Vol. 17 - Orchestrating collaborative writing
Writing with many people is hard, but a strong process can make it work - read about the guidelines I use!
Writing is hard. Writing with many people is harder.
I’ll be honest - I don’t really believe in writing in a committee. I am convinced this is how we end up with so many camel papers (a camel is, of course, a horse designed by a committee). Writing should be about consistency: consistent narrative, consistent rhythm, consistent flow, consistent voice, consistent style. The more people we invite into this process, the blander the writing will be.
For this reason, my default recommendation for collaborative writing is: don’t. Especially when it comes to the manuscripts from my lab, I avoid writing a single sentence because it’s not my manuscript. I edit (as a grad student put it recently, “uh… that’s a lot of ink”), I comment, I provide feedback and high-level advice. But a manuscript has a lead author, and I am expecting this lead author to lead the writing.
But there are a few situations where collaborative writing cannot be avoided, and in these cases, it helps to have a process in place. Reviews are a good example: more likely than not, they will require a breadth of knowledge and viewpoints to have an impact.
One of our COVID-19 lockdown pet projects was to write a review on the ML techniques to predict species interactions. This was a big paper to write in a short amount of time, involving many junior colleagues. We needed a process, which can be summarized as: we turned writing a paper into a pyramid scheme.
Step one was outlining the paper. And by “outline”, I mean a really, really good outline. Down to the point where each paragraph was planned out, with a topic sentence, three bullet points outlining the whole argument, and three to five references. This involved the entire group, together.
If this outline sounds like 75% of the paper… it is! And this is the point - the scaffolding needs to be sufficiently robust that multiple people can tackle the writing independently.
From the outline, we assigned each paragraph to two co-authors, and one reader. This was step two. After the draft of each paragraph, the reader would check for clarity and content. When this was done for the entire paper, we had a first draft. This part includes fewer and fewer people as the text starts to coalesce: not all co-authors were assigned as readers.
Turning the first draft into a paper was the third step: each section was assigned to one editor, who made sure that the style of each paragraph was consistent. When the sections were smoothed out, the section editors got together, and edited the entire text, top to bottom (and also bottom to top - seriously, reading the text in reverse order is a game changer). This step also included fewer and fewer co-authors, down to the moment where I printed the entire draft, lined it across our living room, grabbed my markers, and spent a day doing line edits.
And it worked! It kept working on other papers with even larger authorship! This entire process of collaborative writing was inspired by my experience contributing to book chapters, and chatting with colleagues who had edited books. It works really well for reviews, and especially for narrative reviews where the group wants to advance more opinions. But more importantly, it makes sure that there is no ambiguity as to who should write what and where, and it respects the time of everyone involved by maximizing the productivity of the time spent on a manuscript.
And with all that said, stay tuned for Vol. 18 next week, for a discussion of why I dislike “joke” scientific papers, and why the best they can hope to be is mildly counterproductive.