A pragmatic method for picking PhD research projects

In: How-To, Research
Published on 2025-06-17
Written from the perspective of a machine learning engineer and business owner.

This blogpost applies (as all my PhD related writings) to a PhD in computer science in the Netherlands. It will be different for other research fields and other countries.

In The Netherlands, most PhD students get a contract as university employees for 4 years. After the first 12-18 months there is an evaluation moment. By then, you are expected to have either your first conference paper under your belt (co-written with your supervisor), or at least have a paper manuscript ready for publication.

In my experience you can defend your PhD in computer science if you managed to get 4 conference papers published and gather these in a PhD thesis with an acceptable "narrative arc": a logical string of research questions within a specific narrow research field.

That means that if you manage to do one research project per year and publish it in a conference, you're on track with your PhD.

Too many ideas, too little time

Of course, as a PhD student you're always pressed for time.

Apart from reading papers, writing code, conducting experiments, doing evaluations, and writing papers, you're also spending time on teaching, mandatory coursework, meetings, helping your colleagues, loafing around the coffee machine, traveling, and exchanging ideas with highly-educated strangers. You might do an internship at a company, or perform extra academic service activities to prepare for your next job interview: (sub)reviewing papers, organizing workshops, and so on.

So your research must be reasonably efficient.

To make matters worse, you probably have more research ideas than you can execute on during your PhD time.

You're especially liable to Research Idea Overabundance if:

Choose anyway

The list below describes a very pragmatic approach for selecting research ideas.

You can use this list of statements to find the most promising research ideas, in such a way that will probably allow you to finish your PhD within the fixed term of 4 years.

The checklist is meant for selecting research projects that take 3-6 months, and have a reasonable chance of succeeding and getting published.

If you do one such project every 6 months, you can do 7 full research projects in 3.5 years and then still have take 6 months to finish up your thesis and find your next job. Not all your projects might end up in your PhD thesis, and that's fine. The focus of your thesis will probably shift organically over time, as you collaborate with different colleagues and develop your own research vision.

I would argue that it's even possible to do a 6-month research project in the first half year of your PhD. In fact, I highly recommend it. Just treat it as a more advanced master's thesis. It probably won't be your best work, because when you start, you're still feeling out the nuances of your research field. But you will immediately get your feet wet, and you'll probably learn a lot faster than when you're spending your first 6 months just reading survey papers and worrying about your qualifier exams.

My experience

I used this approach to pick research projects during my PhD, and I was finished within 4 years. I would strive to do two 6-month projects per year, and have time left for submitting papers, waiting for reviews, and dealing with unforeseen setbacks.

In the end I finished with 10 publications -- written between 2018 and 2021 with various wonderful co-authors -- of which several were suitable for inclusion in the narrative arc of my PhD thesis.

Checklist for selecting a 6-month research project

  1. Something similar has been done before, but not exactly the same. List similar papers.

  2. You can find peer-reviewed literature on this topic. List papers.

  3. It's clear which subfield of your work domain this research falls into. List the most important keywords you will use to describe the research project.

  4. You can come up with a good validation method for your research. List papers that use same methodology.

  5. All necessary data is available or can easily be created. List links to datasets.

  6. All necessary code is available or can easily be created. List links to git repositories, models, Python packages etc.

  7. All necessary hardware, infrastructure and other resources are available or can easily be obtained. List resources and how/where you will get them.

  8. It's clear what problems you will encounter during your research. List of problems and how you will tackle them.

  9. There is someone who can help you if you get stuck with your research. List names, contact details and their expertise.

  10. The research is interesting from an academic perspective. Do ancient men with beards established people in your field find this interesting or remarkable? Describe your reasoning.

  11. Bonus points: the research is interesting from a commercial/economic perspective. Do entrepreneurs, companies, multinationals benefit from it? Describe your reasoning.

  12. Bonus points: the research is interesting from a societal perspective. Does your mother, grandmother, or neighbor benefit from it? People from other countries? Specific groups? Extend this question according to your personal values. Describe your reasoning.

  13. There is a journal or conference where the paper that describes this research will fit in. List names of publications. You get bonus points if your paper will cite other papers from this publication.

  14. The research can be completed within an estimated 6 months. Incorporate time for unforeseen problems: plan for 4 months and the work will probably be done in 6 months. Add a rough weekly planning.

  15. The research project cannot be broken up into sub-projects worthy of a full conference paper. If your research projects can be divided up even further: that's great! Write down your vision for the larger project, and go through this checklist for each subproject. No cheating.

  16. John Regehr wrote: "Every good research idea is a departure from the accepted wisdom, but it's important to depart at the right level." So... is your departure from accepted wisdom at the right level? Find the middle ground between "Everyone is wrong! I know better!" and "This small insignificant change has no impact on the world whatsoever." Describe your reasoning.

  17. The paper fits into your envisioned narrative arc of your PhD thesis. Describe how the main research question of the project relates to your other papers or research topics.

I hope this list helps you with finishing your PhD on time. Feel free to also send this to your master students when they're struggling to pick a worthwhile topic for their master thesis.