Saturday, December 02, 2017

Claim: The Mathematics Subject Classification (MSC) Scheme Needs a Revamping

It's getting to be that time of year: it's the season to look at the dossiers of job applicants.

Every year, when I go through these dossiers, I am reminded at how poorly I fit into the two-digit MSC (Mathematics Subject Classification) scheme, as I my work is split across many of them without any obvious single home. Nonlinear and complex systems (and networks) shows up in the depths of many of the two-digit numbers, without a single obvious parental home. It is very easy for people who work on these (and other) kinds of interdisciplinary topics to fall into the cracks, and the extant MSC scheme and the extreme reliance upon it exacerbates this problem. It needs a revamping.

Simply put, the MSC is very poorly suited for classifying applied mathematicians. And because people have to use an American Mathematical Society (AMS) cover sheet when applying to US math departments and they also have to list MSC numbers on that cover sheet, people who work in such areas are hurt very severely on the academic job market for postdocs and junior faculty positions.

In my early applications, I put the two-digit number 37 ("Dynamical systems and ergodic theory") as my main subject area, but my application then often ended up in the hands of people proving theorems in those areas, and as a dirty applied mathematician I was often sunk at that point. I think that listing 37 in this way was a very serious mistake on my part. Things of course have worked out well in the end — yay, survivorship bias! — and working in between fields is far more fun and rewarding than not doing it, but it can also create an extra hurdle when on the job market, especially because of the major role that the MSC ontology plays in how applicant dossiers are doled out to different people to evaluate (and thus in who is reading the dossier). Thankfully, if it does work out, there is also the potential of benefits from the proverbial first-mover scientific advantage (or, to generalize, an early-mover scientific advantage), but first one has to get to that point and there are extra hurdles in the way.

While now it's more of a cosmetic thing for my personal situation (though other interdisciplinary people have to deal with this serious issue when on the market), this was very frustrating when I was applying for postdocs and my initial faculty jobs, as my application could so easily get siphoned off into the wrong spots and not find the right ones, purely as a result of the MSC ontology, which I was required to use in applications to US math departments as part of the AMS cover sheet.

Update: To illustrate my subject areas, take a look at my MathSciNet listing, although it only indexes a small subset of my papers and it captures a vanishingly small (well, not literally, but take a look on Google Scholar, and you will see that it is a factor of more than 50) fraction of my citations. I don't have a home within the MSC ontology, and most of the two-digit identifiers in which my work is best represented are still in 2017 often not construed to be areas of applied mathematics. Many people are open-minded, but many others are not, and the issues with the current MSC contribute seriously to this problem.

Update: By the way, as of right now, Google Scholar registers 10146 citations to my work. Meanwhile, MathSciNet registers 184 citations. These numbers differ by a factor of approximately 55.14.

No comments: