When it comes to authorship, how prolific is too prolific?

One of the suggestions we get regularly here at Retraction Watch is something along the lines of “This researcher publishes too much. You should look into that.” But how much is too much?

The phenomenon was the subject of a 2015 paper. It’s also the subject of a new article in Nature by John Ioannidis, of Stanford, and researchers at SciTech Strategies. The new article is unlikely to answer the question of how much is too much. But it provides some fascinating figures on just how often some authors publish, and even more so how they respond when asked just how they manage to publish so much, in the process raising questions about whether measuring productivity and quality in science should involve a ruler for stacked papers.

Ioannidis and his colleagues probed the Scopus database for every author with more than 72 full papers published in a single year between 2000 and 2016 (they excluded Chinese and Korean authors, whose names can be difficult to differentiate in such databases). For the purposes of the analysis, publications were limited to articles, conference papers and reviews.

They found more than 9,000 such authors who, as they note, published at least one paper every five days, on average. Most of them — nearly 7,900 — were in physics, where, as Ioannidis and colleagues point out, “In high-energy and particle physics, projects are done by large international teams that can have upwards of 1,000 members.” Counting those authors therefore seemed unreasonable, so they were excluded.

The final tally? 265 — a figure which included at least one researcher with numerous retractions:

Materials scientist Akihisa Inoue, former president of Tohoku University in Japan and a member of multiple prestigious academies, holds the record. He met our definition of being hyperprolific for 12 calendar years between 2000 and 2016. Since 1976, his name appears on 2,566 full papers indexed in Scopus. He has also retracted seven papers found to be self-duplications of previously published work.

(Inoue now has eight retractions, according to our own data; his most recent retraction was in March, 2018.)

But in general, Ioannidis and colleagues stress:

We must be clear: we have no evidence that these authors are doing anything inappropriate.

On being reduced to a middle author

The Stanford group also surveyed some the most prolific authors, asking them about their publishing habits and their views on authorship. Their answers are illuminating.

One respondent, Rinaldo Bellomo, seemed to shrug:

My first comment is that some investigators seem impossibly short and some impossibly tall. Some investigators seem impossibly non-productive and others impossibly prolific. No mystery for either: it’s the normal distribution curve with people at each tail nd. To people in the middle, each tail end will look improbable. They are right. By definition, they are. Gauss would be proud.

Another, a member of the “prolific class of 2016” who is anonymous in the survey, wrote:

These articles do show up in Pubmed under my name, but I personally don’t count them as ‘my papers’ and don’t have them on my CV as such, as there is a distinction between being a ‘named author’ versus a ‘consortium member’ authorship.

Another in the 2016 class, also anonymous, wrote:

I have my name on about 70 publications per year, reflecting the cohorts I have helped establish and make freely available to many researchers and consortiums. About [one-third] of my publications are ones I am corresponding author on and are undertaken by my group, and led by me or by staff under my supervision. Another one-third are led by local colleagues and use the resources I helped establish. Some of these are led by people who I trained. The other third are where our data has been “used’ by external researchers and large consortiums (whose leaders think they own our data because they organised finding for measuring the latest SNPs), I am reduce[d] to a “middle author”, and that even gets held AGAINST me by some nasty (local) grant reviewers who are jealous of my success.

These and other comments led Ioannidis and colleagues to conclude:

Overall, hyperprolific authors might include some of the most energetic and excellent scientists. However, such modes of publishing might also reflect idiosyncratic field norms, to say the least. Loose definitions of authorship, and an unfortunate tendency to reduce assessments to counting papers, muddy how credit is assigned.

Turning the tables

Reading the paper, we found ourselves wondering how the team had come up with the number 72. Ioannidis tells us by email:

There is nothing magic about >72 full papers, clearly there is a continuum of being prolific and 71 is not really that different from 73. When we looked at the frequency curves (how many authors have 1, 2, 3, ….n papers), we did see a power law, and two types of excess deviations seemed to emerge, one major one due to physics (high energy and particle physics) and then, after removing physics, there was a smaller excess signal beyond about 70-75 papers. Given that >72 is equivalent to “at least one every 5 days”, that rounding seemed nice as well.  Most important, we had to choose a convenience sample that we could work with in depth and perform also surveys of authors. With >72 we have about 9,000 authors and after excluding Physics and Chinese/Korean names, we had 265 authors to work with in depth which would be manageable.

Ioannidis is, it should be noted, considered quite prolific in many quarters. And at the risk of being impolite, we pointed out that the threshold excluded him. Would that be a problem for critics of his work? True to self-deprecating form, said it was an “interesting question!”

My annual peak is at about 50 eligible papers in a single year (which is a bit of an overestimate because several of my viewpoints still count as full papers), which means there are almost 20,000 people with higher peaks than me, several thousand people even excluding physics. I wish I could lower the threshold enough to include me, but then the project would have been unmanageable.