January 3, 2022
I'm at an exploration phase where I'm deciding what problem I want to spend the next few years of my life working on. My background is a mix of software, engineering, and biology, so you might think the natural place for me to work would be at the intersection of robotics/software/biology. But if you choose your future work solely based on your past experiences, you can suffer from heuristic and sunk-cost bias.
Recently, I’ve taken a step back from my past work so I could explore a broader set of problems. I've found that this open-ended exploration is very different from my previous explorations. For example, when I was building my real estate start-up, the goal of the exploration was to identify a hair-on-fire pain point in the market. This type of work was low in execution-risk, but high in market risk. When I was building algorithms to improve cancer treatment at Princess Margaret Hospital, the problem was already defined (i.e. deliver radiation doses more precisely), and I was exploring a set of solutions. Jumping back to the present day, instead of exploring solutions, I am exploring problems worth pursuing. My goal is to narrow down a high-impact problem amongst the million things you can work on in the world. The problem can be part of a field I'm already familiar with, or a new field where I’ll be bringing my transferable knowledge.
In performing this exploration, I have developed a set of personal principles to help me explore new domains. My hope is similar souls might find these principles useful too on their own journeys towards meaningful work.
Principle 1: I’m patient with myself.
Exploring a new domain is hard. As an impatient person, I’ve had to cultivate patience with myself.
I frequently struggle mentally with things that I have yet to understand. Getting a bird’s-eye-view of a field might require scanning through research papers that were hard to understand to outsiders. Often, papers lack the scientific context necessary to incorporate them into my previous understanding, so I end up having to contact scientists to ask follow-up questions.
When you are struggling, it’s reassuring to realize that in that moment, you're doing the hardest thing you’ll ever be doing on your intellectual journey to understanding something. Instead of feeling frustrated, I switch my brain to be excited: How amazing is it that I don't already know everything that exists out there? I'm on the steepest slope of the learning curve, where I'm experiencing the highest level of growth.
I have also come to appreciate that certain fields (e.g. chemistry, biology) require more patience than others. In software, you can launch a feature and gather customer feedback in 1-2 days. But you do not get the same privilege in experimental fields like chemistry or biology.
Also, in fields that require physical engineering, you literally need to be patient while things happen. Let's use a simple example: Say you want to design a guide RNA to knock out the GFP gene in E. coli. At the comfort of your home, you run a convolutional neural network that predicts a guanine and cytosine content of greater than 80% would lead to off-target effects, and having a lot of T bases in the middle of your gRNA would increase binding affinity with the target protein. You jump on your bed screaming: "Wow, I did it! I discovered something new." But did you?
Any prediction or simulation serves as a good start, but is not trustworthy until you get hard data from experiments. You now have to find a lab space, wait for a bacteria culture to grow, send your sample for DNA sequencing, and wait for results to come back. 90% of the time things fail, and you have to restart.
By being patient, you may temporarily move slower in the short-term. But it's worth it because rushing things comes at a great cost: you risk giving up, take mental shortcuts, and end up working on something at the local maxima.
Principle 2: I don’t use difficulty as an excuse.
This might sound contradictory to the point above. But as soon as you understand the difficulty of something (so you can give yourself patience), it becomes much easier for yourself to make excuses for going slow, or not finishing projects. I have found this to be especially true with projects that involve physical constraints.
Here's a non-exhaustive list of common reasons, and how to mentally disprove them:
|I don't have enough experience.||Firstly, assuming you die at the average of 80, you likely still have multiple decades to dive into the field, if you care enough about it. That is a lot of time for skills and knowledge acquisition. Secondly, you don't need to become the best in the field to solve a specific problem. You also don't have to do everything yourself. Milan Cvitkovic wrote an article on things you are allowed to do but didn't know you could.. Thirdly, many difficult problems can have suprisingly elegant, simple solutions.|
|Working on engineering projects is capital-intensive.||You can start by building a Proof-of-concept prototype. Say you want to convert food waste into clothing materials, like polyester. Your first step might be drafting a series of chemical reactions and conducting an experiment in your kitchen. Running computer simulations also serves as a useful starting point. Once you have something concrete, it becomes easier to reach out to people and ask to borrow their lab space. My friend Brianna serves as a cool example: she built a fusion reactor in her basement and then, a couple months later, got the opportunity to work at General Fusion leveraging the knowledge she built. She was also funded by Emergent Ventures for her prototype.|
(Don't get me wrong. I do think these are, to a degree, valid reasons. They are not "bad" per se. But if you want to do something unconventional and you're not affiliated with any institutions, you have no choice but to build mental fortitude and figure out how to not let these common reasons be your limiting factors.)
Principle 3: I’ve stopped feeling guilty about learning.
After sampling my friends, I was surprised to find that feeling guilty about learning is a common feeling while exploring. Sometimes, there’s a sneaky voice in my head, saying: "I'm learning Stokes theorem? What the hell am I doing? Shouldn't I be doing something more impactful?"
Learning fundamentals and pursuing your curiosity can feel like self-indulgence. If you’re immersed in start-up culture, everyone around you is telling you to move fast and take shortcuts. If everyone around you is executing, learning can feel like a waste of time: you’re absorbing someone's knowledge without doing something immediately useful with it.
It's helpful to remind yourself that obsessively learning the fundamentals when you dive head-on into a new field can be extremely high-leverage in the long-term. It offers you intellectual independence, which means you have a better chance of seeing flaws in others' logic and challenging popular beliefs.
Another observation is there seems to be a correlation between increased age and increased guilt. It's easier to commit to learning when you're younger. My hypothesis is that this comes down to societal expectations: you see your now adult friends are doing tangible work, and you are... learning. There seem to be two solutions to this:
More on rewiring: my approach has been to reframe “learning” in my mind’s eye. Instead of just “learning,” I’m doing a job just like anyone else who’s working at a company. The learner’s job is to seek understanding and ask questions with the hope of discovering something novel. This is just as rigorous and as valuable work as people with traditional jobs.
Principle 4: Commit to understanding, not an outcome.
Sometimes when I tell myself to "discover something new", an ingenuous insight feels so foreign and far away. Asking myself "What's the next paradigm shift in computational biology?" feels big and vague. I’d get intimidated and paralyzed before I even begin.
I find it much easier to ask myself well-scoped questions: What is a cell? Can you simulate a single one in a computer? What about 3 trillion cells in the human body? What computational tools exist out there to predict how different proteins, DNA, RNAs interact with each other? Do they work well, and why or why not? Are there certain pathways that are more difficult to predict than others? I've found these basic questions are often ones that lead one to surprising insights.
(Isaac Newton also asks himself a series of deceptively silly questions like "Why doesn't the Moon crash into the Earth?" to illuminate his understanding of gravity.)
While foreseeing an outcome is hard and nothing more than a conjecture, it is simpler to allow yourself to be a 5-year-old kid again and ask questions as your way to construct your model of the world.
Another limiting factor to asking basic questions seems to be ego. I used to fall victim to this, where I would get scared of asking fundamental definition questions in fear of being judged. At one point, I attached my self-worth with whether my experiments work, if my algorithm performs the best, etc.
In retrospect, this is incredibly silly: Wanting something to be true for yourself doesn't mean that it is. Nature doesn't care what you want. Your job is to find some predetermined truth that already exists. Once I've detached myself from ego, I find myself more productive at asking "dumb" questions and finding the answers to them.
Principle 5: I add structure to my exploration.
I often find that I have the highest learning rate when I'm time-constraint, emotionally invested in the problem, and have a reliable way to measure my understanding. Here are some examples of goals that I would set when I started diving into a new (sub)field:
I aim to write a landscape overview of state-of-the-art research and companies that exist out there. I was inspired to do this after reading The Current State of Machine Intelligence by Shivon Zillis.
I develop my own list of questions instead of blindly following a curriculum. For example, If you are into drug discovery, you can see Celine Halioua's example modality questions here. Personally, I make it a goal to understand the historical picture of technological developments and how translatable they are to solve a specific problem.
I build a repository of key insights and their supporting data.
I develop particular hypotheses or sets of questions and reach out to N experts. This is a forcing function for me to make sure I have a foundational knowledge of a topic, at least enough to be able to converse and think critically about what's said by the other person. It also helps me figure out the unknown unknowns, as when you go deep on a particular idea, you end up dredging up interesting tangential details.
I try to avoid idolizing an expert, no matter how accomplished they are. Being starstruck is a surefire way to waste a valuable conversation. Take as long as you need during conversations to pause and think about what's being said. I found this awkward at first, but effective when aiming to nurture and develop original thoughts. The best experts often feel like they are not experts and aim to engage in an active discussion with you.
I reflect weekly on how much tangible progress I’ve made towards my goal.
What new insights have I learned? How does what I’ve learned change my understanding and execution plan going forward? If I learned less than expected, how come?
People often talk about having empathy for customers and reiterating on a solution, but no one talks enough about reiterating on oneself. Being self-aware and honest about the progress you've made (or lack thereof) is just as important.
Structure and accountability helps me stay focused. For example, I’ve thrown around the idea of writing a Typeform once and automatically emailing it to myself at 8AM every Sunday. You can also find a close friend at a similar stage like you and have weekly accountability calls.
Principle 6: I work with thought partners.
When I was working on my previous start-up, I got to work with incredible sounding boards through the Interact Fellowship, such as Sarah Nahm and Michael Akillian. They asked really great questions which challenged my hypotheses, and they called out implicit assumptions in my thinking that were invisible to me.
When I was learning the concept of a tensor, my friend Adithya and I started brainstorming ways where we can visualize an N-dimensional matrix. One way we came up with was to view it recursively: Each element of the matrix is another matrix. Although not completely rigorous, it serves as a useful mental trick to push forward to the next sentence in the text. During our Internet rabbit hole, we also discovered an interesting thread on Math Exchange.
In both cases, without a third eye, I could have easily wasted time and resources on something not fruitful.
So how do you find a thought partner? Here are a couple ways that seem to have worked for me and others:
I used to struggle with the concept of "exploring" or "soul searching" - It’s hard to resist the urge to focus on one thing. But open-ended exploring is incredibly high-leverage, and almost no one does it.
Without cross-domain exploration, you’ll never switch to focus areas due to a fear of "being behind". You can spend years working on the wrong thing only to regret it in the end.
My genuine belief is that with the right ambition and discipline, you and I can pretty much learn anything at an extremely fast rate and have enough knowledge to translate a technical insight to something useful for the world. Hopefully my experience is useful for you when you attempt to do this.
Thanks to Jay Parthasarthy, Raffi Hotter, Shagun Maheshwari, and Adithya Chakravarthy for editing this article.