Daniel Lemire's blog

, 2 min read

Good students find questions, not answers

It is often believed that learning is a simple matter of collecting answers and replies. I suspect that “learn mechanistically how to answer the questions” is a great way for weak students to pass courses, and for smart students to ace courses. However, I believe that if you really want to learn the material, you have to learn to ask good questions and stop focusing on answers.

If you ask experienced professors, they will probably admit that some students who do very well on tests (A+ all the way) make uninspiring graduate students, while less stellar undergraduate students can end up being superb graduate students.

Part of the difference is that in graduate school, you have to ask questions first. A researcher’s job is to come up with new questions, not necessarily new answers. That is because once you have a good question, the answer can often be derived mechanistically.

It is somewhat counterintuitive if you have been trained in conventional schools. For example, you see mathematics as a set of “obvious questions” followed by “non-obvious answers”. And you think that, surely, the hard part was coming up with the answers.

This leads you to think that the way forward is to work really hard on really complicated answers to simple questions. But where will the questions come from?

What is a good question?

  1. A question should be at the frontier of the domain of knowledge as you understand it. To really master the material, you have to go beyond the surface and understand its limits.
  2. It should trigger interest from the part of the recipient. Your question should be interesting, not just to you, but to the person receiving the question. You may wonder why it should matter: it does because it puts a psychological bar, ruling out tempting but shallow questions.

A good question is usually prepared, not improvised. Most people need a few minutes to come up with a good question. But there are some quick recipes that often deliver good questions…

  • Questions about definitions… It is common that a speaker will use terms without defining them clearly. If you are unsure about how you would define a term, it might make for a great question.
  • « How do you know » questions are often great. If the expert tells you that programming in a given manner reduces memory usage, you may ask « How would I know ? »

The real reason for asking good questions is that the process itself will force you to learn. That is, if you are interested in learning, in really learning, you will not focus on coming up with answers, rather you will focus on coming up with questions.

You can see it play out if you pay attention. Go to an event where “expert learners” are trying to up their game. For example, a good workshop with people at the top of their game. If the event is well managed, you will observe that people are asking each other good questions.

Experienced professors do it effortlessly. The best engineers are great at it.

As an experiment, next time you encounter an expert in a field that you do not yet master, focus on trying to come up with good, non-obvious questions. I bet that you will learn a lot.