forcing them to be disingenuous? How should we think about authorship in cases of radically collaborative research, like multi-site medical trials or high-energy physics, where there is no unified perspective from which claims can be justified? Do these cases require a notion of collective authorship? Or do they force us to give up on a conventional notion of authorship?

In ‘What’s the Point of Authors?’, I argue that many of the problems of authorship emerge because of tensions between the different functions of authorship attribution, and that we see this in particular in big science. Some of these tensions appear in the early history of scientific authorship: part of the reason why ‘invisible technicians’ in early modern science weren’t ascribed the status of authors was that they lacked the social credibility and gentlemanly independence required of a full participant in the scientific community (Shapin [1990]). But as research has become more collaborative, this has generated more cases in which the different functions of authorship generate contradictory predictions about who should be an author.

Drawing on work in epistemology, the history and sociology of science, and the philosophy of science, the article distinguishes between five different functions played by authorship ascriptions: First, authorship ascriptions allow us to allocate credit for the intellectual achievement embodied by a paper, divvying it up between the people who contributed. Second, they allow us to construct a speaker who can take responsibility in accordance with the norms governing scientific assertion. Third, authorship ascriptions allow readers to make credibility judgements, regarding the importance and reliability of the claims made in the article. Fourth, authorship ascriptions support practices of accountability, in which researchers are incentivized to conduct better research by being answerable for their claims. Finally, authorship ascriptions function to create a pseudo-marketplace, in which the allocation of credit incentivizes research.

Having distinguished these functions, the paper diagnoses several of the difficult cases of authorship as conflict cases where these functions make different predictions about who ought to be an author. For example, someone who has contributed to a project but does not believe its results should be allocated credit, but ought not to be held to the same epistemic norms as someone who also puts credence in the results.

Based on this diagnosis, the paper argues that we ought to replace the over-laden status of authorship with a bundle of different roles that are tailor-made to the different functions of authorship. I call this the CSWG proposal. The roles proposed are contributor, spokesperson, writer, and guarantor.

Contributor
Someone who has put in labour to the collective project

Spokesperson
Someone who is held (partly or fully) responsible to the norms of assertive speech

Writer
Someone who has put together the paper, and takes responsibility for the consistency and coherence of the whole paper

Guarantor
Someone who vouches for the headline claim of the paper, and puts their credibility up for the whole paper

Distinguishing these different roles allows us to reconstruct the authorial cultures of different academic disciplines in a perspicuous way. In the humanities, where researchers typically write alone, one person might be listed as spokesperson, writer, and guarantor. And those who have asked questions, given comments, and suggested ideas might be listed as a contributor rather than merely thanked in an acknowledgement section (for an idea of how this looks, you can look at the way the article’s by-line is organized).

In a lab-based discipline, the people who contribute labour to an experiment and its analysis will be listed as contributors (perhaps with their roles distinguished by a taxonomic system, or described by a contribution statement; see one example here), the people who write up the paper will be listed as writers, and laboratory heads (or perhaps labs themselves) might be listed as guarantors. If one person can understand the whole paper, they can be listed as spokesperson, but if it involves a division of epistemic labour, several people may be listed as spokespeople for different parts of the project. Different research cultures may want to distribute these roles in different ways, and this proposal allows researchers to represent diverse research cultures without squeezing them in to the confines of authorship.

The CSWG proposal allows researchers to clearly represent the different roles that they play, both before and after the paper’s publication. It reveals the conceptual confusions that have made debates about authorship appear insoluble. And it puts discussion of authorship in research ethics on a firmer footing, helping us to make clear what is at stake in debates about how to organize authorship, whether these debates happen at the level of a particular collaboration, within an academic journal, or across a whole discipline. The goal of this proposal is not to solve the crisis of authorship. Rather, my hope is that what I offer here makes the problems more tractable and lays the groundwork for better conversations about how authorship should be organized. With this groundwork in place, we can collectively think about how to organize our authorial practices in ways that better recognize creditworthy contributions, make researchers accountable, and promote the collective pursuit of knowledge.