First thing we do, let's kill all the authors. On subverting an outmoded tradition (Force2015 talk)
This is a rough approximation (with some esprit d’escalier) of my speaking script from my talk at the “Credit where Credit is Due”: session at Force2015, January 13, 2015. We were asked to be controversial, so I tried to oblige.
Contents
Introduction
I’m not sure that this paper is going to introduce anything new to the discussion of authorship issues, perhaps just raise some reminders couched in the terms and methodology of a discipline that is only beginning to grapple with problems natural scientists have had to deal with for years. I’m also not going to solve anything, but rather walk through the origins of the problem and propose some potential avenues for change. But I’m also not going to be discussing tweaks or improvements to the system. Instead, I’m going to be arguing that our current author attribution system for scholarly and scientific publications is fundamentally broken and that the only route forward is sabotage.
Would we create scientific authors if they didn’t already exist?
The question I’d like to begin with is the following:
“If we didn’t have the concept of the scientific and scholarly author, would we create it?”
The answer, I think, is that we would not.
The International Council of Medical Journal Editors’ definition of authorship vs. a traditional dictionary definition
This is because what we currently describe as a scientific author actually looks nothing like almost anything else we would describe using the term “author”—as you can see if we compare the definition of scientific authorship as described by the International Committee of Medical Journal Editors and a relatively standard definition of regular authorship taken from an online dictionary:
A typical dictionary definition: Author, n., A writer of a book, article, or document.
ICMJE definition The ICMJE recommends that authorship be based on the following 4 criteria:
* Substantial contributions to the conception or design of the work; or the acquisition, analysis, or interpretation of data for the work; AND
* Drafting the work or revising it critically for important intellectual content; AND
* Final approval of the version to be published; AND
* Agreement to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved.
….
Contributors who meet fewer than all 4 of the above criteria for authorship should not be listed as authors, but they should be acknowledged. Examples of activities that alone (without other contributions) do not qualify a contributor for authorship are acquisition of funding; general supervision of a research group or general administrative support; and writing assistance, technical editing, language editing, and proofreading. Those whose contributions do not justify authorship may be acknowledged individually or together as a group under a single heading (e.g. “Clinical Investigators” or “Participating Investigators”), and their contributions should be specified (e.g., “served as scientific advisors,” “critically reviewed the study proposal,” “collected data,” “provided and cared for study patients”, “participated in writing or technical editing of the manuscript”). (Emphasis added).
In other words, while in the world outside scholarly and scientific communication, we normally think of the author as the person who actually does the writing, in the the world of research communication, it is entirely possible to have writers who are not authors and authors who are not writers. And that, it seems to me, means we are fairly deep down the rabbit hole.
The nature of the problem
There has been a lot of excellent work why our definition of authorship in research communication is the way it is, by Michel Foucault, by Roger Chartier, Mario Biagioli, Mark Rose, and others (see especially Biagioli and Galison, eds., Scientific Authorship: Credit and Intellectual Property in Science). They have tied it to issues of authority, early intellectual property rights, aesthetics, and economics.
I like to think, however, that the problem really comes down to four main issues:
- The inertia of words
- The conservatism of aesthetics
- Economic utility
- Lack of scalability
The inertia of words
The first major problem with scientific authorship, in my view at least, is that our practical definition is changing faster than our term’s connotative implications.
That is to say, while it is entirely possible for us to adapt and bend the term “author” to match our current scientific practice—even if that scientific practice results in such abnormal beasts as the “writer-who-is-not-an-author” and the “author-who-is-not-a-writer”—we cannot as easily let go of our term’s original connotations. Somewhere in the back of our minds, we still believe that authors should be writers, even if our heads and contemporary practice tell us that this is simply neither reasonable nor practical for our biggest projects in the age of Big Science.
We can see that this is so, indeed, if we read through the rest of the ICMJE definition, to get to the bit where they discuss how their definition of authorship should not be abused in order to unreasonably exclude participants who deserve credit for authorship by denying them opportunities to participate in the writing and editing of the article:
These authorship criteria are intended to reserve the status of authorship for those who deserve credit and can take responsibility for the work. The criteria are not intended for use as a means to disqualify colleagues from authorship who otherwise meet authorship criteria by denying them the opportunity to meet criterion #s 2 or 3. Therefore, all individuals who meet the first criterion should have the opportunity to participate in the review, drafting, and final approval of the manuscript. (Emphasis added).
There are two things significant about this proviso. The first is the nature of the abuse that the ICMJE is attempting to prevent: the case of somebody being denied authorship credit on the basis of the second and third criteria (i.e. because they were prevented from particiating in the drafting of the article or were not given a veto over its contents) despite the fact that they met the requirements of the first and fourth criteria (i.e. made “substantial contributions to the conception or design of the work; or the acquisition, analysis, or interpretation of data for the work” and agreed to be held “accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved”). In other words, the ICMJE is worried that “the writing” might be used as a technicality to deny authorship to those who made otherwise major contributions to the science on which this writing reports.
But the second thing about this proviso is that it doesn’t protect against the opposite eventuality—that is to say that somebody who did participate in “the writing” might be unfairly denied authorship credit because they were prevented from making substantial contributions to the design or performance of the research or because they were not allowed to claim responsibility for the work. In other words, the ICMJE does not (for obvious reasons) think that preventing somebody from “designing the experiment” might be used as a technicality to deny somebody scientific credit. Or again in other words: while the ICMJE is prepared to accept that somebody could be deserving of authorship if they were unfairly denied access to the writing and editing, they don’t think the same thing about somebody whose only participation in the article was, well, “authorship” in the sense that everybody but academics understand the word. In scientific authorship, writing is a technicality in a way that participation in the actual experimental design and conduct is not.
The conservatism of aesthetics
This brings me to the second cause of our problems with the concept in research communication: the conservative nature of aesthetics. Because the connotations of the word are so strong, even if our practice has bent the definition almost completely out of shape, we also have a strong aesthetic feeling for where authorship attribution has to go in a scientific article: on the “byline,” between the title and the abstract or introduction. Indeed, a surprising amount of criticism of (and resistance to) “author inflation” rests on the simple idea that it looks ridiculous if you have eight hundred or a thousand authors between the title and the abstract of a scientific article—something that has affected even our opening Keynote speaker, Chris Lintott, in his attempts to accurately credit the profession and citizen-scientists who participated in his experiments with crowdsourcing.
Economic utility
The third reason why it has proven so difficult to let go of the idea that a scientific author must be a “writer” in some way has to do with economic utility. As the great historians of this topic have demonstrated, the decline in anonymous authorship came about in large part through the efforts of booksellers and publishers to create a mechanism for asserting copyright. If a work is anonymous, then it belongs to nobody (or everybody). If it has an author, then this author can alienate their rights to a publisher in exchange for money and their identity can be used subsequently both to brand the works in question (i.e. “another play from the great Shakespeare”) and identify fraudulent reproductions.
Although the situation in scholarship is not exactly analogous, the basic idea still pertains: naming the author of scientific works allows us to use such work as a mechanism for calculating economic value and reward and assigning scientific responsibility. In the specific case of academic researchers, authorship become something could count and use to make comparisons between individual researchers (your worth, of course, rises with the number and prestigiousness of your articles) and something you could use to certify against fraudulent science.
Scalability
And finally, there is the issue of scalability. There have always been epistemological differences between creative/literary and research/scientific authorship. The one is an act of creation (you create rather than discover poems), while the other is an act of discovery (good scientists discover rather than create their results). But the importance of these differences was obscured when both types of authors collaborated in similar sized groups (i.e. of one or two at most). In the age of single author science and scholarship, it was easy to see an equivalence between “doing the science” and “being the writer of the paper” since both activities were usually performed by the same person and because individual experiments could be completely reported in at most a few individual papers.
But this equivalence does not scale. As experiments got bigger and more complex, as projects began to involve many more experiments and participants, and as individual papers increasingly began to report on smaller fragments of the total scientific output of a project, this rough equivalence between “person who wrote the paper” and “people who did the science” became increasingly untenable and you end up with problems like that the ICMJE proviso quoted above is trying to head off—the case of participants in a project being denied access to the byline of an article solely because they weren’t given a change to wrangle a few words in the article reporting on their work.
Why words can hurt us
The point of all this is to show that the real cause of the “authorship crisis” is not the ever-increasing number of authors, but the fact that we are using the wrong word to describe what these people are and identify the quality we are trying to capture. And, unfortunately, that we are using a word that brings with it a lot of connotative and conceptual baggage that we simply cannot easily get rid of. While the term “author” (with all its connotative history) makes sense as a way of describing creative writers, for whom the actual act of composition is their primary activity, it simply does not work as a way of describing scientific authorship, for whom the act of composition is, ultimately, secondary to the research that precedes it. Poetry does not exist unless it is expressed in some form and its expression is in some sense at least coincidental with its composition (you can’t think of a poem without thinking of the words you will use to tell others about it). But while science ultimately requires communication, this communication cannot occur without prior activity: you can (and indeed probably should) do scientific research before you have decided on the precise words you are going to use to report on your results.
(And as a brief aside here, it is worth noting that science is not the only field of endeavour in which writing is secondary: the same is true, for example, in the legal or policy worlds where the actual writing is less important than the goals behind it, except that, in contrast to scientific credit systems, we don’t have any problem in these worlds in distinguishing between those who develop and those who merely draft legal bills or policy documents—there is no rule in parliament or congress that says that MPs or Senators can only be listed as the authors of a bill if they participated in its drafting).
So what to do?
This brings us to the problem of what to do. If the term “author” is bringing with it too strong a set of connotations to allow us to accurately capture what we want to capture (which is presumably participation in science rather than participation in typing), what can we do to change the current situation?
Accept that we don’t actually care who wrote the article
The first thing we need to do is accept that our current concept of scientific authorship is both an abuse of the term and brings with it far too much unhelpful baggage. That is to say, we need to recognise that we don’t actually care all that much about who wrote the scientific articles we read—beyond perhaps in the limited sense of making sure that those who did write out the results are rewarded for that writing. What we are actually trying to capture with our current credit/reward system is not participation in writing, but participation in communicated science.
This recognition is important, if for nothing else, in that it should free us of our aesthetic objections to long author lists. As long as we think that scientific authorship is actually about writing, then aesthetic objections to long authorship lists remain at least somewhat valid: it is indeed absurd to think that 800 people could possibly be responsible for the composition of a 6 page physics article. But if we stop thinking that what we are trying to capture is who wrote the article instead of who did the science reported on in the article, then the aesthetic objection becomes far less significant: if the people being credited are not actually authors, then we can stop thinking of their names as belonging on the byline; or we can stop thinking that the “byline” on a scientific article is in any way analogous to the byline on a newspaper article or novel.
Recognise that “authorship” is really just a special form of acknowledgement
Once we accept that scientific authorship systems are not actually about who wrote the article, it becomes easier to understand the next conceptual adjustment we need to make: recognising that “authorship” in a scientific sense is really just a special form of acknowledgement—or, in more concrete terms, that the “byline” in an article is really just an arbitrary, privileged, “above the fold” extension of the acknowledgements section.
You can see this if you compare the case of scientific authorship against that of poetry. Both books of poetry and scientific articles name authors and, commonly, have an acknowledgements section. The difference, however, is that where there is a clear epistemological difference between those mentioned in the byline and acknowledgements section in a book of poetry, there is (despite many attempts to develop one) no such clear demarcation in a scientific article. In a book of poetry, one will often find the author acknowledging the help of librarians who aided them in finding specific reference works, friends who hosted them while they were writing, thank yous to their agents and fellow poets for support, and so on. While these are all arguably people who helped the poet do his or her work, there is still a pretty clear distinction between helping a poet compose and actually composing yourself: nobody thinks the bartender at the poet’s favorite watering hole is actually an acknowledged coauthor of the poetry itself (well, not seriously, at least).
The people acknowledged in a scientific article, however, are, for the most part, those specifically responsible for conducting the science upon which the article is reporting: the people who did the calculations, who designed or ran the apparatus, who built the detectors, and so on. These are generally not people who had a purely social connection to the scientific work in the article but instead were directly responsible for its results. Our hypothetical poet would probably still have been able to compose poetry without the assistance of his or her agent. A scientific “author” would have nothing to write about if it were not for the people who helped make the discoveries in the first place.
This means, in turn, that the authorship-acknowledgements distinction in a scientific article is different from the similarly-named distinction in literary contexts. In contrast to the quite concrete distinction between the “person who composed the work” and “people who assisted the composer” we find in a literary work, in a scientific work, the distinction between “named as author” and “acknowledged as helper” is far more arbitrary, despite attempts such as those of the ICMJE to come up with discriminators. Instead of being able to make a clear binary distinction between those who have primary responsibility for a piece of science and those who merely assisted (as we are, in many ways, able to do in the case of literary authorship), what we are really doing in the case of scientific authors is attempting to determine the point on a scale of participation at which we decide to stop rewarding those who participated in our research. People to the left of the line we draw get their names put in the byline and are able to use the article on their CVs; people to the right of it get our best wishes, but little else, from their participation.
Understand that this conceptual problem is not amenable to tinkering
Since the problem with scientific authorship is conceptual—i.e. we are using the wrong criteria in attempting to determine credit—it is also not amenable to tinkering around the edges. Once you accept that an author might be somebody who doesn’t write and that a writer might not be an author, you are far beyond the power of any definitional tweak to save your system. Since the problem is the fact that we maintain an arbitrary distinction between those acknowledged on the “byline” and those acknowledged in the “acknowledgements,” reforms that improve the granularity of either without addressing the fundamental problem that we make the distinction at all is going to fail. Such reforms are attempts at refining the accounting for responsibility for “the article,” when what we really need is a system that recognises both that “the article” is only a second subcomponent of the scientific endeavour and that it is participation in reported science, not participation in the reporting of science, that our reward systems should be attempting to capture.
In fact, the only solution, in the end, is to stop using “authorship” as the primary measure of scientific participation. In the age of Big Science, the article is a better indication of the success of a project than the success of any individual within that project. We will only solve the issue of credit when we stop privileging authorship over participation.
Realise that there is no opportunity for external change
Although the problem is not amenable to tinkering, it is also true that it is not amenable to fiat. Because so much currently rides on authorship credit, we will find it almost certainly impossible to change formally in a top-down fashion. As advocates of Open Access discovered in the early years, change only comes when there is a critical mass that is comfortable with the innovation, but a critical mass only develops when the change itself is already understood to work within the current system. As various people have pointed out, Academia is very much a prestige economy and prestige markers are extremely resistant to change: scientists may want to publish in Open Access journals, but they need to publish in prestigious ones—and prestige seems, in large measure, to be a function of time and familiarity.
This is where the “sabotage” comes in. If you can’t change a system externally, then the only option left is to change it from within. And if the problem that we are facing with our current authorship systems is that they force us to make arbitrary distinctions among participants, then the solution is to refuse to make those distinctions. Since scientific authorship measures the wrong thing and excludes people who should be credited as participants solely on the relatively arbitrary grounds of whether they participated in the drafting of the article, then the solution is to stop using “writing” as a criterion for authorship: in other words, move the line that distinguishes acknowledgements above the fold from those below to put all of the people whose participation made the science possible above. It is only when the byline becomes indistinguishable from the acknowledgements section that the system will have been modified to the point where we can begin to work on more granular systems of identifying (and rewarding) actual scientific participation. Because, as Syndrome argues in the The Incredibles, “when everyone’s super, no one is!”
Conclusion
“Sabotage” is a strong word, but we are actually facing a pretty fundamental problem with our current attribution system. While equating “authorship” with “scientific productivity” made rough sense in the age of single-scientist experiments (and still does, to a large extent, in the current age of single scholar humanities research), the concept simply does not scale. It is difficult to apply to even moderately large collaborative teams and it is simply impossible to apply to the gigantic teams responsible for today’s biggest projects.
The reason for this, however, is that the concept is simply wrong. When we count authorship on scientific papers as part of our evaluation of an individual researcher, we are actually counting the wrong thing. We do not, on the whole, actually care that much whether a given scientist wrote the papers involved. What we are really attempting to capture is how productive and effective that scientist is as a participant in the science that is reflected in those papers—i.e. in the communication and discovery of communicable results. This does not mean that the article itself is irrelevant to science—you can’t have science without the communication of results. But it does mean that authorship of papers (authorship in the sense of “writing”) is no longer an adequate metric of scientific participation. The PI who conceptualised the project, the researchers who designed the equipment or methods, the people who calculated and reported the results—all of these are necessary to the production of good science whether or not they participated in actually typing or editing the articles in which this science is reported. Systems that fail to recognise this, such as that of the ICMJE with its fetishisation of, in essence, typing, are ultimately not going to solve a fundamental problem that has to do with the very term we use to describe the metric.
The answer to my question at the beginning is that we would not create the concept of the scientific author as a credit metric if it did not already exist. Now that it is causing serious trouble, it is time to kill it off.
