The other day I posted some remarks on Stan Liebowitz’s analysis of coauthorship in economics. Liebowitz followed up with some more thoughts:
I [Liebowitz] am not arguing for an increase or decrease in coauthorship, per se. I would prefer an efficient amount of coauthorship, whatever that is, and certainly it will vary by paper and by field. If you feel you are more productive with many coauthors, that is not in contrast to anything in my paper. My point is that you will pick the correct number of coauthors if you and your coauthors are given 1/n credit (assuming you believe each author contributed equally). If, however, all of the coauthors are given full credit for the paper (and I have evidence that, in economics at least, authors are far more likely to receive full credit than 1/n credit), authors will get credit for more papers if they use more coauthors than would otherwise be best for total research productivity. My criticism is in the inefficiency induced by not using 1/n as the reward system. [I assume, I think reasonably, that increasing the number of real authors eventually causes too big a team and that productivity falls (too many cooks).]
A particular form of excessive coauthorship consists of gratuitously putting friends on a paper. This form of excessive coauthorship doesn’t actually hurt research productivity since gratuitous authors have nothing to do with writing the paper. Gratuitous coauthorship is a rational (although unethical) response by authors to a reward system that does not give 1/n credit. We shouldn’t be encouraging this type of behavior.
I’ve been involved in almost every authorship situation. I’ve added coauthors who’ve written just a couple paragraphs. I’ve done this not to do a favor for a friend, but just to make the paper better. For Bayesian Data Analysis, I keep adding new coauthors. I don’t remove the old authors even after they’ve stopped contributing. I once had a project I did entirely myself, but at the time I was feeling paranoid so I approached two colleagues—not friends, just two people I slightly knew—and invited them on to the project. They contributed a lot to the papers that got written, but originally I included them only because I was afraid that any paper with myself as sole author wouldn’t get taken seriously. I was once deeply involved in a research project—at one point I was dictating formulas to my coauthor over the phone—and then I was stunned to find that my coauthor did not want to include my name on the project as a coauthor, he just wanted to thank me in the acknowledgments. I wrote Red State Blue State with four coauthors, all former students. The publisher recommended that, even though all five of us were officially authors, and we were all on the title page, that my name alone should be on the cover, so it would sell more books, because nobody wants to read a book with five authors. I discussed it with my coauthors and we followed the publisher’s suggestion, but I regret it. I think all authors should be on the cover. And we didn’t sell a lot of books anyway. Once or twice I’ve done collaborative projects and then seen papers floating around with our work but without my name on it. Conversely, I’ve been coauthor on collaborative projects where I’ve done very little. Once I wrote a paper with an economist—it was a joint effort, he had the original idea but I wrote most of the paper. I thought I should be first author but he told me that in econ it’s always alphabetical. Once I insisted on a reverse-alphabetical authorship for a paper just to ensure that my end-of-alphabet coauthor would get the appropriate share of credit. I listed George Romero as a coauthor on the zombies paper, just because. I like adding coauthors; they almost always add something useful, and I don’t really worry about the distribution of credit.