Posted by Jesper on August 1, 2023
Good writing is important, and it takes time to get it right. Over the last couple of months, I have been reading and giving feedback on way too many papers written by bachelor students, master students, PhD students, and people who should really know better. To help both my students and my future self, I am thus inspired to write down some of the issues that I encountered again and again. I tried to keep them short and sweet, but I might come back to this post and expand it if there is interest (just shoot me an email or a toot). Also, I included some links to other sources that I find helpful at the bottom.
Lack of context
When you write a paper you do so from a very specific context: you are at a specific university, working with a specific supervisor, working on a specific goal in mind. But if you want your paper to be read by people other than your supervisor or direct collaborators, you need to step outside of that context and think about questions that might be obvious to you. Where is this problem situated in the broader academic field? What are the questions that are considered hard or important in your sub-field? What kind of people would benefit if this problem were solved? What are commonly used techniques for working on these problems?
Lack of motivation
It’s just as important to explain why you are writing this paper as it is to describe what you did. In fact, it’s more important, as a reader who’s not convinced of the goal of the paper won’t read any further. Readers usually do not yet know about the problem you want to solve, so you do really need to explain what you want to accomplish, why this is important, why this is useful, … So motivation should be your first concern when writing the paper, and should ideally already be clear from the abstract.
Lack of focus
Your paper should have at its core a main idea for which you want to convince the reader of its importance. All other parts of your paper should be assessed through the lens of this main idea: if it does not contribute to it, then it’s often better to throw it out.
Lack of examples
Examples are always good, especially when the general case of the thing you are explaining is complex or general. They force you to be concrete and precise. Make sure that each example is as simple as it can be and does not rely on anything that is unexplained or not relevant to the thing it is an example of.
Lack of signposting
A good academic text needs to make its structure obvious. The reader should be able to navigate easily to the parts of the text that are most relevant to them. Of course some of this is accomplished by a proper structure of sections and subsections, but this on itself is not sufficient. Before jumping into the specific, you first need to explain the general thing. For example, if you have a section with three subsections identifying three different challenges, you should name the three challenges before the first subsection, and explain how they are connected.
This applies on the level of the whole paper as well as the level of an individual paragraph. When you explain a set of typing rules or some pseudocode, don’t just go through it from start to finish, but instead start from the high-level view of the overall structure and then zoom into details.
Keeping the reader in suspense
Your text is not a work of fiction so you don’t have to worry about keeping the reader in suspense. If the reader has gotten out of your paper what they need after the first two pages and stops reading, then that’s a success. So put the most important results and conclusions front and center in the introduction of your paper so readers don’t have to look for them.
Explaining all solutions that do not work before giving your solution
I understand the urge to explain all the previous attempts at solving the same problem and where they went wrong. However, this is both confusing to readers unfamiliar with the previous work (since they don’t want to learn about solutions that don’t work) and frustrating to people who do know it (since there’s nothing new for them). Instead the right place for such comparisons is in the related work section, which should come after the main technical part of the paper.
Assuming the reader already knows technical terms
If you use a technical term, then you should introduce it either by explaining what it is and/or by giving a citation to where the reader can find it. What exactly is a “technical term” is dependent on the audience but it’s good to err on the side of safety. Your text might be read by someone from a different research area, or by someone reading this in 30 years when the terminology has changed.
Assuming pictures or code blocks or inference rules are self-explanatory
Spoiler alert: they are not. Explain them in plain text.
Using imprecise or convoluted language
These are some more micro-level issues that seem inconsequential but can very much distract from the main point of the text. I recommend keeping a list of the ones that often appear in your own writing and doing a search-and-replace for them during proofreading.
- Using comparatives without a point of comparison: writing that your approach is “more expressive”, “simpler”, “more secure”, … without mentioning what is the point of comparison.
- Imprecise language such as “some”, “a lot”, “many”, “significantly”, “greatly”, “drastically”, “extremely”, “a number of”, “very”, “important”, “crucial”, “something like”, “things”, … Instead, be precise about what you observe and how it compares and let the reader judge for themselves.
- Superfluous turns of phrase: “The reader might have noticed that…”, “We argue that…”, “Note that…”, “It is important that…”, “We say that…”, “We know that…”, “As previously mentioned…”, …
- Using passive voice: “our approach was evaluated” instead of “we evaluate our approach”.
- Using future or past tense where present tense would be just fine: “In this paper, we will show…” and “In this paper, we have shown…” are both needlessly wordy. “In this paper, we show” is more direct and works just fine.
- Using words that serve to make the reader feel stupid: “simple”, “elementary”, “straightforward”, “basic”, “trivial”, “just”, “of course”, …
- Demonstrative pronouns without a clear referent: “This is…”, “these things are…”
- Book: The Scientist’s Guide to Writing by Stephen Heard
- Book: Trees, Maps, and Theorems by Jean-Luc Dumont
- Video: How to write a great research paper by Simon Peyton Jones
- Video: How to write papers so people can read them by Derek Dreyer
- Video: How to explain things real good by Nicki Case
- Web page: Writing tips by Eelco Visser
- Blog post: Some Research Paper Writing Recommendations by Arie van Deursen
- Blog post: My top ten presentation issues in other’s papers by Andreas Zeller