Paul Alper points to this news article on a clash between two scientists at the same institution fighting for credit on a research project and writes:
That dispute retriggered a question I long have had. In the case of multiple authors, how to know who the main author is. That is, what is the current protocol in various disciplines: First named? Last named? Alphabetic? Prestige? Variations within a discipline? Variations between discipline authors? Changes over the years? Is there confusion when an author has a “de” or a “van” or a “von”? Or even “aa” which is last in the Norwegian alphabet.
An interesting example is the Metropolis algorithm which takes its name from the alphabetically first author listed and I guess his was not the largest contribution.
Pontryagin’s maximum principle is famous in control theory but have you ever heard of L. S. Pontryagin’s coauthors, V. G. Boltyanskii, R. V. Gamkrelidze, and E. F. Mishchenko?
To take these points in reverse order:
3. That last question, coupled with the topic, reminds me of the famous song (“I have a friend in Minsk who has a friend in Pinsk . . .”).
2. We discussed the invention and implementation of the Metropolis algorithm in past posts here and here.
1. As an author who’s had hundreds of different collaborators, I’ve thought a lot about the rules that are used for authorship. Here are a few rules I’ve seen:
– In economics, authors are listed in alphabetical order. At least that’s what Aaron Edlin told me when we wrote a paper together.
– I’ve also seen in economics a paper with many contributors listed under a single author, with all the other contributors listed only in the acknowledgments.
– I had four collaborators on the Red State Blue State book. I wrote most of the words and did most of the work, but their collaboration was absolutely necessary, so all of us were authors, with me first. But the cover just had my name—I had this idea that the book would be a bestseller and the publisher said that a book with five authors will never be a bestseller. It wasn’t a bestseller anyway, and then I felt bad, because many people took it to be a single-authored book because they just saw the cover.
– The authorship of Bayesian Data Analysis is pretty much in decreasing order of the authors’ contributions, but the last author, Rubin, is a special case. He didn’t put much effort into the writing of the book, but large parts of it are derived from his earlier writings.
– When I collaborate, I generally follow the rule that the person who does most of the writing should be first author. I like to encourage student coauthors to take the lead on a project so that they can be first author. Sometimes the lead shifts. Expectation Propagation as a Way of Life was a paper that was originally my idea, and I wrote much of it, along with a large number of collaborators, but at some point it got stuck in the review process and Aki took over, adding more content along with another collaborator, so it made sense for Aki to become the first author.
– Two of my most influential papers are Inference from iterative simulation using multiple sequences by Gelman and Rubin, from 1992, and The no-U-turn sampler: Adaptively setting path lengths in Hamiltonian Monte Carlo by Hoffman and Gelman, from 2014. The idea and method in the 1992 paper was entirely mine, but Rubin was helpful in contextualizing it. It was fair for him to be a coauthor—I invited him to do so!—but it’s funny when people talk about the Gelman-Rubin statistic, because I came up with it on my own. The idea and method in the 2014 paper was entirely Matt’s, but I was helpful in inspiring him to come up with it. It was fair of me to be a coauthor—at least, Matt didn’t seem to object!—but whenever the NUTS algorithm comes up, I make it clear that it was Matt’s idea alone. I guess these things sometimes balance out.
– Sometimes one author will take a collaborative work and write it up and publish it under his own name. That gets me really mad.
– I didn’t know that “aa” is last in the Norwegian alphabet. But we do have this story about a ticket to Baaath.
That reminds me of this:
https://carlstrom.com/publications/scsh-manual.pdf
There are two authors, but one of them, Olin Shivers, has an extremely bitter acknowledgement section where he claims he did it all himself (for his whole career). Is the other one slacking? Or is Olin unjustly taking credit? Unclear.
I clicked through and took a look, and . . . are you sure that this acknowledgement section isn’t just a joke?
It could be. I saw it in isolation (a picture of it is traveling around the internet) and didn’t really look car the context carefully. Strange place to put a joke, I guess.
” – I didn’t know that “aa” is last in the Norwegian alphabet. But we do have this story about a ticket to Baaath.”
I don’t think anyone knew it, because it isn’t true. In Norwegian and Danish, the final three are Æ, Ø, and Å. In Swedish, the final three are Å, Ä, and Ö, because we just couldn’t agree.
Aa is used for Å in some Norwegian and Danish names, and when you don’t have a Nordic keyboard, but it is not an actual letter.
In my day, the 1960s, when I lived in Trondheim, there were three topics of conversation: the weather, the lack of discipline of the Norwegian national soccer team and the pronunciation/spelling of certain words. Definitely, back then, “AA” was to be seen as one letter as was the upstart equivalent, Å, and it was sort of at random–depending on age and home village of the speaker–which one was used in written Norwegian. The language books for English speakers pointed out to be aware of the two versions, especially when it came to “AA” which was to be found, not at the beginning of the alphabet but at (just about) the end.
To give a taste of the ever ongoing language wars in Norway, back then, the daily weather report depended on the presenter because one insisted on always using “sne” for snow, instead of switching to snø every other day. My local friends referred to him as “The Abominable Snowman.
On this day, in 2025, as I live in Trondheim, people still discuss the weather, are pleased with the current discipline of the national soccer team, and (still?) seem quite proud of their local dialect and spelling. Often to the point of mimicking their particular dialect in text messages and on social media.
As for the topic at hand, surnames beginning with an “Aa” are sorted together with “Å”, since they represent the same letter. The letter “Å” officially supplanted the “Aa” spelling already back in 2017. However, I could imagine that government decree taking some time to permiate Trondheim.
As for “sne”, that still holds for some names, and in dialect, although the correct form is “snø”.
Why, there we are! I wasn’t even a topic of conversation myself back in the 1960s and would have guessed that “Aa” was superseded much earlier than 2017.
I can confirm that the lack of discipline of the national football team is now one of three *Swedish* topics of conversation.
Sorry, typo. I meant to say 1917!
Edlin told you the truth.
When you see economics papers where the authors are not in alpha order, it says something. Sometimes the senior people promoted a junior who did most of the work to give that person more credit. Sometimes a jerk senior person insisted on being first because they were a jerk.
Becoming a first author must have, at the time, been about 38 years in the making
Regardless of the possible sub-optimal title, the paper may have caused some shaking
I’m talking about that 2015 paper about reproducibility, or was it really replication
And if you look at the dozens of authors on there, it’s my name first on that publication
Now, I wasn’t directly involved in an experiment being reproduced, or replicated
If I remember correctly, I mostly just coded, verified, and intensely concentrated
Because when the replication crisis coincides with a personal crisis,
it simply might result in trying to make the most of being and/or feeling directionless
So, while being there where I was, I just tried to make the most of it all
I thought perhaps after a fall, one might first begin to crawl, and start again small
And maybe it felt like some sort of responsibility to try and help improve matters
And maybe an authored paper could be useful for my CV, or regarding some other letters
It’s now about 10 years later, and I don’t think it helped me concerning the academy
But it may have helped out with the crisis, or even crises
It did help make clear to me that A can lead to B, C, D, and up to, and including, Z
And that there might sometimes be ways to achieve something, or contribute,
without first being able to fully plan and see
At least in some cases, for instance when your first, middle, and last name starts with an A
and the authors on a paper are listed alphabetically
In my wife’s field (virology), the senior author gets listed last. I usually look for the author for correspondence as the best clue for who should get most of the credit.
So far, there seems to be little uniformity regarding authorship sequence when credit is due. What about the sequence when things go off the rails because of errors, accusations of—well, most anything?
Before I forget, recall the famous fictitious female coworkers of Cyril Burt–Miss Conway and Miss Howard. Quite a saga that.
In most of physics and most of the parts of biology I intersect with, the usual convention is that the first author is the student or postdoc who (probably) did most of the work, and the last author is usually the professor / group leader whose role in practice may be anything from fundraising to project management to data analysis to writing. The last author is almost always the corresponding author.
An interesting development is that it’s increasingly common, at least in biology, to have co-first authors. It’s understandable given that it’s often the case that two people contribute a lot to a complicated project, and everyone knows that the middle authors don’t get much credit. This sometimes leads to debate about who should be the first first author, and whether that matters.
I’ve also seen multiple corresponding authors.
Raghu:
There’s some research in economics about the total credit given to all the authors of the paper. The idea is that if a single-authored paper gives the author a credit of 1, then a two-authored paper would give the first author a credit of a1 and the second author a credit of a2, where 1 > a1 > a2 > 0, and a three-authored paper would give credits of b1, b2, and b3, where 1 > b1 > b2 > b3 > 0, etc., but that 1 < a1 + a2 < b1 + b2 + b3 < c1 + c2 + c3 + c4 < etc. I can't remember the details but they did some empirical analysis to estimate these numbers, and the punch line is that the authors in total benefit by increasing the number of authors. For example, if I write a paper and you write a paper, there's an incentive for me to add you as coauthor on my paper and for you to add me as coauthor on yours. Whoever wrote this was concerned about the incentives here. It did not bother me so much because I like working with lots of coauthors. Hmmm, let me do some searching . . . ok, the economist in question who did this work is Stan Liebowitz. See these two posts from 2013:
Is there too much coauthorship in economics (and science more generally)? Or too little?
Econ coauthorship update
Given what’s happening now in computer science and related fields (millions of students publishing multiple conference papers each year, with some presumably increasing number of these papers being entirely unoriginal but presented as original work, thus overloading the system in so many ways), these concerns about coauthorship just seem quaint at this point.
It’s mildly interesting that for things like academic hiring, being a middle author doesn’t count for much, but for citation metrics (such as h-indexes), which every sensible person thinks are silly but which are religiously followed by some, all author positions count equally and it’s certainly the case that “authors in total benefit by increasing the number of authors”.
I also notice that (thanks to backspacing the autofill of my full name) I misspelled my own name in the comment, and so I will receive zero credit for it.
Yes, I’ve often thought this. It would be instructive to look at something like the Clarivate Highly Cited Researcher list and see how many are on it due to huge numbers of middle author papers. Seems sometimes like once you get established you can look great on paper just by networking and writing opinion pieces. The rich get richer etc.
This equation makes me think of supposed gender effects on credit. One is the concern that women get less authorship credit when co-authoring with men relative to solo authoring or authoring with other women. Tthere’s at least one econ targeted study that claims this: https://onlinelibrary.wiley.com/doi/epdf/10.1111/ecin.13047.
The other is that women are less likely to get listed as authors: https://www.nature.com/articles/s41586-022-04966-w
But I haven’t read either closely.
If I am not mistaken, Fong and Wilhite (2017) published a paper in PlosOne titled “Authorship and citation manipulation in academic research”. Some quotes that might be relevant in light of the post and discussion here:
“Consistent with our hypothesis, we found support that females were more likely to add honorary authors as the estimated coefficient on males was negative and statistically significant. The odds that a male feels obligated to add an author to a manuscript is 38% lower than for females. As hypothesized, authors who already have several co-authors on a manuscript seem more willing to add another; consistent with our hypotheses that the decrement in individual credit diminishes as the number of authors rises. Overall, these results align with our fundamental thesis that authors are purposively deciding to deceive, adding authors when the benefits are higher and the costs lower.” (p. 11)
“Focusing on manuscripts, there is more variation in the stated reasons for honorary authorship. Lab directors are added to papers in disciplines that are heavy lab users and junior faculty members are more likely to add individuals in positions of authority or mentors.” (p. 29)
“It is difficult to examine these data without concluding that there is a significant level of deception in authorship and citation in academic research and while it would be naïve to suppose that academics are above such scheming to enhance their position, the results suggest otherwise.” (p. 30)
For me these findings are consistent with my view on academia and publishing: it’s rigged, and sad, and stupid to a certain degree.
It also confirmed why I keep hammering on new possible publication ideas, which I have mentioned on here a few times now (the “earn authorship” idea I am calling it).
And, the paper contributed to my view that certain sub-optimal or bad or questionable research practices might be done consciously and purposefully, and might be done more by people with certain personality characteristics. In this light, possibly see my manuscript titled “Why psychopathy might be present and prosperous in present-day psychological science” which can be found on SSRN and which cited the Fong and Wilhite (2017) paper if I remember correctly.
In pure mathematics the custom is to list authors alphabetically, and it has been so for a long time. In the rare instances where this is not the case, it usually has been done intentionally to reflect that what is reported is somehow really due to one author and that the other others have acted in support (and the first author has insisted on including him). The paper of Pontryagin is probably of this nature (I don’t know in this specific case, but Pontryagin was a mathematician culturally) – by the way Gamkrelidze is certainly very well known – Pontryagin was blind since childhood and depended on others to write for him, so perhaps that is relevant.
Other famous examples of non-alphabetic ordering of math authors include Beurling-Ahlfors (quasiconformal maps) and Sato-Kimura (classification of prehomogeneous vector spaces) and Selberg-Chowla (Selberg’s unique paper with a coauthor!) (there are more examples, but probably only 10-20 well known ones in the past 100 years). In the case of the Sato-Kimura paper, Sato developed extraordinary mathematics but was not much into writing it up, so his students, such as Kimura, wrote it up for him but put him as first author out of respect and to signal intellectual origin of the ideas. But in mathematics this is very unusual.
The idea that underlies the custom is that it is impossible to coauthor a math paper without contributing equally. This is probably true for papers with real depth and surely false for many papers with multiple coauthors.
Such customs also extend to ordering of references. Mathematicians order references alphabetically while physicists and genuine applied mathematicians order them by order of citation within the article.
And of course for naming whimsy, there is always the Alpher-Bethe-Gamow paper:
https://en.wikipedia.org/wiki/Alpher%E2%80%93Bethe%E2%80%93Gamow_paper
A Few Goodmen: Surname-Sharing Economist Coauthors
Allen C. Goodman & Joshua Goodman & Lucas Goodman & Sarena Goodman
https://scholar.harvard.edu/files/joshuagoodman/files/goodmans.pdf
“We also note that, for papers with more than two coauthors, surname-sharing eliminates the “et al” penalty documented by Simcoe and Waguespack (2011). Though the many expected citations to this paper will refer to it as Goodman et al. (2015), such citations will provide equal amounts of publicity to all of us coauthors.”
Authorship is academic currency. Once one is tenured and respected, my goal is to pass as much as possible of that currency to the hard workers, either students or junior faculty, and among faculty, the more junior the better. What does a senior person get by being first, or among the first, authors? If I’m any good, most folks in my field are at least aware of me, if not a friend (or competitor/frenemy). Speaking as a old fart known in my field.
In molecular biology (and I think natural sciences in general), the first author does the majority of the work/writing. It’s a usually a PhD student or postdoc, sometimes it’s shared between two people when they both made major contributions. The last author is the senior author, probably whoever runs the lab, gets the grants, etc. Then the other names are roughly ordered according to contribution but there may be more last authors if it’s a collaboration between different labs.