Academic papers provide a durable snapshot of our thinking about a particular topic in a particular point in time. They provide a great outlet for a creative synthesis of the different strands of a research project, starting from the relationship to previous research, through the methods that are employed, centering on the empirical findings, and ending with a meditation on the state of the world (also known as a discussion section). Because of that, papers can give their authors a tremendous amount of satisfaction. They are also important common currency in academic careers. So, we write papers. The bad news is that writing full papers is one of the hardest things we do.

Where you start writing the paper is often one of the hardest things to decide. The good thing is that by the time you are writing the paper, you have probably already done a lot of work that could serve as a starting point.

A good place to start writing is deciding on the figures that would appear in the paper. They don’t necessarily have to be finalized or fully polished, but a sequence of figures that contain the main results that will appear in the paper often start to tell the story that the paper will eventually tell.

This then allows you to write down the main results in sequence and see whether the logic flows.

Another good place to start is by writing down the methods.

Share ideas with co-authors and collaborators early on in the process. A google doc, Overleaf doc, or a GitHub repository are all good places for people to read/comment/suggest.

Quarto offers a really interesting option for weaving together computations, figures, and text into rendered publication-ready documents.


Authorship and credit are a sensitive and touchy subject. Authorship should be discussed early and often, to avoid some of the pitfalls that arise from making assumptions about authorship. Ideally, project leadership/”ownership” will be clear enough, so that the identity of the first author(s) will be obvious, but particular caution needs to be taken in cases where different people are collaborating closely to generate results that require extensive methods and/or infrastructure development. In these cases, we can consider shared first authorship. But this needs to be clarified early on in the development of the project, so that different people can determine the level of their involvement if a single first authorship is determined to be the best course of action.

Regarding secondary and collaborating authorships (authorship positions other than the first and last): Overall, the general philosophy we take is to be inclusive in authorship. This means that folks who contributed to the work described in a manuscript while it was being developed can be co-authors if they are interested. This rewards and encourages people to be involved in other people’s projects and to help move that work along. It is primarily the responsibility of the first and last author to pay attention to the role that different people are playing.

Other resources#

Matteo Carandini wrote a short, succint and memorable essay that explains how to write a paper (with him, and more generally).

One of the main take-home messages of this essay is that a good place to start writing your results (maybe once you have those figures mentioned above) is by writing down a series of topic sentences for the paragraphs that will make up that part of the article (and then then other parts as well).

If you have more time, and want to Carandini’s recommendations to heart, you can also read “The science of scientific writing” by Gopen and Swan (1990).