Highly Opinionated Advice for (CS) PhD Happiness

January 7, 2022

I’ve been miserable for almost my entire PhD. Fortunately, this year (my fifth year!) things started to improve. And by things I mean my outlook, mood, and happiness levels, not my publication rate. After some reflection, I decided to record advice I never received that I wish I had, both before choosing where to go to graduate school and during my PhD itself. I believe this advice would have made me happier (not necessarily more “successful”) earlier. I have no idea if this advice will generalize at all.*

Note that the Internet (mostly HN) will cheerfully tell you that getting a PhD is a largely worthless endeavor both in terms of your career and overall well-being. But your reasons for getting a PhD are not the subject of this essay. I assume that you already want a PhD in computer science.

I also assume that you want to be happy during your PhD. Happiness is an important and, in my experience, neglected aspect of the process. I can tell you firsthand that most of the students that drop out of the program do so because they’re miserable, not because they’re not “smart enough”. Contrary to the PhD’s reputation, I don’t think misery is a prerequisite for publication success (and I can tell you firsthand that it’s certainly not sufficient). And if it is, why do it?

Consider Seeing a Therapist Before Starting

During your PhD, it is too easy to correlate external feedback with personal self-worth. Many PhD students come from undergraduate environments in which they received almost exclusively positive external feedback (e.g., straight As). That kind of positive feedback is, for many reasons, hard to come by during the PhD.

First of all, undergraduate work—in which the student is given a task and graded on how well they complete that task—is unlike research, where 90% of the work may be determining what exactly that task is. Second of all, professors (the same ones giving out those As!) typically have higher standards for graduate-level research compared with undergraduate work. Finally, conference paper reviews, the most consequential feedback of the PhD, are simply not predictable—hard work simply doesn’t (necessarily) lead to to positive feedback (paper acceptance). And frankly, there is a lot of luck involved. For instance: Is your work timely? Were you scooped? Did the reviewers even read your paper?

I think the PhD does a good job of seeking out and agitating underlying pathologies like over-achievement. Take a few students who have been conditioned to anchor their self-worth to external feedback and put them in an environment in which feedback is vague, harsh, and non-deterministic and you’ve got a recipe for misery. Addressing these issues before/as they come up is better than addressing them after they’ve festered for a few years. Many schools have (free) mental health resources and relationships with external, often subsidized, therapists and psychiatrists.

From personal experience, seeing both a therapist and psychiatrist has helped me immensely with dealing with failure, putting too much pressure on myself, and more.

Evaluate the Lab Culture, Not Just the Advisor

I think the value of the PhD can be boiled down to a single experience: sitting in the computer lab with one or more of your fellow students, tossing markers at the whiteboard, riffing on interesting and often silly ideas. Maybe you pair-program on a problem for a couple hours before grabbing lunch. Those couple hours often amount to nothing—the PhD should, at least, give you the freedom to explore stupid ideas. Most professors I’ve talked to that look back on their PhDs with fondness have these types of memories. They remember working hard, but they mostly remember the freedom to explore ideas with other, like-minded students.

But most professors will tell prospective PhD students to evaluate the advisor, not the school, when choosing a program. This is insufficient advice—evaluate the lab, not just the advisor or the school! After all, the students and post-docs in the lab are the ones you’ll be spending most of the time with, eating lunch with, and (hopefully) working on projects with. Is there a culture of collaboration, or does work happen in isolated silos? Is there a strict hierarchy for publications (senior students are always first authors) or are there younger students leading projects? Is there a toxic hustle-culture? Is there collaboration between different professors’ students?

You should think carefully about what type of work environment you’d like in terms of the people you work with. I personally believe PhD students work best when they are surrounded by like-minded students interested in similar topics. Further, I think it makes the transition from undergrad/industry easier when first year students can join an ongoing project under a senior student or post-doc. This fosters a sense of teamwork and provides a nice on-ramp to research. I think prospective students should avoid environments in which students rarely collaborate and tend to work in isolation.

(As a side note, I don’t think these collaborative experiences can be adequately digitized, and as a result I believe the pandemic hit us PhD students especially hard.)

Seek Out Advisor Fault Tolerance

Choosing the right advisor is a comically difficult task. You’re expected to choose the person who will mentor you, almost exclusively, for 4-7 years after meeting them for maybe a few hours if you’re lucky? You can read their publication record online, but it’s much harder to evaluate whether your working styles are compatible and whether they’re going to treat you with respect.

I suggest giving yourself the freedom to make the wrong choice by choosing a school with many potential advisors. Choose a school where there are at least three professors working in your area, and don’t be afraid to engage them in your early PhD years. If your prescribed advisor does not work out, don’t be afraid to switch.

Take Up a Highly Engaging and Social Hobby

Many PhD students feel immense pressure to “live in the lab” during their PhD. This lifestyle is rarely explicitly prescribed, but is reinforced through student behavior. There is a dangerous hustle culture in the PhD—how little sleep you’re getting, how many hours you stay in the lab, how many weekends you worked, all become points of pride. I don’t think most professors force their students into this lifestyle (though some do) but I don’t think enough of them try to instill healthy lifestyle habits into their students. They are, after all, a product of the same system, and breaking such cycles is hard.

The PhD can feel like your world. One of the best things you can do for your mental health during your PhD is to engage with another world. This means finding a hobby that:

  1. Happens regularly (multiple times a week).
  2. Includes regular interaction with other people not in your PhD program.
  3. Is not related to your PhD at all.
  4. Includes physical exercise (optional of course, but highly recommended for the obvious health benefits).

Think local sports teams, martial arts clubs, group piano lessons, running clubs, language-learning clubs, etc. Engaging in this new world can provide new friends, a crucial mental break from all things associated with the PhD, and (if you follow rule 4) some refreshing physical activity.

I personally train Brazilian jiu-jitsu 3-4 times a week, and I consider it non-negotiable for my mental health. It’s important to “pay yourself first” with your time here—it’s all too easy to think that because your fellow PhD students don’t leave after 6pm, you shouldn’t leave (to engage with your other world) either. Try to commit to your activity before you plan how much you’ll be working, and consider your activity a crucial component of your overall well-being.

Go On Summer Research Internships Early

I did not go on an internship until the summer after my fourth year, and I regret it. I was generally discouraged from going on internships during the summers until I and my research was “ready”. I think this was completely backwards—going on an internship early would have, among other things:

  1. Exposed me to exciting open problems, according to industrial research.
  2. Helped me form early connections with researchers not at my university.
  3. Informed my decision regarding whether to pursue academia or industrial research (or neither).
  4. Helped me not feel so isolated—industry teams typically collaborate much more than PhD “teams” (labs).
  5. Earned me more money. Money is one of the biggest stressors, not just for PhD students, and PhD stipends are abysmally low. Supplementing your yearly stipends with industry paychecks can provide much-needed relief.

Conclusion

Professors, PhD students, and PhDs often promote the idea that the PhD is a miserable trial-by-fire. It is certainly challenging—the most challenging thing I’ve ever done. But this reputation has sinister consequences: PhD students start to feel like they should be miserable. I think that’s silly. Something can be extremely difficult without destroying your well-being. It’s something I’m still working on, and these tips would have helped 23-year-old me.



* Who am I?

I’m a fifth year PhD student at a private, Ivy-League university in an urban setting. I study computer science, broadly systems and web security, in a pretty small (in terms of people) program (for my university) in a smaller sub-discipline (for my university). I started my PhD immediately after graduating from my undergraduate degree, though I did three 6-month internships (co-ops) during that degree.

Everything I say here is framed by my personal circumstances, and might not generalize at all. PhD students in the humanities likely have vastly different experiences and perspectives. In fact, PhD students in theoretical computer science likely also have vastly different experiences.

I am also not a doctor (MD), and this is not medical advice.