A Year of Working on Scientific Papers

I started my PhD in January 2023, after doing some work in industry. While a lot of my work involved doing things that were novel from the engineering point of view, and sometimes involved reading research papers and implementing ideas from them, I had never worked towards a paper myself.

In some ways, the day-to-day work doing research was a lot more like engineering than I expected. You still have some engineering objectives, and you need to write the code to achieve them. A lot of the difference comes from setting the engineering objectives, and the constraints you have as a team member or coordinator when planning out work. As eloquently described in this excellent talk on writing research, the aim of a paper isn't to present data or outperform the state of the art, but rather to convince the reader, presumably already an expert in the field, that they will benefit from reading and understanding your work. All the data, explanations, and arguments have to be carefully picked and iterated on to satisfy this objective.

One aspect of research that surprised me was the resource allocation on paper projects. Researchers often have a lot of flexibility when picking their projects, so to attract collaborators you need to provide some sort of an incentive. One popular measure of a researcher is the number of publications at top conferences or journals, so you have to convince a potential teammate, potentially before any real work has started on the project, that there is a realistic chance that the work that they contribute to will be prestigious enough to benefit their Google scholar profile. I'm lucky that my professor has an excellent track record and can leverage his reputation to attract collaborators, a perk I hadn't considered before starting my PhD.

After a year of doing this work, I feel like I've made enough missteps to be able to impart some advice and guides to others starting out on their research career. I'm a big believer in checklists in complex processes, so I've formulated them as simple step-by-step guides. You can find them here. Please write to me if you have comments or advice to improve them.