For the past two years, I have been teaching undergraduate computer science at the University of Cambridge as a supervisor1. Over six academic terms, I have taught 48 students in pairs, each pair participating in three intensive, hour-long sessions where we discuss their submitted work and understanding of the given topic.

Many of the students I taught are much smarter than I am. Practically all of them excel in at least one area where I don’t. How then was I supposed to teach them? How was I supposed to answer questions that I didn’t have the answer to, even after hours of effort? It required some clever tactics, but mostly it needed a drastic change in the way I perceived teaching.

For most of my time as a student, I saw teaching as consisting of a supervisor who knew things, and a student who needed to know those things. The lesson was then the simple, structured process of moving the information from one brain to another. I thought that in becoming a supervisor I could simply pass the information that I’d been taught down the chain.

But as soon as I started preparing to supervise, I was hit with the startling realisation that I had no idea what I was doing. My understanding of content that got me decent grades in exams suddenly seemed woefully inadequate. I assumed I would be able to answer questions correctly on the spot, but when running through old lectures, I kept coming up with imagined questions that sent me down endless rabbit holes. I was trying to achieve an impossibly perfect understanding, with the assumption that other supervisors had this.

However, jumping into my first supervisions, I realised that they were much more dynamic than I had assumed. Instead of just transferring knowledge, I was providing a space for the student to work things out for themselves, gently guiding them. I often had to overcome the urge to blurt out the immediate answer that came to me, and basically shut the hell up while they rolled the problem over in their heads. This is difficult to do, but the students (and I) are often rewarded with a much deeper insight than I would have come up with alone. I can’t stress enough how productive it can be, in response to a question, to ask “what do you think?” and then say nothing else.

There is also a strong asymmetry between the supervisor and the students: they may be smarter, but I have the home-field advantage. Students have to juggle many different topics with content that they’ve only just become aware of, whereas I’ve had multiple years to mull over specific areas, making those all-important subconscious connections. A student sees each supervision only once, whereas I see it over and over again, noticing the common misunderstandings. What I learn in one supervision can be shared with many others in future sessions. These are not tactics to make myself look smart, but opportunities to guide the students.

Most importantly, I do not present myself as an unquestionable authority. I have to drop my ego and be OK with the students knowing that there’s an awful lot I don’t know. Doing this feels so liberating, and it changes the feel of supervisions from adversarial to collaborative, though you have to accept that some students will look down on you for breaking this illusion that they’ve come to be used to.

Once you answer a question with “I don’t know, but let’s work through it together,” it feels good to combine your experience with their intelligence. The process benefits both sides, and I can attribute a lot of my current understanding to my students.


Tips for supervising

  • Only supervise one topic per term. The vast majority of your work time is spent before supervisions, understanding the lecture notes and related reading to the point that you can teach it. Once you have done this work, the marginal cost of doing more supervisions on the same topic is comparatively negligible. It is almost always preferable to do more supervisions in one topic than to diversify. For the same reason, try to repeat the topic the next year instead of switching to another.

  • It goes without saying, but you should run through your set questions yourself a couple of times to see how much of a pain in the ass they are to properly answer. If you won’t do it, why should your students? This will help you to preempt common pitfalls by running into them yourself.

  • Be empathetic. A Cambridge degree is a hard process to go through — think about what you would have wanted your supervisors to do when you were in a bad spot, and do that.


  1. Oxford has an equivalent system where tutors give tutorials to tutees