Authorship: time for a paradigm shift?

The authorship system is broken and may need a radical solution

Anthony Trollope, one of England's greatest nineteenth century novelists, rose at 5 am every morning, wrote for several hours almost every day of his life, and so completed more than 50 books. That was authorship. The words, characters, and plots all came from him, and his was the glory and the criticism. Producing a scientific paper is completely different. Some people conceive the study, often within a broad programme of work conducted by others. Different sets of people may design it, collect the data, and analyse and interpret them. The paper may include techniques as diverse as molecular biology and economic evaluation, all carried out by different people. The person who writes the paper may have done nothing but the writing. Who then will be an author? This becomes a matter of politics, not science. Often the powerful will be authors and the powerless ignored or simply acknowledged. We need to scrap the notion of authorship in science and try something else.

Disquiet about authorship in science has been growing for years. In the early 80s John Darsee “coauthored” papers with distinguished researchers.1 When the papers proved fraudulent some coauthors refused to accept responsibility. This was clearly unsatisfactory: authorship must bring accountability as well as credit. The International Committee of Medical Journal Editors (Vancouver group) thus drew up guidelines (see p 1009) based on the principle that each author should be able to defend the work publicly.2 Several studies have shown, however, that the guidelines are not working.3 4 5 Many “authors” do not meet the criteria. Work that we publish today from Newcastle shows that many researchers did not know of the Vancouver criteria and (when told about them) did not think them workable (p 1009).6 Most researchers had experienced problems with authorship: many had assigned inappropriate coauthorship, and many had been excluded when they thought they deserved it.

Perhaps we need further data on the problems with authorship, but all the studies so far have found problems. A meeting on the subject held in Nottingham last June also concluded that the concept of authorship was broken,7 while all the conversations I have had with researchers convince me that the current Vancouver criteria are not working.

What action might we take? One option is to publicise the existing criteria and work harder to enforce them. But the Newcastle study confirms that most researchers think the Vancouver criteria too restrictive. Furthermore, the BMJ's attempts to enforce them–by asking all corresponding authors to sign that the criteria are met by all authors and that nobody meeting the criteria has been excluded–have been unsuccessful: almost no changes in authorship result, despite our knowing that many authors do not meet the criteria.

Secondly, we could tinker with the criteria–make them clearer and, according to taste, more or less restrictive. But any system that depends on separating people into sheep (authors) and goats (non-authors) will lead to arguments and will be decided ultimately on the grounds of power.

A third, radical, response is to scrap the concept of authorship. Instead, we would have a descriptive system something like film credits and talk about contributors rather than authors. This solution was advocated by Drummond Rennie–deputy editor (West) of JAMA and doyen of researchers into scientific publishing–at last year's meeting in Nottingham7; his paper is likely to be published soon in JAMA. It should be possible for researchers to agree easily on who did what, particularly if they keep a record from the beginning. Readers can then judge for themselves the relative importance of the contributions.

One problem with the radical solution is over who will take ultimate responsibility for the study. Without a “guarantor” (Rennie's term) there is a danger that overall responsibility will be lost. Clearly the contributor who analysed the data must take responsibility for a wrong analysis or for doing it badly, but who will take responsibility when it emerges that the data were invented? The idea of ultimate responsibility is not a difficult one. Ministers must take ultimate responsibility for everything done in their departments and editors for all that is in their journals.

Another argument against “film credits” is that they “take up too much space.” But this is trivial. A stronger objection is that it will undermine systems of academic credit–such as citation indices. But undermining these would be no bad thing: credit should depend more on thought and less on number crunching.

At its last meeting the Vancouver group decided only to encourage debate on authorship. In May it will consider the three options outlined above. This editorial, the paper from Newcastle, and two letters are the BMJ's contribution to the debate. We want to hear from readers about this issue–to avoid editors proposing a solution that is unacceptable to readers and those who produce papers. We also encourage those who send us papers to experiment with the system of contributors and guarantors. We will be happy to publish these credits–in addition, for now, to traditional lists of authors. Depending on what you tell us, we may soon ask all who send us papers to try describe themselves as contributors and guarantors.