What’s the Point of Authors?

When an academic paper gets published, who should be credited as an author? If I have an idea and do all the writing up myself, the answer seems obvious: I am the author. But what if I had idea, but needed some maths help with the write up? What if I had the idea during a particularly engaged late-night drinking session, but then wrote the paper up by myself. What if the idea required empirical testing, and I had to get my friend to run some experiments? What if the empirical work requires setting up a research institute involving thousands of scientists, technicians, and administrators? What if the work of having and developing the ideas was all done on a blog with anonymous contributors?

These possibilities are real: work in CERN has led to a paper with over 5,000 authors https://www.nature.com/news/physics-paper-sets-record-with-more-than-5-000-authors-1.17567 and a paper from the recent observations of the collision of two neutron stars had something like a third of the world’s living astronomers as authors http://iopscience.iop.org/article/10.3847/2041-8213/aa91c9/pdf. And, the Polymath project publishes maths papers that come out of a blog with many anonymous contributors https://en.wikipedia.org/wiki/Polymath_Project.

It bear stressing that peoples’ livelihoods can be made by inclusion on the author line, and failures of attribution can be morally pernicious, especially in cases where the excluded parties are members of underrepresented groups. Perhaps the best-known case of author exclusion is Rosalind Franklin being left off the author line for her work on DNA with Crick and Watson, but it’s easy to multiply examples. The awarding of Nobel Prizes exacerbates this exclusionary dynamic, since the science prizes can be awarded to at most three (typically male) individuals. When the prize in physics was awarded for the discovery of the Higgs bosun in 2013, it went to Higgs and Englret, who proposed the existence of the particular, rather than to the many thousands of individuals who were involved in making the measurements. The prize seems set up exactly not to deal with the kinds of large-scale projects which typically produce ‘Nobel-worthy’ results.

To allow us to properly attribute credit for collaborative work, and to understand what goes wrong in cases of excluding authors, it would be good to have a grip on what it means to be an author. The immediate problem is that different disciplines have such different approaches to attributing authorship. Rather than offering an account of what it means to be an author, I want to set out a menu of practical and epistemic functions which our attributions of authorship might play.

Labour: One first thought is that the function of authorship is to allow us to track who put their labour into the production of a piece of academic work. We care about this presumably want to have a marker to allow us to give positive feedback for putting in effort. This is the gold star for effort proposal. Of course, the immediate question is going to be exactly what kinds of labour should determine authorship. The most permissive policy — including anyone who has put any labour in at all — will clearly grossly over-generate authors, so presumably we will need to restrict either to certain kinds or certain amounts of labour.

Traditionally, there has been a distinction between intellectual and practical labour, with authorship being restricted to intellectual labour. Technicians and administrators do not — typically — receive authorship credits (although, I’m sure there are exceptions). I’m pretty sceptical about whether we can make any good sense of the intellectual labour/practical labour distinction. I’m impressed by Antonio Gramsci’s point that there is intellectual labour in pretty much any of the perceived manual occupations (Stanley’s How Propaganda Works has a nice discussion of this point in Chapter 7). Furthermore, if we privilege intellectual labour, then we will wind up with weird situations in which an academic has an idea in an afternoon, sends her technicians off to perform labour-intensive testing for years, then authors the paper they write. This ‘Renaissance painting’ model of production is pretty widespread, including in Robert Boyle’s laboratory (memorably described by Stephen Shapin in ‘The Invisible Technician’), and presumably in various contemporary natural science labs.

The other version of this proposal would be to have authorship determines not by type, but by amount of labour. Notwithstanding the difficulties in establishing a metric for labour (is it person-days, words written, number of tests performed, or what?), we would need to decide whether to have some threshold for the amount of labour put into a paper, or whether to have some sliding scale for degrees of authorship. A further difficulty is that since different papers take different amounts of labour, there will be no simple way to read across from authorship to the amount of labour put in, even in the case of single-authored papers.

Credit: A related idea is that the function of authorship is to allow us to make intellectual credit attributions. Compare intellectual work with winning running races. In both cases there is a distinctive goal — gaining significant truths, and coming first in the race — and we credit people with a distinctive kind of achievement for realising those goals — we say that someone won the race, or that they found out such and such significant result. We might think of authorship as being a bit like awarding a medal: as a formalised way to recognise credit for achieving the goal in the relevant domain.

It’s worth quickly distinguishing the credit and labour conceptions of authorship. On the labour proposal, authorship recognises the amount of effort put in to a project, irrespective of the significance of that labour in reaching the result. Authorship marks sweat. By contrast, on the credit model authorship is a kind of responsibility-judgement: being an author means being responsible (either completely or partly) for knowing the relevant result. For the most part, the predictions of the two proposals will line up, but there will be cases in which people put in a lot of labour which is not significant for reaching a result (as in the case where a researcher goes down a blind alley that takes a lot of labour, but doesn’t lead to finding out a result).

This proposal leads to some interesting predictions in the case of collective credit. Just as individuals can achieve goals (whether practical or epistemic), so too can groups. No individual wins the FA cup: the achievement is due to the team working together. Similar, in a case like the huge joint projects in CERN, we may want to think about discovering the Higgs bosun as a team achievement. How should we attribute credit for team achievements? We could list all of the people who were involved in the team. In many cases it may be complicated to work out exactly who was ‘in the team’ so an alternative way to attribute credit would be to allow ‘team’ authorship, in which a research team or centre takes credit for finding out a particular result. Rather than listing all of the collaborators, we could list the ATLAS project as the author of a particular project (much as the Polymath project uses ‘Polymath’ as a collective author for the papers which stem from its project).

One nice feature of this proposal is that we can import work from virtue epistemology about the notion of intellectual credit, and work on team credit (for example: http://ndpr.nd.edu/news/the-social-contexts-of-intellectual-virtue-knowledge-as-a-team-achievement/) to understand authorship. That said, we’d need to tweak things a little since the notion of intellectual credit used in virtue epistemology is credit for discovering truths, and many papers assert false claims (obviously).

Both the credit and labour proposals help us to see what is wrong in the cases of excluded authors. According to the labour proposal, excluded authors are failing to have their labour properly recognised, and according to the credit proposal, they are being denied responsibility for their part in an intellectual achievement. Both failures are clearly both morally and epistemically pernicious, which means that we might think of the wrong of excluding someone from authorship as a kind of epistemic injustice, or a harm done to someone distinctively in their capacity as a knower.

As far as I can see, the problem with Credit is that in the majority of cases intellectual credit will be too widely distributed to reasonably keep track of who deserves credit for what. I think that something like this idea motivates the bloating of acknowledgements sections of academic papers and books. It’s pretty common to find huge acknowledgements sections, which try to spread out the credit for the ideas in a paper or book across the people who provided comments on the paper, or asked questions at presentations. Of course, there are various different kinds of acknowledgements, but in many cases authors acknowledge that some of the key ideas in a paper may have come from other people who did not end up getting credited as authors. If the acknowledgements section is used to distribute credit across audiences and interlocutors, then the function of the attribution of authorship winds up not being to attribute credit for the results but to mark credit for finding a way to put together different ideas from various different people.

This proposal also runs into problems with reliance on other peoples’ work. If one paper puts forward an idea, and then another paper provides empirical testing for that idea, who deserves credit? Presumably, we want to say that the author(s) of the first paper deserves credit for having the idea, the author(s) of the second paper deserve credit for providing evidence for the idea, and that both papers deserve joint credit for establishing the idea. This example demonstrates that when we’re thinking about who deserves credit for a paper, we need to have a really good grip on exactly what the upshots of the paper are. I doubt that there is any general answer to the question of what the upshot of a paper is, so Credit will need to depend on case-by-case judgements about what the credit-worthy upshots of a paper might be.

Both of these worries might motivate a tweak to Credit so that claiming authorship is less like claiming credit for the key idea of a paper, and more like claiming credit for putting those ideas together into some kind of coherent whole. According to what we might think of as Writing-Credit authors deserve credit for putting together a set of ideas into a paper, rather than the credit for the significant discoveries (which might be radically distributed).

Speaker: One way to make sense of the possibility of having a single authored paper (or just a small number of authors) in cases of radically distributed credit is to switch from thinking about the intellectual processes which are involved in making a paper to thinking about what it is to publish a paper. One way to think about publishing a paper is as a special case of making an assertion, involving a special institutional context, and perhaps some fancy academic norms. In order to make an assertion, we have to have a speaker who is the agent of that assertion. This is a pretty dry logical point, but it has some interesting implications. Assertion is a social act, which is governed by various norms: plausibly the backward-looking norm that what the speaker asserts better be known (true/justified/…) by her, and the forward-looking norm that the speaker is committed to either defending or retracting her claim. I don’t see why things should be any different in academic publishing: we censure researchers for making claims which turn out to be false or unjustified (think of the reaction to studies which purported to establish a connection between vaccines and autism), and we expect authors to defend their claims by publishing follow-up papers, responding to questions in presentations and so on. So, we might think that the point of having an author for academic papers is to have someone who can play the role of speaker, with the attendant normative commitments.

This proposal is well-positioned to make sense of why it might be helpful to have just one author in cases of distributed intellectual credit. In publishing an academic paper, one typically doesn’t just make one claim (joke papers aside). One makes various different claims, presumably in the service of establishing some significant claim. One of the normative commitments incurred by a speaker is that the claims she makes be consistent (at least in the sense of not asserting contradictory claims, and perhaps also in some stronger sense of not leading to contradiction given plausible background claims). Whereas non-authors can contribute ideas into the mix while disagreeing on basically everything else, the author of a paper is committed to crafting a somewhat coherent point of view, which she can then defend. (Making a coherent point of view is itself a creditworthy achievement, as I pointed out above).

One of the interesting things about this proposal is that — unlike Labour and Credit — it detaches the question of who is an author from the question of who put in the intellectual legwork. For Labour and Credit, by the time that a paper or book is written, there is an objective answer about who counts as the author of a paper (i.e. the people who put in the right kind of labour, or who deserve intellectual credit). However, Speaker makes being the author of an academic paper more like signing up to an online petition: one needn’t have written the petition to count as one of the signatories. (this seems pretty close to the writing strategy used to produce the new paper ‘Redefine statistical significance’ https://www.nature.com/articles/s41562-017-0189-z as described in a post on the Brains blog http://philosophyofbrains.com/2017/10/02/should-we-redefine-statistical-significance-a-brains-blog-roundtable.aspx). According to Speaker, the interesting question to ask isn’t ‘who is an author?’ but rather ‘who should we bring on board as an author to allow us to best fulfil the epistemic responsibilities that come from publishing a paper?’.

Filtering: A third proposal is that the point of authors is to facilitate a certain kind of credibility judgement. As any academic who vaguely tries to keep up with even one area of literature knows, there are way too many papers being published for anyone to realistically read and digest. We face a difficult question: which papers should we pay attention to? Although there is something intellectually heroic about trying to take on every paper on its own merits, realistically, we need some quick and dirty heuristics to help us read the most important and interesting papers in an area. There are various heuristics we can use: looking at the title, and reading the introduction and conclusion usually providing a good starting point. But maybe the most important heuristic is to look at the author(s) of the paper, and to think about their epistemic credentials. Did they go to a prestigious university? Do they have a famous supervisor? Have they written other papers in good venues? Although this kind of reasoning might veer dangerously close to the ad hominem fallacy, it’s difficult to see how one could possibly avoid using some kind of heuristic in thinking about what to read, and as long as the characteristics of the author are epistemically relevant (i.e. whether they have written other good papers in the past, rather than whether they are a a member of a minority group), I don’t see too much to object to.

One upshot of this proposal is that the authorship of a paper ends up being more about lending credibility than reflecting the processes that went into its writing. On this proposal authorship by a respected member of a discipline plays a similar role to being published in a prestigious journal: it’s a thumbs up to the ideas presented in the paper.

At this stage, I’m more interested in spitballing functions than definitively endorsing one of these proposals, but I want to note a couple of quick things. I think that we can see three kinds of role played by authorship: i) being a record of the intellectual process which went into producing a paper (whether that be labour or desert for credit), ii) being a speaker, with the normative baggage that involves, and iii) providing credibility to the claims made in the paper. I think that in current practice, we use authorship to play all three of these roles. However, they are not mutually consistent. In particular, Labour and Credit make different predictions about who should be an author, and it’s easy to imagine people who don’t deserve credit for the ideas in a paper coming on board in order to help constitute an epistemically better speaker, or to positively certify the results of a paper.

I’m absolutely at the beginning of the process of thinking about this, and there are tonnes more questions to think about. How should we understand lead authorship? Do these functions exhaust the practices of author attribution? How does authorship actually get ascribed in practice?