How to find a research topic worth your next three years

Choosing a research topic is the single highest-leverage decision in graduate school. Most students underweight it — they pick something "interesting," start reading, and discover eighteen months in that the topic is either un-finishable or already done. There is a better method. This guide walks through the two-week search process we use in Research Goal's methodology cohort.

The three intersections#

A topic worth three years sits at the intersection of three things. Drop any one of them and the work either stalls, gets scooped, or burns you out.

1. Curiosity — do you still want to think about it on a Sunday?#

You will think about this topic every day for three years. Curiosity is not a luxury — it is the energy that gets you past the inevitable trough. If the question doesn't make you want to read about it on a weekend with no deadline, it is not yet the right topic.

2. Contribution — does the field need this answered?#

Curiosity alone doesn't pay. The topic must close a gap the field can name. The cleanest test: when you describe the topic to a senior researcher at your university, do they say "interesting" politely, or do they immediately suggest one or two papers you should read? Suggested papers means the field has a conversation you're stepping into.

3. Feasibility — can you finish it with what you have?#

The third intersection is the one most students skip. Can you actually finish this work in three years with the resources, data access, supervision, and method-fluency you have right now? "I'll learn the method as I go" is the most common failure mode — the topic that requires you to also become a different researcher is rarely the topic that gets defended.

A two-week search method#

Most students search for a topic by reading widely until something resonates. That is passive and unbounded. A better method is active and time-boxed.

1. Days 1–3 — list 20 candidate topics. Write them down as one-sentence questions. Quantity over quality. Aim for 20.
2. Days 4–6 — score each on the three intersections (curiosity / contribution / feasibility) from 1 to 5. Drop anything with a single 1.
3. Days 7–9 — for the top 5, find one canonical paper each and read it slowly. Does the paper open questions you'd want to answer, or close them?
4. Days 10–12 — talk to three people about your top 2. Your advisor, one senior researcher in the area, one peer. Listen for which question keeps coming up unprompted.
5. Days 13–14 — write a one-paragraph proposal for your top topic. If you can write it in 90 minutes, the topic is real.

Questions to ask any candidate topic#

• Who is the audience for this work? — name three real readers, by role
• What changes if I'm right? If I'm wrong? — both answers should be non-trivial
• Is the data accessible from where I sit? — and if not, do I have a clear path to it within six months?
• Can I write a single-sentence contribution claim? — if not, the topic is still too vague
• Has someone in the last five years done something close? — if no, ask why; if yes, what does mine add?

The red flag every weak dissertation shares#

After supervising and reading hundreds of theses, one signal predicts a weak dissertation more reliably than any other: the topic and the method were chosen at the same time. Strong dissertations choose the question first, then the method to answer it. Weak dissertations choose the method ("I want to do machine learning") and shop for a question. If you can name your method before your question, pause.

Wrapping up#

Two weeks of disciplined search beats six months of casual reading. The right topic is the one that sits at the centre of all three circles — curiosity, contribution, feasibility — and holds up to conversation with the three people you most respect. Spend the time now; it pays back every week of the three years that follow.