Recently I have been asked by my School, and my Research Group, to give a talk about how to write a good paper. Here are the notes of my talk, but please do not take them too seriously. There are many ways to write a good paper, this is just my personal viewpoint.
Abstract for the talk.
I consider the ingredients of a good paper. a) problem inspired by a practical setting b) a clear contribution c) a strong methodical background d) co-authors that contribute fully to the design of the research and its execution e) a complete analysis of the problem, even if it means discarding work or starting again f) incorporating feedback via lectures, seminars, and conference presentations g) a comprehensive review process, where all of the referees comments are fully explored and dealt with h) a considered approach where ideas are left to mature so a fresh perspective can be gained.
Inspired by a practical setting: The topic should be empirically based, relevant to the real-world. It should be an important problem. Data should be available and observable/measurable in principle.
Clear Contribution: What literature is it building upon? Nothing is completely new, if you think it is new, you have not read your literature properly. But, what is new in this paper? Make it clear and compelling. Create a nice hook at the beginning of the paper to draw the reader in.
Strong methodical background: Appropriate and rigorous method. Use maths, logic, reason, statistics. Make sure your arguments are repeatable by others, causally-linked, not open to misinterpretations/misunderstanding. Your logic should be irrefutable.
A complete analysis of the problem: Do not salami slice the problem into the smallest publishable units-solve the complete problem. Don’t be afraid to throw it all away and start again, it will be quicker, cleaner, and better next time. Build up the story from first principles, simple example, complete solution, and then the extensions. More suppliers, constraints, optimality gaps, different industries/countries/products/other generalizations. Where and how could your research findings be applied in other settings?
Horsepower: A great team of the right size needs to be assembled. All team members need to contribute to all stages of the paper’s design and development; everyone needs to put his or her capacity to the problem pushing the envelope at all stages. A post-box style paper, where one author does one section, and another does the next, just does not bring the breadth of knowledge to bear on the problem, niether will it read very well. Professors should contribute more than the final proofread before submission!
At least one co-author should have a profound knowledge of the research area, (others can be technical experts), otherwise, how could you properly identify the contribution? Collaborate, but not too much. Get the right-sized team, get a strategic focus.
Present your work: Present at research group meetings, teach it in-class, present to the general public, your professional bodies, academic conferences, and invited lectures at other Universities. Get feedback, get insights into what makes a good story. Make your ideas concrete by verbalizing them, use the feedback to make your paper better, stronger, easier to understand, more accessible, shorter. Remember the paper is there to convey an idea, not to make you look clever.
- Clear, coherent, and concise.
- Short sentences.
- Use plain English. Avoid ornate, pompous, or waffly prose. If the reader had to reach for the dictionary you failed.
- Precise. Avoid vague and imprecise statements.
- Active doing words.
- Tell a story, write a compelling narrative.
- Avoid too much jargon.
Give it time: A good paper can take 2-3 years to write/re-work/resubmit or submit to another journal. Be patient. Let the paper rest, like a good bottle of wine. When you write, you hold ideas in your head, and you read more into the words than what is written down. Put your paper aside for a few weeks, work on another. When you come back and read your words afresh, you can be really sure about whether you have written it correctly. Can you reconstruct your findings/argument from what is written?
Proofreading: You will need to read your paper 30+ times before it is properly proofread. You need to read it again and again until you can find no more improvements/corrections. Try swapping the beginning and end of a sentence around, often you can find unnecessary words to remove. Spell check, delete your dictionary every other month, use software like Grammarly, or proofreading services like Scribendi or word analyzers like the WritersDiet.com and the Hemingway Editor. Read the paper aloud-Adobe can do this for you-or get someone else to read it to you. The paper is finished not when nothing is left to add but when nothing is left to take away. Write every day, the more you do, the stronger you get.
Review process: The reviewers have given up their time in good faith to help you improve the paper-we have to believe this! If they don’t understand/have got confused it is because you did not explain it properly. Address all the reviewers’ comments, don’t just bat them away by saying “for the next paper”, this is the next paper! Aim for a top journal, as if the paper gets rejected, address the referees comments, send to another journal. Eventually, the paper will become good enough to be published somewhere.
Less is more: Switching between too many tasks makes you inefficient. Learn to say no. Prioritise. Obsessive focus on a few good papers is better than trying to write all the papers. Do less, then obsess. If you don’t obsess, then you will never do as well as those that said yes to everything because you did not do enough.