There were many moments during undergrad where I did not believe I had the strength or capacity to do a PhD.
On the other hand, I’ve always loved this part of Walt Whitman’s Leaves of Grass:
Do I contradict myself?
Very well then I contradict myself;
(I am large, I contain multitudes.)
As I write this, I’m currently wrapping up the second year of my PhD at UC Berkeley’s School of Information. I’d like to tally up several reflections on things I’ve learned so far. This is not as good of a guide as others have written in the past, since it is very much tailored to my own experience, but there may be bits and pieces that are helpful for others.
Time
It feels like I am still stuck in my first year, suspended in time by the monotony and isolation of the pandemic. Taking control of this warping of time while navigating an obstacle course of meetings, classes, and unstructured research/homework screen time is still something I’m learning to do.
My ability to handle unstructured time was especially tested during the summer between my first and second years, in 2020. I was doing research full-time that summer, and a lack of classes meant my calendar was a bit emptier than usual. I knew that structure had helped my sanity a bit, so I tried to introduce some by hosting a regular cross-institutional PhD reading group. Remote work nearly erased the divide between work and home, but hobbies like gardening helped bring a bit of it back.
Still, it’s taking a lot of practice to learn how to work productively and consistently. The Pomodoro technique doesn’t work for me, because once I’m in the flow of things, I usually don’t want to leave, and the moment I allow myself too much of a break, I derail easily. This means my rhythm of working involves longer spans of time, with shorter breaks. Overriding the initial resistance to work is the biggest challenge for me. To handle this, I break tasks down into very manageable pieces. Sometime I plan my mornings the night before so that my sleepy head the next day will have fewer things to think about. I write pseudocode with enough detail so that my future self just needs to fit it with Python syntax. I create checklists on Trello containing small tasks so that I feel tiny bits of victory when I finish and check each box. I acknowledge that sometimes I plan more than I can do in a day, in a week, or in a month, and forgive myself for being ambitious.
I‘ve tried multiple times to stick to a 9–5 schedule, but my love for sleep clashes with my love of late nights, and the hectic tempo of data processing scripts starting/failing/finishing prevented that from ever sticking. Plus, there is a wondrous calm when grocery shopping on weekday mornings. My peak work time is often in the evenings, when meetings don’t interrupt my flow, especially during the winter months when the daylight outside is precious. The challenges of this schedule is that sometimes I don’t align with my non-student friends, and I still need to have a very clear mental division of work time and non-work time. During non-work time, I try to not think too much about work. If I worry too much about something work related, I write it down, and reassure myself that my notes are thinking about it and storing it for me, so that I can pick it up again later. I only work on weekends if it will make my overall quality of life better during the week (maybe the lines are shorter at that popular ramen place on Mondays!).
Saying “no” is a great skill to have in academia, and I learned that everything, no matter how small, takes time. There is also a gray area between “yes” and “no”, where I decide what to pour more effort and time into, and what to do quickly. This isn’t a free pass for me to be lazy about everything, but it allows me to pour more into things that matter. This means I get to say “shhhhh” to my internal perfectionist. Example: the reading response paragraph I write for a class discussion doesn’t need to be my magnum opus, unless writing it sparks unbelievable amounts of joy.
How do I know when to say “no”, “heck yes”, or “sure ok”? I’m still figuring out the heuristics for this, but I think central to my time are three things: research, people, and wellness.
Research
When I was in undergrad starting out in research, I would struggle to know what to do next when working on projects. I remember asking a professor during a PhD visit how they knew what to work on, and they said, “Read lots of papers.” This is advice that’s worked pretty well for me, though I do have a habit of saving papers faster than I can read them. I skim papers like a kid reading a picture book, focusing on the figures and tables first and appreciating ones that are particularly well-labeled with the main takeaways. I’ve recently switched to Paperpile to manage my highlighting and notes.
When picking papers to read, I imagine myself situated in a network of subfields and larger disciplines. It’s tricky knowing what papers to prioritize and where in this network to pick papers from, but I typically categorize papers as part of two general groups: papers directly relevant to my current research, and papers I read for breadth. I’ve noticed that people tend to cite their friends, or in the United States, researchers from other American institutions. I’m actively trying to read more papers from other countries and authors that I do not yet know about.
I was lucky to start my PhD with some knowledge of what I wanted to do already, and some graduate-level coursework already under my belt. I’d already gone through a crisis, near the end of college, of deciding between cognitive science and social NLP. I knew about a lot of researchers in the field and my subarea. Still, the early stages of a PhD involve a lot of exploration. I can feel my center of gravity shifting a bit, especially now that I’ve been exposed to my peers’ interests in HCI and AI fairness. For students who enter a PhD with a less defined path in front of them, I’d suggest initially working with more senior students and finding mentors who are not just your advisor. Your initial work does not need to be the best you’ll ever do during your PhD, and writing up anything is an achievement to be proud of.
I’ve had plenty of bumps along the way (tbt when I forgot to delete a pricey TPU instance one weekend…). I experienced two journal review processes during the past two years, and doing R&Rs was challenging but great for my intellectual growth. Reading reviews on OpenReview is helpful for seeing how others critique research. It’s incredible how much work goes into something that might end up being a footnote or single sentence in a paper.
Sebastian Ruder, in his 10 tips for research and a PhD, says:
A good indication of whether something you are working on is impactful is whether you’d be excited if it was published by someone else.
I treat paper writing like a game of mad libs, where my rough drafts usually involve some structure but lots of blanks for final results and half-baked ideas. There’s storytelling involved as well, because research is supposed to tackle problems and questions (the conflict) and researchers should provide some scientific, engineering, or theoretical contribution (the resolution). I like to think of the potential story at the beginning and throughout implementing a project. I tend to get overly curious and fall down many time-consuming rabbit holes, and so remembering the main point is useful for pulling me back to more productive tasks.
I think a bit of my struggle during the summer of 2020 included a feeling that I was stuck in limbo. I was working on an R&R that didn’t seem to have a clear end in sight. The eventual acceptance of my first paper written entirely during my PhD gave me confidence, and also a bit of motivation to continue other projects. I’ve found that too much context-switching can impede my research, especially as a beginner still learning the ropes, so I try to stick to leading only one or two projects at a time.
For one-on-one research meetings with my advisor, I usually organize an agenda of notes beforehand. I try to prioritize these notes by leaving less essential items at the end, in case we run out of time. I think our most effective meetings usually involve me showing deliverables in the form of text printouts or graphs. Sometimes I make rough and slightly silly slides (usually I’d start by saying “This is not a presentation!”), just because slides provide bite-sized pacing of information and help us literally be on the same page.
Applying to internships sucks. Perhaps I’ll do a rant or guide on this later, but I don’t look forward to repeating the app and interview cycle in future years. (Get a referral before applying!!! Deadlines happen from Sept-Feb!) I’m incredibly lucky to have an internship for this upcoming summer.
People
When deciding where to situate myself in the world, I like to think about who I’d like to be around. Would these people be kind and supportive? Would they discuss interesting questions and ideas? For example, my advisor is very receptive to my local dessert recommendations.
The pandemic turned me into an extrovert, and I’m often flailing around trying to organize meetups and activities with others in pandemic-safe ways. Many of the best memories I’ve had of the past two years included other people: a trip to Point Reyes, an embarrassing number of boba runs, ice skating in Oakland. It’s saddening that the social momentum I gained during my first year was cut short, but I’m fortunate to have multiple pools of people to rely on: friends from college, EECS PhD students, and Information School PhD students. When conditions are safe to do so, I’m ready to make up for lost time and the memories-that-could’ve-been.
I also spend a bit of effort on connecting with other researchers outside of my institution. Partially this is because my PhD program is small, and the number of people working in my subarea of NLP is even smaller. Partially this is because I find interesting people everywhere across the different disciplines that make my work interdisciplinary. Knowing less than others about some topic means there is so much potential for me to learn new exciting things. I especially like the Sociologists of Digital Things Slack. Twitter is okay for connecting with others, in measured doses to prevent doomscrolling. A web presence seems to be necessary for PhD students these days, and I’m sure plenty of opportunities I’ve been given in the past two years were possible because I have one.
My relationships with non-academics help me stay grounded in the bigger picture. There is so much more to love than work and school, such as rare houseplants and the spicy kick of homemade hot pot. My friends and I once spent a day in SF seeking boba and free pizza, getting lost, and unintentionally coming across Coit Tower as the sun set.
I have also reflected on the meaning of my PhD to my family. My grandmother, who played a large role in raising me during the first ten years of my life, did not complete school beyond elementary. My parents immigrated to the United States, leaving the familiar and comfortable to be seen as perpetual foreigners. I’m an only child, which means the set of ideals outlined by the American Dream is a weight that I carry alone.
Wellness
In college, I was surrounded by very busy and very high achieving people. Students were often described as ducks: calm on the surface, while furiously paddling underneath the water. Though my overall college experience was positive, there were plenty of moments of panic, misery, and numbness. I remember working from the moment I woke up to the moment I went to sleep during the school year, sometimes timing alarms for the middle of the night to check if my models had converged. Sometimes I’d finally achieve something, but I was too tired to celebrate, or too anxious to believe that it was true. One of my friends asked, “Why do we work so hard to be unhappy?”
A fixation on academic and career performance brings along with it a great deal of stress, bad habits, and burnout that can be detrimental in the long run. The gradual realization that I can get happiness and meaning from things unrelated to school or career is one of the biggest lessons I’ve gained so far.
I learned that I’m affected by seasonal changes in weather and daylight, especially during long periods without in-person interaction. I’m grateful that I’m in California, and I spent a lot of time this winter trying to work outside in 50 degree weather just so that I got enough daylight. The side effects of the second dose of my Covid vaccine were pretty debilitating, but I was most surprised by how similar the fatigue I felt then was to how I felt while handling the darkness of winter.
Conclusion
The first two years were defined by many disruptions, but I was privileged in so many ways. I’ve found blogs written by researchers that reveal some aspect of how they balance personal with work has made the challenges of academia and industry more transparent and humanized, and perhaps increased the solidarity we feel when managing our own lives. While writing this blog post, I needed to be careful to not to make it seem like I’ve solved all of my problems with managing a PhD. I’m sure many of the themes and challenges I’ve discussed here will come up again repeatedly in the next few years. In fact, I’m slowly realizing that by writing this blog post, I’m procrastinating and sitting in denial that I have lots of work that I’d like to finish before tomorrow…
