The Writing Rules

These rules are for my research lab.

Do not use ChatGPT or any other genAI tool. Since 2022, my students and my research team have been increasingly using ChatGPT to help (re)write sentences, abstracts, or even entire paragraphs and sections of a paper. The argument is that it looks more professional and has fewer errors. As a reader, I am tired of reading generated text: It is often repetitive and shallow. It can miss entirely the point you are trying to make or worse make up an entirely different point. You can write it better. Even if it is rougher, or seems less polished, we understand you better when we hear your voice. We also spend more time editing out the generated fillers than refining rough text. 

The bigger concern here is that “writing is thinking”. If you let a tool do the writing for you, you are not clarifying your ideas, figuring out the bigger picture or logically constructing and validating your arguments and algorithms.

This rule applies even to abstracts. If you can’t distill the main contributions and claims of your paper into a single paragraph, how will you give an elevator pitch to anyone who asks about your research?

Do not fill space. If you can write it clearly in a sentence, don’t write a paragraph! If you need a paragraph, don’t write a page, etc. Brevity is an art and one you should perfect. From now on, ignore whatever notion you may have about how much space something should occupy. Some of our conferences penalize papers that are unnecessarily wordy. Gassy writing stinks!

Do not use passive voice. Experiments were not done, you did them. Show some professional accountability by owning up to the work. I have zero-tolerance for passive voice. It has no place in scientific writing. If you are intentionally obfuscating attribution, we have bigger problems. Also, active voice is easier to read and flows better.

Do not start with “Data is exploding …” or other tropes. I get that it might help you get over the empty page intro block, but how about you just get into it and tell us what the problem is; the rest will naturally come together.

Do not cite papers you haven’t read. If you can’t meaningfully describe a system in related works and how it compares with your work, don’t put it in. Everyone is tired of lazy citation lists. Also don’t use citation lists to back common sense claims such as “database systems are used in healthcare [1, 2, 3, …100], finance [100, … , 200], blah, blah, blah …”

Do not be pompous. Write simple and clear. Think “KISS — Keep it simple, stupid.” This rules also applies to algorithms, mathematical formalisms, etc. Strive for simplicity. I will institute a pomposity jar: you will pay a dollar for every¹ Greek alphabet, ten for every subscript or superscript, and a hundred for every convoluted or over-the-top sentence. All proceeds will go towards a coffee fund for the readers who have to go through your paper.

Disclaimer: Writing rules are not legally binding. I bestow the rights of judge, jury and enforcer to myself. Violators can pay in the local currency: 1 US Dollar=3.65 UAE Dirhams. No students have been charged … yet.

1. I considered adding “extraneous” to the use of Greek notation but then if I’m going over a proof or algorithm, I sure do need the coffee. ↩︎