Writing A Paper

Recently I submitted a paper to my first international conference. I thought that writing the paper would be easy, seeing as all of the ‘actual’ work was done already, but in the end it took 16 drafts over 6 weeks to get something that my supervisors were happy with.

That’s not because my supervisors were being unnecessarily harsh or perfectionist (the first draft was a bit rubbish and the current version could still be improved) but a paper takes much more thought than I realised. I’ve learnt there are at least four things a good paper must be:

  1. Logical
  2. Accurate
  3. Precise
  4. Entertaining

Now keep in mind that I’ve written one paper and am far from an expert on the matter. Unfortunately though writing good papers comes so naturally to most experienced academics that they struggle to convey what is that they are actually doing. These notes are an attempt to make plain the criteria that they follow for my future self and others.

Logical

Science is a process of examining a problem, working through it logically, comprehensively examining and following the available evidence to come to a reliable conclusion. A research paper is meant to be a demonstration that you have followed this process and that your conclusions are in some way justified.

The typical structure of a research paper falls out naturally from this. The root of the paper is your introduction which connects your research with the rest of the world. Every subsequent paragraph answers some question that was implicitly raised in the previous one:

Introduction

  • What problem are you trying to solve?
  • Why is it worth solving?
  • Have others attempted this?

Literature Review

  • What were their results/approaches?
  • What hasn’t been tested/tried?
  • How does your work fit in?

…and so on.

This isn’t unlike how you might structure any piece of writing but a scientific paper requires more rigour. A break in the logic or an unanswered question implies you might not have fully considered all of the evidence and damages your conclusions.

Accurate

There are only three things you can include in a scientific paper:

  1. Tautologies, things that are true by definition
  2. Things that were proven in some other work that you reference
  3. Things that you prove in your paper

Anything else is suspect, perhaps even dangerous since others may reference your assumptions as if they were facts. Accuracy issues can slip in easily. At several points when drafting the paper I found I had made a statement that was based on what I considered common knowledge in the field. More often than not I found these statements were true but with important details that were not obvious.

Precise

Suitably precise data and figures are obviously necessary since people are going to be making conclusions from them. Less obvious is the precision required in the wording.

When I write, I naturally use a lot of adverbs but they’re often difficult to quantify. How good is ‘much more effective’? How frequent is ‘quite often’? While imprecise statements might not be outright false they are open to interpretation.

Precise language leaves no doubt. Rather than ‘much more effective’ it is more useful to say that something is ‘2.5x more effective on average…’. Of course you then need to ensure there is no doubt what ‘better’ and ‘average’ mean in the context…

Entertaining

The very best papers I read have some extra magic that makes them really enjoyable to read. I’ve not managed to work out exactly what the secret ingredients are but here are the few I’ve picked up on:

Beautiful

Pretty papers are nicer to read. A healthy mix of figures, text and tables looks much less intimidating than a wall of text on a first glance and it keeps things interesting throughout.

Expressive

Expressive language is allowed so long as it remains precise. Emotive language when introducing the problem can give the paper a bit of go, pulling you into the paper and carrying you through the tricky bits. There is a balance to strike here though as most papers are written for an international audience. My rule at the moment is to use simple language for the hard parts of the paper and be a bit more playful elsewhere.

Short

No one likes their time being wasted. The best papers I’ve read can flip your perspective in a few good pages of text and some well placed figures. A short, insightful paper is a joy to read.


There is one thing that every academic I have met has been very clear about. The real trick to writing a good paper? Practice.