It is absurd that the second year of my PhD has come to an end. I am taking a minute to reflect on what I have learnt and want to improve upon.

---

Overall I am super happy and feel very lucky to be in my PhD program. On an average week I have ~2 hours of commitments in total. The rest of the time is totally unstructured where I am free to learn and pursue whatever it is that I’m interested in. My interests are centered enough around my research that I manage to keep my supervisor and collaborators happy (at least to date!). But I still have lots of free time to explore and learn about so many other things. I doubt I’ll ever have this degree of freedom and flexibility again in my life.

Lessons:

• Make feedback loops as tight as possible. This is both in the quality of data produced from an experiment and the speed at which it can be run. Your first experiment is never going to work, everything in the real world is messy and requires fine tuning. Build a system so that it can scale and run very fast.
  • Getting more meta, your first system to run things quickly will certainly not be all that fast so also invest in maintaining and refactoring it over time. Expect growing pains.
  • 20% of the things you do will likely have 80% of the impact so take a lot of shots on target.
• If there is a problem with your system, fix it now before it becomes a problem later. In other words, prepare for the worst.

His years at sea had taught him that if you don’t fix something when you first see it beginning to fail, it is very likely to finish failing just when it is the most dangerous and the hardest to deal with, such as in the midst of a storm. Source

• Nothing is new under the sun - the more old papers you read the better.
• Use Twitter but only sparingly. It is a delicate dance between knowing what the state of the art is but also avoiding flavour of the month topics that are overly crowded.
• Be very careful when taking on mentees, especially time-poor and inexperienced undergrads. More hands do not always make lighter work. There are real benefits to doing everything yourself. The mentorship may be a net time sink for your research.
• Rotate with your highest priority labs first. I assumed that everyone would rotate for the first year and there would be an official point where everyone bid for and committed to a lab. In reality, it is far more ad hoc and if you don’t rotate with the labs you are most excited about and likely to join first then they may not have space by the time you rotate.
• If you can always work more than four hours in a day then you might be doing the wrong kinds of work. It seems like most people (including Fields medalists) can only do really deep work for approximately four hours a day. Other kinds of work where you are more on autopilot like implementing an idea or replying to emails you may be able to work for much longer. However, if you are always doing this kind of work and never thinking super deeply then you might be doing something wrong.
• Be explicit about taking time off. Flexible hours can be great until you feel like you can and should be working all the time. This drains the enjoyment from time not spent working unless you are explicit about actually taking it off.
• Think for longer, act more hesitantly. In hindsight there are a number of results that I could have foreseen if I had just thought for a bit longer and more deeply about the problem first. There are no hard deadlines in research (unlike in school). Use this to your advantage.
• But if something is cheap to test then don’t theorize about it, just actually go and test it, immediately. Right now.
• Most insights come unconsciously but you need to support their appearance. Read widely, go on deep dives into topics, ramble on and on in your notebook about ideas and see where it goes.
• Life is cyclical. Moods, motivation levels, the type of work needed to drive a project forwards. Accept this rather than pretending the cycles aren’t there.
• If you do ignore them… you might work at really suboptimal times and try to take shortcuts that hurt in the long run (see the bullet point about solving problems now rather than when they actually become problems).
• But don’t let perfect be the enemy of good. It is too easy to decide not to start something that seems hard because you don’t feel at your absolute peak right now. E.g. starting to write that paper or reading that publication. Everything is iterative and you will probably need to read it again or re-write it to get it perfect anyways, but you can still capture a lot of the variance the first time around.
• The publication itself and the main text in particular hide an incredible number of details, this makes talking to the paper authors really overpowered.
• Working alone can have its advantages. Collaboration is great but if you never work alone and never force yourself to come up with your own research ideas then those muscles will never grow. They certainly didn’t in school where there is always a right answer and a very small number of inductive leaps are required to solve any problem.
• However well you think you understand something, you certainly don’t until you go through the process of re-creating it for yourself.

What I cannot create I do not understand - Richard Feynman

• Everyone is naked. This is a dramatization of the phrase “the emperor has no clothes”. The more I climb ivory towers the more I realize that the people who you assumed knew everything and had everything under control often don’t. They are humans too. This is both terrifying because nobody is in control but also highly motivating because you can make a difference.
• Just because you can work on something doesn’t mean you should. Taking an idea all the way to a publication is incredibly time consuming. And there are fixed costs to the endeavor that don’t scale and aren’t particularly educational, e.g. making Figure 4 for the 30th time with a slightly different color scheme this time.
• When deciding if a project is worth pursuing think “if this went perfectly according to plan, what comes of it?” If the outcome in this unrealistic scenario isn’t all that great then look further.

My favourite science quote:

The Most Exciting Phrase in Science Is Not ‘Eureka!’ But ‘That’s funny …’ - Isaac Asimov

Oh and if you haven’t read You and Your Research then you should.

Goals:

• Read more textbooks. I need to gain new tools for thought and invest more in the long term instead of immediate projects.
• Find more collaborators. I have been working alone for too long. Moving to the Bay Area and working with the Redwood Theoretical Neuro Institute should really help with this.
• Do more theory and less engineering.
• Spend more time on maintenance. Unit tests, refactoring code, etc.
• Have a good answer to the Tyler Cowen question: “What form of routine practice do you do that is analogous to the way a pianist practices the scales?”
• Share my research more openly via blog posts and get things on ArXiv more frequently.
• Get better at accepting how the world actually is. There is something really powerful about seeing the world not as it should be or could have been but how it is right now. You’re in the cockpit, everything that has happened has happened, now what is going to happen next? I feel this most acutely with investment decisions, “ok yes I should have sold before the market crash. But what should I be doing right now?” Yet I think this generalizes to accepting that I am two years into grad school and turning 25 soon. What projects are sunk costs and what do I want to change going forwards? Meditation and journaling are two ways that I think can really help with this “world acceptance” and future planning.

---

Thanks to Max Farrens for reading drafts of this piece. All remaining errors are mine and mine alone.