Why Write?

Tags: academia, research, musings

Published on
« Previous post: How to Write a Scientific Paper — Next post: Most AI Interactions are Like Junk Food »

As a reaction to the previous article on how to write a scientific paper, I got some reactions from friends and colleagues, the underlying message being that they would rather like to understand why writing can be important. There is indeed a tendency to see academic writing as a tedious task that one should try to get done as quickly as possible lest it causes unbearable pain,1 or, even worse, that writing in general is a talent that you have to be born with.

That is of course rubbish. Writing is a cultural technique and a skill, and like any skill, it becomes better by practising it. To accomplish that, I want to provide three different2 answers as to why writing can be useful. Pick yours and have fun!

(Notice that I will be focusing mostly on academic writing, but the with some minor changes, I think the same answers can also apply to other types of writing; more about that at the end of the article.)

Answer 1: Writing Sorts Your Thoughts

The first answer to ‘Why Write?’ is that writing is a wonderful way to sort out your thoughts in a concise manner. Virtually all researchers enjoy talking about their research, and I understand why: having an interlocutor that prompts you about specific details is a marvellous way to outline your own research!

Writing, by contrast, forces you to assume a much more concise perspective: most venues have a limit on the number of words or pages you are allowed to use, and I argue that these restrictions are fundamentally a good thing because they force you to be more creative with your explanations and miserly with the number of words that you use.

But even if you have no such restrictions, putting things into writing is a marvellous way to sort out your thoughts about the work and ‘clear your internal cache,’ as it were. I posit that while writing down your ideas, you will experience at least one of the following things:

  1. You will get ideas about how to improve your research.
  2. You will improve your understanding of the context of your research, i.e. the ‘bigger picture.’
  3. You will find it hard to express certain aspects of your method, forcing you to question your understanding of certain parts of the materials.

The last part is of particular relevance: I find it easy to trick myself into believing that I understand certain techniques or methods, but only by actually writing them up do I have a way to see if I internalised the knowledge well enough to actually generate something with it.

Answer 2: Writing Increases Your Audience

Most academic research will not be widely read, unfortunately. At the larger machine learning conferences, your paper, when accepted, will be one of thousands (!) of other papers. Unless you are extremely lucky, chances are that few people will interact with your research. Thus, my second answer to ‘Why Write?’ is that any investments into a well-written abstract and a well-written paper can pay off by enticing readers from other disciplines, which often leads to unexpected outcomes.

For example, my work on dimensionality reduction algorithms and their evaluation languished in my previous research community, but it was picked up by different communities—and I even received some offers for collaborations because people liked the way I tried to motivate certain aspect¬†(I am a strong believer in building the intuition for methods).

If you polish your writing and keep on doing it, you will increase the probability of other people interacting with your work. This is all the more true for ‘outreach’ writing, i.e. writing that targets a more general audience. My blog posts on manifolds was read by more people than most of my papers—and can actually inspire others to engage more with your research field. What’s not to love?

Answer 3: Writing Improves Your Research By Narratives

The last answer to ‘Why Write?’ can be seen as somewhat utilitarian, so bear with me. I believe that writing an academic paper should always involve an underlying narrative. Just like a play has its character and a story, so should any good research paper. By that I do not mean that you invent things or be imprecise in your speech, on the contrary! A paper’s underlying narrative needs to be based on the facts of your discoveries.

If you study outstanding academic papers, such as Shannon’s A Mathematical Theory of Communication, you will find that they all have a narrative writing style. Your writing will benefit if you start adding your own narratives into a paper: you start by explaining the problem—introducing your characters—and why it is important. Then you show how your method works—how your characters overcome obstacles—and what properties¬†(and limitations) it has. I used to react with immediate disdain for people that said that ‘The paper needs to tell a story,’ because I always understood this as an exercise in creative make-believe. I have since changed my mind and accepted the meaning behind the idea of a story: a story is compelling and makes it easier for others to understand your research. Even great ideas can flounder if no one recognises them as such.3

On top of that: new research is being published every day at ever-increasing rates. Try to captivate your readers if you want to stand out!


You now have some seen some answers to ‘Why Write?’ but you may be wondering how to actually get better at it. The answer is simple: read and write more. Read the best papers you can find, either from your own field, or from others,4 and study their writing style. Be bold in emulating whatever you like—it is a matter of personal taste anyway—and be prepared to rewrite your own papers multiple times. One day, you will be deep in the zone, making it look easy. Trust me on that.

(By the way, all improvements should be measured relative to where you started from. When I read my first papers and compare them to my recent ones, I see definite improvements. That is all that matters in the end.)

Final Thoughts

Writing is a marvellous cultural tool; writing well feels like a superpower because clear communicators will always be relevant, in particular as we are moving towards new hybrid teams, both inside and outside of academia.

Learn to use that superpower wisely and responsibly, until next time!

  1. This sentiment certainly explains some low-effort and low-quality papers out there! ↩︎

  2. The answers are not necessarily fully orthogonal, though. ↩︎

  3. I think the missing exposition and narrative might be one of the reasons why ideas are constantly re-discovered. ↩︎

  4. Personally, I like The Oxford Book of Modern Science Writing↩︎