“Et al” is unethical

So apparently the AMS has a document on Ethical Guidelines. It’s actually remarkably well done. It has lots of tips that can help young mathematicians learn how to behave professionally. I was also impressed by the way that the guidelines avoid making too controversial of stands (which would go beyond the basics of ethics) while still not being milquetoast. For example, “No one should be exploited by the offer of a temporary position at an unreasonably low salary and/or an unreasonably heavy work load” is certainly an ethical obligation, but one that may be difficult to live up to.

I also thought that the guidelines about correct attribution were well phrased. For example:

To give appropriate credit, even to unpublished materials and announced results (because the knowledge that something is true or false is valuable, however it is obtained);

I have my own suggestion for a guideline on ethical use of citations: you should never ever use “et. al.” citations. Furthermore, if journal typesetters add them you should ask them to replace them with full citations.

If a bibliography just says “et al.” many readers are never going to get around to looking at the other names thereby effectively failing to properly attribute everyone. People at the end of the alphabet are already at enough of a professional disadvantage (see What’s in a Surname? The Effects of Surname Initials on Academic Success by Liran Einav and Leeat Yariv,), the use of “et al” just exacerbates this.

Hat tip: I learned about this document in a MathOverflow comment by Bill Johnson

44 thoughts on ““Et al” is unethical

  1. I agree with this wholeheartedly. (Well, maybe not that the use of et al. is “unethical” per se — that’s a strong claim — but certainly that it is disrespectful and should be avoided.) This advice was given to me a few years ago by one of my most stalwart guides to increased professionalism: Dino Lorenzini.

    @Felipe: with respect, I don’t think you should do this, although I doubt that either Joe or Barry would take any offense.

  2. What drives me crazy is that Mathscinet uses “et al.” This strikes me as insanity for a database of papers, authors, et cetera.

    For example, MR1184114 (93h:57005)

  3. Et al. also has an author page on MathSciNet, with 562 papers since 1981. It appears that none are single author, and yet the Erdos number of et al. is infinity. That’s got to be a record.

  4. Probably this is some problem with how mathscinet automatically pulls out metadata from papers, not actually a policy decision on their part. I agree that it would be a good idea to write them and point out that there’s a problem with their metadata and use of et al. My guess is that they’d be happy to fix it and just haven’t realized that this is happening? If someone else emails them please comment here saying so, and otherwise I’ll try to get around to it.

    1 is hilarious and probably not problematic because people presumably get the joke and know who the other authors are.

  5. I’m not saying it was malicious on their part, but it does seem slightly ridiculous from people whose job it is to get bibliographic data. If these papers aren’t showing up in people’s MathSciNet searches, that is a scandal.

  6. Noah’s story is true, except Zworski approached me at tea and didn’t knock on my door.

    According to the Mathematics Genealogy Project, I am last alphabetically in the history of mathematics! If you must use et al., may I suggest the variant “al. et”.

  7. The example of Zygmund (currently the last author on my bookshelf) shows that one can still succeed against overwhelming odds…

    Concerning the use of “et al.” in Math Reviews, the example given in #4 seems to be an anomaly among mainstream research papers: almost all the other papers listed seem to be either (i) articles written in honor of mathematicians with many remembrancers, or (ii) closer to biology or physics or other fields where the tradition is quite different than the alphabetical one in mathematics, and there are often many more authors.

    (E.g., in “Nature” nowadays, authors seem to be listed in roughly decreasing order of importance of contribution, and there is an “Author contribution” section describing explicitly the contribution of each; for instance

    “Author Contributions S.A. performed project planning, experimental work, data interpretation and preparation of the manuscript. Y.-H.L., I.-H.P., J.H., E.M.M., J.D.M., R.M.R., M.O., H.H. and S.L. performed experimental work. H.-H.N., F.D.G., D.L.K., A.J.K., L.L. and G.Q.D. participated in project planning, data interpretation and preparation of the manuscript.”

    which I copy-pasted from the first paper I saw on the Nature web page right now).

  8. One the one hand I know for a fact that the five authors of the famous “ATLAS of finite groups” abhore the practice of referencing the book as “Conway et al.”

    On the other, do people seriously believe that the many papers being churned out by the LHC with literally *thousands* of authors e.g.


    to pick a random example from the arxiv, should be referenced as fulley as possible???

  9. @14 Different fields necessarily have different attribution conventions. As explained in 13 “author contributions” are not given alphabetically, but instead certain authors contribute more importantly and get listed as such. In that setting “et al” makes a lot of sense.

    In fields where the number of authors is often small and where the author names are listed alphabetically with no differentiation (like math and economics) using “et al” is terrible.

    I did not mean my post to apply more generally than fields with authorship patterns similar to those of mathematics.

  10. Emmanuel-

    It’s true that it’s an isolated example, but to me that almost makes it worse. It would take ten minutes for someone to fix that, but no one has in the years that review has been on MathSciNet.


    There’s a reason that people come up with terms like “The ATLAS Collaboration” which at least isn’t explicitly favoring one author over another (I wouldn’t call it Aad, et. al.).

    On the hand, I still remain completely mystified as to how having a publication with 2500 coauthors could further someone’s career. I guess there must be some way of sorting these things out, but it boggles my mind to try to figure out what it is.

  11. Is it always the case that there is an alphabetical sorting of author names? I never heard of such an explicit convention in mathematics(though it seems to be a rule in computer science, etc.).

    A counterexample which immediately comes to my mind is the Vlèduts-Drinfel’d paper, MR0695100, improving Ihara’s results in the direction of Weil’s bound on number of points on a curve over a finite field. It seems it is mainly the first author’s work.

  12. #17: Sure, that bad example should certainly be corrected, and it’s likely that the MR people would do it if pointed out to them (there is an email address to submit corrections to MR, and in my experience, it is reactive, at least if MR numbers are wrong or things like that). Did someone already email them about this?

    To come back to ethical guidelines, one thing I (and many others) were very happy when my institution (ETH) recently issued its own guidelines was the explicit mention that it is unethical to purposely spread out results over multiple papers when they could be presented in a single contribution.
    (Article 13 in http://www.rechtssammlung.ethz.ch/pdf/414_Integrit%C3%A4t_Forschung_engl.pdf )

  13. As far as I know math and CS have the same convention: Alphabetical except in unusual situations where one author contributed significantly less to the project, but enough that they deserve some credit. In CS you have RSA.

  14. Noah: theoretical CS uses the mathematical convention (alphabetical order) but more applied CS uses the scientific convention (order somehow related to who secured the funding, seniority, etc.). This paper of Andrew Appel attempts to use this fact to determine which CS conferences are theoretical and which ones are practical.

  15. To follow up some of these comments, I just wrote
    a note to MathSciNet support mail about their “et al.”
    author page. Only a few math papers have had many
    collaborators, such as those published in recent years
    by faculty and grad students at U. Georgia working as
    “University of Georgia VIGRE Algebra Group”. Here all
    authors are listed on MathSciNet, but I’m not sure what
    most people would do if they wanted to cite such papers.

    Concerning the order of author names, alphabetical order
    is still fairly standard for mathematics (modulo issues
    with transliteration from other languages). But two
    papers in JAMS 10 list the authors as Andrei Suslin, Eric
    Friedlander, Chris Bendel [former student of Eric]. This
    seems to have been a matter of ranking the input of the
    authors, but has certainly raised eyebrows. People I’ve
    encountered like Asger Aaboe (science historian)
    and Gregg Zuckerman (mathematician) no doubt had
    their own views about the whole question.

  16. #18 – The article you cite was originally published in Russian. In the Russian alphabet, the letter V (or rather “в”) comes before the letter D (“д”).

  17. @#23. Sigh …

    Ok, here are the next bunch, non-Russian.

    MR0989230, Zagier and Kramarz.

    MR1492594, Hirzebruch, Demailly and Lannes.

    MR1189136, Hirzebruch, Berger and Jung.

  18. GS-

    I have no idea what point you’re trying to make. Yes, not every single paper ever published in mathematics has the authors in alphabetical order, but the vast, vast majority do, and it’s understood as a standard convention.

  19. @27. I had no idea that there is such a standard convention, and I spent time digging up counterexamples. However I will trust everyone’s word here that it is indeed a convention.

  20. I’ve heard it said (though I’m not sure how much I believe it) that deviating from the standard alphatbetical convention can “mess-up” future attempts to reference a paper and in doing so have bad implications for things like citation statistics.

  21. Further to Ben’s comment: they are generally very well-known *outside* their field, I would have hoped. (I’m sure it can’t just be me, right?).

    Mind you, I have a sentimental hankering after a possible return to mathematics of John Rainwater

  22. I can confirm that applie CS tries to list papers in order of “direct-work” contribution (creative/insight contribution is always arguable). However, there’s also the obvious point that with respect to material FOR HUMAN CONSUMPTION you ought to only use “et al” in cases where it doesn’t matter because the reader’s eyes are going to skip over the author list because it’s sufficiently long so they won’t take in all the names anyway. However, if names for citations databases are being harvested driectlly from papers, rather than the article name being used to find the original paper, clearly et al should never be used.

    (In the past in applied CS I’ve used et al after more than five authors, although I ought to rethink that,)

  23. Just the other day I was told about a study which found a strong alphabetical bias in tenure decisions, etc., in economics, which as pointed out earlier has a tradition of alphabetical author lists just as mathematics does. That is, the closer you are to the beginning of the alphabet, the more likely you are to get tenure, grants, awards, etc. People near the end of the alphabet were also found to collaborate less.

    (I have no idea whether the investigators considered the possibility that people late in the alphabet have less career success as a result of reluctance to collaborate. It’s worth considering, but my personal guess is that’s at most a very weak effect.)

    Furthermore, no similar bias was found in some other fields in which authors are listed according to contribution.

    This suggests that use of “et al” may indeed have serious consequences.

  24. One case I remember from many years ago. Some mathematicians talking among themselves came up with a proof of an interesting result. The resulting paper was 2 pages long or so, and it was felt that a 2 page paper with 5 authors would be ridiculous. So they published under a pseudonym.

  25. @36: possibly Peck, G. W. (1979), “Maximum antichains of rectangular arrays”, Journal of Combinatorial Theory, Series A 27 (3): 397–400, doi:10.1016/0097-3165(79)90035-9, MR0555816 ?

    It’s a pseudonym for six authors and a short paper but I don’t know whether it fits the rest of your story.

  26. Many people use et al. with full citation. I don’t see anything wrong with that at all. “ In Whatever et al. [63] … ” seems perfectly fine.

    If you don’t cite properly that seems wrong but who does that? When does it happen? In my field the bibliographies are dictated by latex style files.

  27. @15&18: perhaps a good way is to come up with Team Names for your collaborations! Colliander-Keel-Staffilani-Takaoka-Tao refers to their joint papers as by the “I-team”, and the name does seem to stick.

  28. Yeah, the the problem is that a team name makes you feel ridiculous. I’ve toyed with possiblility of christening my collaboration with Braden, Licata and Proudfoot as “team blowup” but I can’t really say that with a straight face.

Comments are closed.