Citation Style

date: 2018-05-07
status: notes

To collect my thoughts on citation styles in different contexts. I’ll list the places where I want to cite effectively (and quickly!).

makefile strategy

If I’m citing the same source (e.g. Simon Jones) repeatedly, I’d like just to [@sjones], in the pandoc-citeproc style. I’ll need a .bib bibliography included and also a thorough understanding of the compilation process from whatever plain text source to the desired .pdf or .html. I suspect pandoc-citeproc and CSL citation styles will become my tools of choice.1

Kieran Healy models one workflow here. And I’ve taken the Makefile (see also discussion here) from his pandoc-templates repo.

Joseph Reagle gives a stellar example for marking up a curriculum vitae (markdown and corresponding html). Reagle opened two relevant issues (which are interesting forums to read for anyone trying to incorporate .csl files with a pandoc-citeproc workflow):

  1. Including styles with inline-references (instead of bibliography) for publication lists
  2. apa-cv.csl not working with pandoc-citeproc

citation style languages

In undergrad, I was comfortable using Biber and $\rm\LaTeX$ for papers in the Chicago footnote style and in the elsarticle.cls document class. I never really moved beyond \footnote{} for references in my mathematics problem sets.

And yes, it seems Chicago (as footnotes or inline author-date) for informal writing and “numeric (or mnemonic), with titles, sorted alphabetically” for mathematical writing remain the best choices for my intended audience.

There’re also the IEEE and ACM proceedings styles, but these are further out of my scope. (Although I see the advantage to IEEE in handling mixed media and link rot.)

Now to give examples of different .csl styles in (approximate) use.

computer science

N. Polikarpova and J Yang et al., Enforcing Information Flow Policies with Type-Targeted Program Synthesis (2018).

Note that Polikarpova expanded the reference keys to be more legible, which I like quite a bit.

P. J. Guo, Non-Native English Speakers Learning Computer Programming: Barriers, Desires, and Design Opportunities (2018).


elsevier-with-tiles-alphabetical.csl (i.e., numeric)
T. Tao, What is good mathematics? (2007).
M. Mirzakhani and A. Wright, Full Rank Affine Invariant Submanifolds (2018).

I fear, in general, mnemonic bibliographies are formatted by hand. For example, from the front matter of “[1801.07530] A Guide for Computing Stable Homotopy Groups”, we find

% \MRhref is called by the amsart/book/proc definition of \MR.

J.~F. Adams, \emph{On the structure and applications of the {S}teenrod
  algebra}, Comment. Math. Helv. \textbf{32} (1958), 180--214. \MR{0096219}

\bysame, \emph{On the non-existence of elements of {H}opf invariant one}, Ann.
  of Math. (2) \textbf{72} (1960), 20--104. \MR{0141119}

% and so on

which, unfortunately, matches the colloquial instructions given in math departments and recapitulated here:

Some Definitive Style Guides


But how do I manage my citation collection? CiteULike subsumes MathSciNet for bibtex records, which is helpful for citing papers across traditional discipline boundaries. (See also “What are good sites to find citations in BibTex format?”. TeX - LaTeX Stack Exchange. Retrieved May 9, 2018.)

Don’t confound file extensions!


Isn’t the goal to have internalized key references such that I can just prattle off at the appropriate times?

Not necessarily to have memorized a series of author-dates but rather to be familiar with “the field” or “the canon” such that citations are more akin to introductions? And as with introductions to be mindful of who is getting to know who and the relationship that’s therein forged?

(And also, isn’t it curious that mathematicians are using \bibitem as opposed to curating a bibliography database? Maybe it’s just one of those inherited quirks, but it also evinces that the mathematical canon, for many professionals, develops much slower than the spit-fire developments in, say, programming language research. I suspect, also, to lower the signal to noise ratio, professional mathematicians might regard parts of the canon as immutable.)

(Aside: Rosoff pointed me to Spivak’s Differential Geometry at a point in my life where the text didn’t make much sense. Still doesn’t, but I now get the reference.)


  1. I imagine I’m going to run into trouble implementing these in gitit.