Research and Adventure
During my PhD, I went hiking alone in a remote region of Iceland. Over the years, I've come to view this trip as analogous to the PhD process. Graduate school was hard, but on the warm days, the views were spectacular.
The summer after my first year of graduate school, I went backpacking alone in the Westfjords of Iceland. I had planned the trip because I was tired of being lost in my research, of living in a dorm room, of doing homework on weekends. Before starting a PhD in computer science, I had worked as a software developer in New York for a few years, and I knew what I had given up to go back to school. I had wanted an intellectual independence that I found difficult to have in my job, but I discovered, unsurprisingly in hindsight, that independence has its own challenges. So after a rough year back at school, I decided to go for a long walk.
I settled on a place called Hornstrandir, a remote peninsula in the northwest corner of Iceland. I don’t remember how I found out about the place, but I was searching online for remote destinations when I found a photo of beautiful basalt cliffs, shrouded in fog and teeming with birds. I learned that the place had a strange name and was a nature preserve that had not had a permanent human settlement since the 1950s. Unlike many popular backpacking destinations, I could find almost nothing about Hornstrandir online. The less I could find out about it, the more I wanted to go.
Six months later, I found myself alone, trudging across muddy trails under a steady drizzle and wondering why I had picked this place. I hadn’t seen another person in hours. My waterproof boots had long-since given way to the onslaught. I was colder and wetter than I had ever been, and I had no way to dry my gear except the sun, which rarely appeared. In every direction, I saw a treeless, gray-green landscape. Why had I come here, I wondered? I could have spent less to go to Paris. I could be on a beach somewhere warm. Why did I pick somewhere cold, wet, and remote?
In the years since, I’ve grown to love hiking and backpacking and on every trip, I inevitably find myself asking a similar question. Why do I spend my free time walking with a loaded pack, often alone, through remote regions. Why am I drawn to these kinds of activities in these kinds of places? As I have pondered this question, I have come to suspect that my answer is the same as my answer to the question, “Why did I go to graduate school?” This post explores this connection in my mind.
Unknown territory
The goal of a PhD is to make a contribution to the body of human knowledge. What this requires is striking out for unknown territory. The first step to finding unknown territory is learning what is known, and most of my PhD was this process. Even in the physical world, unknown territory is often not as neatly defined as a line on a map or an unclimbed mountain. A young climber, for example, might spend years climbing well-known routes while learning which particular sections or ascents are interesting, hard, or unclimbed. Similarly, in graduate school, I spent the first few years building a mental map of my field, discovering for myself what work mattered to me.
This idea that learning is itself an adventure was put dramatically by the mathematician Carl Friedrich Gauss who, in a letter to a fellow mathematician, wrote:
It is not knowledge, but the act of learning, not possession but the act of getting there, which grants the greatest enjoyment. When I have clarified and exhausted a subject, then I turn away from it, in order to go into darkness again.
Going into unknown territory—”darkness,” as Gauss puts it—is hard, and researchers organize themselves in different ways, depending on their wilderness. A team of researchers, for example, may set up a basecamp at the foot of a large mountain and spend years in the assault—think of big science projects like the Large Hadron Collider. Other researchers wander through their wild places, collaborating freely with whoever passes through—think of the prolific mathematician Paul Erdős. And sometimes, rarely, an ambitious researcher, often young and with little appreciation for what is considered hard or impossible, simply does a remarkable feat—think of Albert Einstein, who as a 26-year-old patent office clerk, published four physics papers that changed the foundations of modern physics. This kind of work is valorized for its audacity, and as soon as a new route is discovered, up the mountain or across the wilderness, a community of researchers follows. Years later, what once seemed an impossible route becomes a well-worn path.
So this is the first part of my answer to the question. I pursued a PhD because I wanted to understand my wilderness, computer science broadly and machine learning specifically, well enough to make my own discoveries. And this mindset is apparent in my hobbies. A big reason I love backpacking over other types of travel, for example, is that it allows me to experience the physical world on my own terms. There are no lines or entrance fees, no tour buses or guides. Backpacking is all about doing things on your own. The biggest task is that you must move under your own power. And then there are the everyday tasks: building a shelter, making a fire, tracking the weather. I like this kind of work. There’s a sense of immediacy that can be missing in mental labor.
Similarly, before going back to school, I had found myself always asking questions related to my work. How does a search engine return results so quickly? How does code that I type into a text editor get translated into something usable by a machine? How does a neural network classify images? Hearing explanations was not enough. That was like being told how a car works without touching it. I spent many nights and weekends teaching myself the basics of computer science in an effort to understand the machinery I was using. (I had studied architecture in college.) I wanted to understand ideas deeply enough to build my own tools, and a PhD seemed like a good way to spend a few years doing just that.
Solitude and clarity
A possible side effect of independence is isolation, because wanting to do hard things yourself often means that few people care to join. In both graduate school and hiking, I have felt far from civilization. I can clearly remember the feeling of talking to someone at a party in Brooklyn and realizing I could not explain my research to them, not in any meaningful sense.
This isolation is often necessary, I suspect, since being alone is perhaps the best way to decide for yourself how to approach a problem. With too many people around, it is too easy to accept the thinking of others. The climber Lynn Hill once said, “To be the first at something, I think you have to have a kind of confidence and a vision that’s objective and not based on what other people think.”
During my PhD, I learned that to make progress on a hard problem, I had to ignore popularity and hype, to not get sidetracked by what others were doing. For example, I once wrote in an early paper draft that a method in machine learning was “popular.” I was trying to motivate why we were building on this prior work. After reading the draft, a senior researcher said to me, “In research, popularity is not a reason to do something.”
This aesthetic is not so different from how the climber Marc-André Leclerc described his approach to alpinism:
As a young climber it is undeniable that I have been manipulated by the media and popular culture and that some of my own climbs have been subconsciously shaped through what the world perceives to be important in terms of sport. Through time spent in the mountains, away from the crowds, away from the stopwatch and the grades and all the lists of records I’ve been slowly able to pick apart what is important to me and discard things that are not.
You do not have to be a researcher or a climber to be subconsciously shaped by others. A few years ago, I hiked the Presidential Traverse in New Hampshire with a couple friends. I remember arriving at the top of Mount Washington and seeing two dozen people in a queue to take a photograph at a nondescript plaque marking the summit. The sun was shining and the air was clean and in every direction we could see the beautiful White Mountains, yet people were standing in a line, looking at their phones, and waiting.
My point is not that isolation is good or that popularity is bad, per se. And my point is also not that one should close oneself off from others or from new ideas. In my experience, the best researchers have lots of problems they care about and are able to integrate new information quickly. My point is, however, that if you want to do new things, then doing things because others do them is a bad process. You must convince yourself that what you are doing is important.
Context
Doing things yourself has limits, of course. While in Hornstrandir, I read Halldór Laxness’s Icelandic epic, Independent People. The novel is about an Icelandic farmer, Bjartur of Summerhouses, who idolizes an independent life to the point of suffering and isolation. He believes that a man should always decide for himself, should never eat the bread of others, and should not even mourn the loss of a child. Throughout the book, Bjartur damages his relationships because he overvalues independence. This independence is an illusion, of course. Even in Hornstrandir, in as remote a location as I have ever been, I was following cairns maintained by others.
If real independence is impossible, what’s the point of trying things on your own? Beau Miles, an educator and outdoorsman, put it well when talking about why he was walking ninety kilometers to work instead of commuting by car:
It’s about putting value on such a thing, much like baking your own bread or taking karate lessons. I think that experiences like this are the essence of being human, which to me is our ability to question everything we do.
For me, a PhD has put a value on human knowledge. Just spending years getting to unknown territory drove home how little I could possibly know. It has made me appreciate experts in new ways and clarified how hard won even the most basic advances have been.
This experience reshaped how I view knowledge. I used to think of knowledge as like a palace, built carefully from the foundations. During my PhD, I realized how little of the world is mapped, and I started viewing knowledge like cairns in a foggy mountain pass. When you’re at the edge of knowledge, you realize that multiple papers give the same incomplete and canned explanation. You realize that a beautiful result in your field was discovered two decades prior in another field, just under a different name. You realize that just as we can gain understanding, we can lose it. Knowledge is not a palace; it is a trail of signal fires in the darkness. What this suggests is that, much like physical trails, these connections must be maintained.
A really good day out
The final part of my answer is that often a great view requires work.
When I bought my plane tickets for Hornstrandir, I was unprepared. As a kid, I had hiked a few mountains and car camped with my family. As an adult, I didn’t even own a sleeping bag or tent. Hornstrandir was wild and remote, a peninsula only in name, cutoff from the mainland by the Drangajökull glacier. To get there, I would fly to Reykjavik; transfer airports; take a prop plane to Ísafjörður; and then take a boat. While there, I would have no cell phone reception. To arrive unprepared, even in the summer, was foolish at best and deadly at worst.
I spent countless hours preparing. I researched tents designed for exposed, high-wind environments and practiced setting mine up in a park in Brooklyn. I took a class on how to use a compass and read a map. I scrambled up nearby mountains with a weighted pack. I built a spreadsheet to calculate how many calories to eat at each meal. I borrowed a friend’s safety beacon and bought travel insurance for emergency medical evacuation and for the repatriation of my remains. In absolute terms, most of the adventure was preparation. Even the trip was mostly work: slogging up a hill in wet boots, drying gear when the sun was shining, setting up camp and cooking a meal in the civil twilight.
Was all that time and effort worth it? Let me tell you about Hornbjarg. Hornbjarg is a headland that juts into the North Atlantic. The sea cliffs are spectacular, roughly twice the height of the White Cliffs of Dover. Thousands of birds live there: Arctic terns, puffins, and Iceland gulls. Standing on Hornbjarg, you are only five miles from the Arctic Circle. I visited Hornbjarg on the second-to-last day of my trip, and I was lucky. After a week of cloudy skies and intermittent rain, the weather broke and the sun came out.
That morning, I was followed by an Arctic fox as I walked across the dark mudflats of Hornvík Bay. I saw three fox cubs wrestling on the rocky shore. I watched a bird dive from the cliffs, joining hundreds of others, becoming another white dot on a field of dark blue. I walked through fields of yellow and purple flowers that swelled into Miðfell, a jagged mountain that splits Hornbjarg into two halves. I walked at the feet of Kálfatindar, a mountain that rises up like a knife edge. Even in July, the spires of Kálfatindar were garlanded with snow.
Remarkably, I was alone the entire day. I felt as if a security guard had opened a museum to just me, before the crowds and tourists arrived. Only once did I see other people. After hiking over Miðfell, I looked back and saw two hikers I had met at the campsite trailing in the distance. As they descended, a low cloud rolled over the mountain top.
Mental mountains
In his essay on training, Nose to the Wind, the alpine climber Steve House wrote
I like to say that the only people who are not enthusiastic about training are those who haven’t trained correctly. The only people who hate lunges are those who’ve never done enough of them to experience what strong legs feel like on the trail. The only ones who hate long, slow aerobic capacity–building runs and skis are those who have never known what it feels like to sail up the mountain, nose to the wind, with ease. Relaxed, poised, moving fast and flying.
I feel this way about physical training, but I also feel this way about graduate school. For me, graduate school, like Hornstrandir, was mostly training and mundane tasks, but this mundanity was been interspersed with moments of wonder. A good idea, properly understood in context, is like a view from a mental mountain. To get there is not much different than physical mountains: preparation and training, risk-taking, a willingness to do something hard and maybe pointless while your friends do something else. Yet on good days, the sun comes out, and you feel the value of training in your lungs and legs. You wake early, hike quickly, and take in rare views.
Maybe you’ve read this far because you’ve been asking yourself, “Should I get a PhD?” Honestly, there are a hundred reasons you shouldn’t, and you can find plenty of people who will supply you with the list. However, if you’ve heard those reasons and are still asking the question, then I think the most important psychological question is: are you compelled to understand and explore? Would you rather sit uncomfortably with confusion than to pretend the orthodox answer makes sense? Are you willing to do something lonely and hard to settle that confusion? My point is not to glamorize the hardships of graduate school, which are real, unevenly distributed, and often have nothing to do with the ideals of academia. All I can say is that, to me, the challenges were worth the views.
Why am I like this? I am not sure. But more directly than one might think, the years I have spent studying computer science, mathematics, and machine learning are directly related to a desire to set up my own tent in a remote corner of the world. Both are driven by a curiosity to understand things on my own terms, to convince myself, to discover.
Company
The day after visiting Hornbjarg, I woke up to the wet world of Hornstrandir. I had spent a sleepless night as strong winds blew across the campsite, breaking tents and lifting stakes out of the ground. Gone was the Arctic fox, and gone was the sun. Again, the world was cold and wet. It was my last day on Hornstrandir. Two hikers who had trailed me on Hornbjarg, Max and Katharina, were taking the same ferry as me, and Max had found a fourth person, Gui, who was heading our way as well. Gui was a Belgian man in his sixties, who said he had done a solo trip every year for the past thirty years. I liked him immediately. We all needed to be in a place called Veiðileysufjörður by late afternoon and decided to hike together.
It rained the entire day, and there was plenty of snow and fog in the mountain passes, but after a week of hiking alone, it was good to have company. The day passed quickly, and in the afternoon, we arrived early at Veiðileysufjörður. It was still raining, and I set up my tarp tent for the group. Sheltering from the wind and the rain, we boiled water for coffee and shared the last of our food, carefree with the knowledge that we had no longer had to ration. Shivering in the summer afternoon, we waited for the boat that would drop off a new batch of hikers and take us back to the mainland.