Wednesday, October 24, 2012

How many authors is too many?

Physics Today recently published an excellent commentary Too Many authors, two few creators by Philip Wyatt. It is worth reading in full.

He compares the number of authors per paper in 8 major journals in 1965 and 2011. The number of single authors has plummeted while the number of papers with 4+ authors has dramatically increased.
In contrast, the number of authors of patent applications has remained steady.
Wyatt argues that this reflects a decrease in creativity. This is all caused by the pressure for everyone to have large numbers of publications and citations to survive in science.
He asserts that many authors have not made significant scientific contributions to a paper and so should only be in the acknowledgements section.
He sings the praises of single author papers, as I do.

Why has not the number of authors on patent applications increased?
If a listed inventor, or “innovator,” did not actually contribute to the invention, the issued patent will be void if such deception is ever discovered. The patents most easily challenged in court may well be those with extraordinary numbers of inventors.
Overall I agree with Wyatt's concerns.
Some people are getting credit they don't deserve.
Multiple authors make it very hard to evaluate the quality and quantity of individuals contributions.
Too much time is being wasted producing papers as opposed to doing creative and productive science of lasting value.

The commentary brought a number of interesting responses. I particularly agreed with his  response included the statement:
so much effort seems focused on writing papers just to secure funding or a job that many fundamental building blocks needed as tools to spur creativity in our young scientists are lost in the process.


  1. I see two possibilities:

    1) Scientists are now genuinely working in a more collaborative way.
    2) Author lists now include more marginal contributions.

    If 1, I would argue that this is a good thing. When I get stuck and go and talk to Max, or when I convince one of my colleagues that they need to consider something they haven't, I think I'm being a more effective scientist than if I stayed in my cubicle all day.

    The flipside is that we could be overvaluing complicated rather than simple ideas, but I don't think that author number really measures this.

    If 2, the question is, why? The obvious answer is that it boosts publication numbers. Surely this could be taken into account by keeping first, middle and last author figures separate when counting publication and citation numbers. There's value to each, and hiring committees, etc. could surely consider three numbers just as easily as one.

    The question is, do they effectively do this already?

    1. Thanks for the comment.

      I agree that 1. (more collaboration) is good. For example, a paper which reports measurements of both ARPES and SdH oscillations on a topological insulator and compares the results is a good thing. This will involve more authors. Hence, some component of the increase in numbers of authors is a good thing.
      However, the anecdotal evidence is that there is also a significant component due to CV padding.

      Hiring committees also have letters of reference which usually (or at least should) describe specific contributions and their value.
      However, referees of grant applications do not have access to such information. It is then difficult to evaluate the significance of middle author contributions. I tend to give them little weight. I realise that in some cases this may be unfair, but I am not sure what else to do.

  2. I quite like in Nature (or is it Science or both?) where it is spelled out the individual's contributions. Perhaps this should be enforced and taken seriously. I don't know what the effect would be.

    There are papers for which I'm second author and made very little contribution, and a couple where I made a substantial (almost as much as the first author) contribution. I feel like the latter cases become meaningless because of the former, and don't like this.

    Yet, those were the most enjoyable papers to work. I am working on a single author right now, for the sake of my career alone, not because I like working on my own.

    The whole system's stuffed.

  3. I think the Nature/Science stipulations are inadequate and superficial. Sometimes they say "All authors contribute significantly to this work."

    In his response Wyatt points out that there is no evidence that Nature/Science ever use this information to question inclusion of an author. If they had for the Schon/Batlogg scandal they could have saved some people (including themselves) a lot of embarrassment and wasted time.

  4. This comment has been removed by the author.

    1. I read and like this blog a lot, and as many people I guess, rarely find time to comment on some of the very interesting posts.

      I wanted to comment the last post concerning the authors list and forgot, so this is a perfect opportunity. As an experimentalist, I have been recently involved in large collaborations resulting in papers with 15+ authors.
      There can be various reasons for this generally speaking, and as far as I am concerned, I would highlight:
      - the use of samples from various dources (e.g. films and single cristals in our case), that are prepared each time by at least 2 persons (for, say, growth and general characterization)

      - experiments performs at different places. In the case of measurements performed on large scale facilities synchrotrons or neutron sources).
      In particular in the case of newly developed techniques, one has to realize that many people are involved. Even tough they do sometime not participate directly in the interpretation or analysis of the data, there work is indispensable for the realization of the measurement, and should be properly acknowledged. Co-authorship is one way to do it. Take an experiment that required measurements on 3 or 4 sources and you get your crazy author list. You might say, well why doing the measurements on 3 or 4 sources instead of 1 ? All together, writing a proposal to get beamtime, having it accepted and performing the experiment takes in the very best case at least 6 months, so when possible, parallel work is indispensable. Some people take the lead, others are confined to more technical tasks, while other simply think about the physics that comes out but all together this is really large collaborative work, and if not everybody shares the same responsibility in the final result (most important are often the 2-3 first and last 2), all of those who are listed contributed at some point, and the final result would not have been the same without this contribution...

      I agree however that this is certainly biased by the trend to rank people by their publication lists and H-index, and this is often a matter of politics and diplomacy.