PhDs and publications

I ran across a post from Tim Chevalier of Heroku about grad school. I’m not arguing with most of his points; I agree on opportunity costs, working conditions, and privilege, and we’ve all seen how universities react to accusations of sexual assault. For job prospects, the typical route is to use internships with large companies that do work in the general vein of your research to secure a position before graduating — something he seems not to have understood.

But one thing he said seemed rather off:

In grad school, I never wanted to spend all my waking hours on research, which meant that if I’d graduated, I would have had at most 2 or 3 publications; when I read CVs for tenure-track faculty candidates who were coming to meet with grad students, they had as many as 20 publications straight out of grad school.

Three publications in six years. Three. PhD coursework takes about two years to finish. Assuming he didn’t want to do any research while taking classes, that still leaves four years of grad school in which he would only manage to publish one paper every fifteen months.

If he were working a typical work week, he should be able to produce a publication at least twice a year while not taking classes. That would be eight publications, not three. Apparently, insofar as research is concerned, Chevalier is a solid underachiever. Twenty publications in five to six years is about one every three months — a fair bit of work, and even a bit grueling if you add on classes, but not “a monomaniacal focus on work”. Certainly it would be hard to get that number of publications on algorithms and data structures, but with a focus on engineering instead, that is achievable.

In short, Chevalier left grad school, felt bitter about it — with good reason — and is now determined to find every aspect of it problematic, even though part of the problem was him.

Leave a Reply