On writing styles in scientific publishing

Today I came across something while working on a paper with a collaborator. Let’s call them Kevin (yes, I just watched Home Alone). You see, Kevin’s great. Excellent scientist, very knowledgeable in their area, and very easy to work with. But Kevin and I think very differently. Normally it’s just a fun quirk that I barely notice, but it becomes really obvious when we write a paper together.

As I’ve learned over the years, people have wildly different styles of writing. Some need the threat of a hard deadline, some prepare months in advance by picking up papers and lines to write along the way. Many older professors seem to start by copying and pasting bits from their other drafts or writing they’ve done recently, then rewriting to suit the current narrative. Some prefer to write exclusively in the passive voice while others push for more use of the active voice.

The difference between me and Kevin isn’t really any of these (though they definitely prefer the copy-paste-edit method, and I’m more of a write-from-scratch guy). It’s really more about the way we think. Science is inherently a creative process. While writing, I get these sparks of thought and sudden ideas, but I tend to look past them in order to achieve some structure.

How I write

For me, a paper needs to follow a logical flow from beginning to end. Ideally, I think each paragraph should behave like a function in your favourite programming language: it should do one thing, and only one thing. This singular purpose, I think, makes for easy reading. The result should be a paragraph that can be summarised in around 4-5 words, or at least less than 10. If you stacked all those paragraph summarise together sequentially, they create one continuous logical argument. I think this is analogous to the “red thread” in writing: the core theme in a piece of written work that can be traced through the writing, guiding the reader through (there’s a nice summary of writing tips here; it’s behind a paywall, but you can find it on Sci-Hub easily).

Structuring the flow between paragraphs

Sometimes I write paragraphs and then edit them to have such singular purpose and rearrange them for a logical flow. Often though, I begin a paper with a rough outline in bullet points of what I want to write. For instance, for a paper applying machine learning approaches to the genetics of psychiatric disorders (a little niche for most people, but the topic of my PhD), it might look something like:

Introduction

• Psychiatric disorders are phenotypically complex
• This is reflected in their genetic architecture
• Linear approaches have identified robust associations
• But have fallen short of addressing this complexity
• We now have a confluence of data, compute and algorithms to apply flexible approaches
• That’s what we do here

Or something like that. To me that’s not perfect, but for something I just came up with it’s a reasonably logical flow of statements. Each of these then becomes the basis for a paragraph.

Structure within the paragraph

That paragraph should set up the point and give some evidence to support it at the very least. During the actual writing and editing part things can get mixed up again, so pruning it back to be paragraphs of singular purpose is sometimes needed.

I know others take this idea even further and follow the “the three S rule” of:

• State - the opening point of the paragraph
• Support - evidence to back it up
• Summarise - conclude the paragraph

I think this is a great idea, but I don’t like to be that prescriptive when I write, and I often leave summaries for later (but perhaps I’m misunderstanding the last S). You’ll also notice I follow none of these rules when writing blog posts…

Working with others

Now I said I felt creative sparks while writing. I think Kevin must be feeling one endless firework display, because he will be pulled left right and everywhere else while writing a single paragraph. This means there’s not really a logical flow in them, and he will often wander off or introduce the odd sentence because he wants to put in the paper somewhere, so it may as well go here, even if it doesn’t necessarily make sense.

I know this sounds like one big criticism of Kevin, but actually I think it’s fascinating how different the process is between people that work together in the same field. It’s also worth noting that Kevin ultimately writes great papers, but the process itself is just very different from mine. I don’t think I’m in a place to say one way to write is better than another by any stretch. It’s an area I’ve noticed I need to work on over the last couple of years, and I’m actively trying to improve on it. The main issue is that reading about writing just reminds me I should be writing instead. This is probably why Write No Matter What has been sitting half-finished on my shelf for a year or more now, despite being a great book.