Background: I have done quite a lot of research work for a particular project. I am working in the field of operations research (i.e. applied math/physics), so this work primarily takes the form of propositions, proofs and numerical experiments. In the process of my research, each day I write up my daily progress in my lab notebook, which in my case takes the form of a very very long LaTeX file. I am now trying to write up my work as an article for submission to a journal.

Question: What is an efficient process which I can use to write the journal article?

Related links:

  • The Mumford method sounds interesting. However, as Mumford is a philosopher, his writing process seems to me to be less relevant to writing in the sciences.

2 Answers 2


I think the process of writing papers is pretty individual to each researcher. What works for me may not work for you. That being said, the following are a few rules that I usually give me mentees and students when we start writing a paper together:

  • As "Not Quite An Outsider" states, do start with an outline. Look through related work for papers with a similar scope / methodology / idea and start by imitating their outline. However, do not do so blindly. Instead, focus on understanding why the authors of your related work chose to structure their paper the way they did, and check whether their (assumed) reasoning is also useful for your paper.

  • Keep in mind that your paper needs a "story" - you know your material and reasoning, your readers do not. Start at the beginning, and end with the conclusions. Avoid statements that are not understandable at the point in the paper where they appear (rule of thumb: when you feel writing something along the lines of "as will be explained later on", you likely have a bug in your structure).

  • Plan the length of your paper. Fill each section with some Lorem Ipsum filler text of roughly the same length as you plan for this section. This allows you to see how much space you actually have for each part of your paper. During writing, when I start a new section I remove the filler text and replace it with what I actually plan to say at this point. I sometimes even go as far as drafting where in the paper I will have which figures, and put placeholder figures there instead when doing the outline.

  • Write the paper in order (i.e., in the same order as it will appear in the final paper). This is a bit controversial - I have seen many experienced paper writers suggest various other orders ("Start with Related Work" - "Start with Conclusions" - etc.). To me, the problem with writing in a different sequence is that it is very easy to lose track of what a reader actually knows at this point in the paper (hence destroying the story of your paper). For instance, you would end up using concepts and ideas that you actually only discuss at a later point. This makes papers unnecessary difficult to understand. I feel it is also much easier to produce a convincing line of argumentation when you produce the material in the same order as it will be read.

  • In any way, later rearrangements will be necessary. After reading the paper, you decide that you need to switch some subsections around, or that you do not need Section 3 at all. Stay flexible and don't be too much in love with your current outline just because it is how you initially wanted to do it. One additional sidenote in this is that you write your paper in LaTeX. This makes later changes in the outline trivial.

  • Write the text the way you suppose it should appear in the paper. Do not draft too much - there is no point writing throwaway text unless you really do not know how to write this section / part properly at this point. Only go on to the next section when the last one is pretty much done.

  • As soon as a section is pretty much done, get some feedback on it. Remember that your paper should already be coherent and complete up to this section, so there is no harm in sending it to colleagues or your supervisor and asking them to tell you whether the paper makes sense up to and including Section X.

  • +1 for "What works for me may not work for you." This method of writing a paper does not work for me at all, especially the advice about mostly finishing one section before working on the next. Crafting a paper is a dynamic process, and each section will require many revisits (and usually several rewrites!) before I am satisfied that it is as good as it can be. Feb 28, 2014 at 2:09

I should think an annotated outline (indented list of titles) would be effective. Do a basic outline of the article with however many titles, and under each title put one or more of:

  • link(s) to the text to use from your journal
  • a short phrase to use in conveying the point of the titled section
  • an idea of how much (word, line or paragraph count) to use to fill this out.
  • links to other outline titles.

Once the structure looks good (or even make a few structures to pick from), go back over and fill in the body under each title. The annotations will help you keep track of certain goals (word length, number of ideas, enough persuasive sentences, logical coherence), and it is easier to manage a high-level version of the article this way. It also makes editing easier if you prioritize which titles to cut.

  • If you are concerned about stylistic issues, again start with rendering existing articles into outline form, and then make your outline similar. Other advice such as style and idiom are useful, but poor structure is less forgiveable (in my opinion) than poor use of style or idiom. Feb 27, 2014 at 20:23

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .