How to write a paper

Felix, one of my early career doctoral students, recently spoke out the undeniable truth:

Reading a paper is much easier than writing one.

He implicitly (and later explicitly) asked for advise on how to write a research article to be published in a scientific journal. Felix met the most important criterion already: he had something to say. So, shouldn’t writing be easy then? Here is a random collection of thoughts I can offer on this from my own experience.

  • Above all, content matters. You need an idea, news, results, progress, a message. That’s a fact. There must be archival value. However, the final results are needed only to finish the paper, you can (and should) start writing much earlier since
  • Writing spawns new ideas. One often thinks of results to be “easily transferable to another area” and recognizes the difficulties only when trying to exactly write them down. Writing thus produces new questions, and new questions produce new answers.
  • Read, read, read. This is what Martin Skutella once said to me. Reading papers does not only inspire your research. You will also develop a better feeling of how to present it. Be it writing style, clarity, examples, notation, length, language, level of detail, rigor, organization, visualization, you will see good and bad examples. Needless to say that you should follow the good ones.
  • Know when to stop. I was taught this by Sándor Fekete, but I am not very good at it. Our expectations may surmount ourselves, and we may wait for that single great idea. It may never come. Thus, realistically assess what can be achieved in a reasonable amount of time (unless you are really going for something big, then never give up!) and know when you are only seeking incremental improvement.
  • Will you regret it? This somewhat contradicts the previous item (I did not claim my thoughts would follow any logic). It mainly applies when your paper is part of a longer term project. Don’t write the paper you will wish to have written again (this time “right”) in a year from now. If you feel this applies then invest in research first. Don’t waste your reader’s time.
  • Marketing. Papers may remain unread when not properly sold. Motivate well why  your paper is interesting, who will benefit, why and how? Tell a story. But of course
  • Be honest. This should go without saying: you must adhere to the codex of science by all means. Give credit to others’ work, be precise and complete, and stick to supported results, don’t even leave room for interpretation.
  • Seek feedback. Be grateful to everyone reading what you wrote and even more when they comment on it. It is good to know what works well, but it is more important to hear about what doesn’t. Don’t take critique personal. Writing is a process, one has to (and can!) learn it. It helps to know that we are all learners.
    I usually use many more colors than just red when giving feedback. I often ask questions and try to “misunderstand as much as I can” — only to point to missing arguments, lack of rigor, ambigous formulations. It is my hope that authors start asking such questions themselves when writing. A good question to ask after every single sentence is “why”? In our area it works well to send abstracts (few lines) or extended abstracts (several pages) to (refereed) conferences first. You will get feedback on the abstract. Even if unrefereed, the presentation at the conference tells you a little about your writing, and even more on your way of preparing and presenting your material. You can literally see whether your current setup “tells a story.”
  • Think of your reader. This is what giant Vašek Chvátal says at the beginning of his great book on linear programming. What is your audience? What prerequisites can you expect? Decide for yourself and stick to it. Don’t forget that you try to explain, to convey a message, in the most didactic way you can think of. Imagine a specific reader, and always ask: what is the most natural piece of information that would now suit the flow of material best? This is incredibly difficult, you need talent and experience for this, empathy certainly helps as well. But investing here turns a good paper into a great one.
  • Don’t mix up. Writing is a bit like coding. The organisation of the material is sometimes the most critical issue. When writing about models, don’t write about algorithms. When presenting algorithms, don’t write about the implementation. Also implementation is different from computations etc. Keep everything that belongs together in one place. It helps the reader a lot. You will still need cross-references but these are on good purpose.
  • Writing is not linear. Again: writing is a process. You just don’t start with the abstract and finish with the conclusions. This is not the way you carry out the research and this is not the way it is written. Write here and there, delete, re-write, let it grow simultaneously, in iterations. This certainly contradicts advise to be focussed, but only reading is linear, writing isn’t. Your writing is best when you have the big picture present in your head.

Finally, but certainly not least:

  • Start now. You think you have nothing to say? Wrong. Have read a paper? Take notes. Had an idea? Write it down. Proved a small result? Don’t believe it until you have the nuts and bolts on paper. Writing is your permanent companion. Writing takes time. Postponing the writing “to the end” is the worst idea you could have. Embarking on a big task is much harder than starting a small one. You would also miss feedback, iterations, and time to digest. Repeat: start writing now and write often!
  • A little pressure sometimes helps, too 😉

There is much more to say, about choosing the right journal (or conference), LPUs (least publishable units, a.k.a. salami tactic), about heaven and hell with co-authors, about competition and doubting yourself, about writing a “thesis in three months” (not!), about procrastination. In a follow-up post maybe.

Postscriptum. Sven Müller pointed to the Elements of Style for Writing Scientific Journal Articles. Thanks!


4 thoughts on “How to write a paper

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s