A very little key will open a very heavy door.
― Charles Dickens
I was a third-year PhD student scrolling through the online publication of a systems conference, looking for papers to read. I was stopped in my tracks by the title of a paper that sounded exactly like my PhD thesis topic. Anxious but curious, I raced through the paper.
It was a published paper from a PhD student in Europe. It wasn’t just the title. He had solved the research problem I was going after. The work was complete. Even brilliant, elegant. I was a fan. He had done and said everything that I wanted to say and do. Digging around, I found more bad news. He had already finished his PhD thesis. This was the thesis that I had dreamt about and that I had wanted to write. He and I had been working on the same problem independently. He got there first.
He had scooped me.
I thought that my PhD was finished. It didn’t matter that I had published papers at that point. I had no PhD problem left. I left the lab, slunk home, crawled into bed, and stayed there for the rest of the day. I was numb, and in a state of shock. I didn’t dare tell anyone that day. Three years of work down the drain, and someone else had beaten me to it.
I told myself it was my fault. I hadn’t been fast enough. I hadn’t researched the literature enough (he was out there, doing similar work, and I hadn’t even known). I hadn’t done my homework enough. I hadn’t worked hard enough. I wasn’t good enough.
That weekend, I decided that I should just quit my PhD. I could deal with throwing three years of my PhD away, but I couldn’t stomach another three years of trying to find a new PhD problem, only to find myself scooped again.
Image by Arek Socha from Pixabay.
I went to see my advisors on Monday morning, to tell them my PhD was a hopeless quest. I started blabbering about working on a new PhD topic, about how I would read all the literature again (and better this time) and how I needed to start afresh. After my rambling, they asked one simple question.
“So, what did he decide to leave unfinished in his thesis?”
Unfinished? The guy had finished his PhD. He had filed his PhD thesis. What else was there to finish?
“No, what do you think he would still do, if he had 4 more years?”
“Research is never finished. A good PhD solves a problem. A great PhD solves a problem and finds 10 new ones. If his work is as great as you say it is, he must have left unsolved problems.”
I went back and re-read the paper and his thesis multiple times over. I read with purpose this time. A different purpose. I wasn’t looking to critique his work. I was looking for the open doors. I was looking to see what he had decided not to solve. I looked for the seeds of things he had decided not to, approaches he had not chosen to take, problems he had chosen not to solve, assumptions he had chosen to make. I was looking for the choices behind his solution, instead of fixating on the solution itself (which is what I had done at first).
You see, I didn’t realize, then, that a PhD thesis didn’t solve things up with a pretty bow. I thought that a PhD was a finished piece of work, a perfect thing, all loose ends tied, all experiments done, every observation made, every graph drawn, every conclusion written. Nothing left to do. Just a perfectly finished piece of work.
The best PhD theses leave doors open for other PhD theses.
A week later, I identified the choices his thesis had made. Looking at his choices, I was able to make different choices, leading to a different approach, different experiments, different observations, and different conclusions. By reframing the choices underlying his PhD, I was able to make more informed choices for my PhD. His PhD thesis, which had thrown me into despair, had closed a door but opened so many more at the same time. I opened a different door. I ended up solving a different, but related, research problem.
I devoured other PhD theses in my field, to understand what doors they left open. I was blown away by the sheer number of open doors in the greatest PhD theses.
This experience made me a better researcher because I was not fanatical about the outcome of my research. I realized that the outcome would be imperfect, and that a PhD thesis was not about closure. It was about opening doors, and being unafraid to open them. This made me more keen on the process, the assumptions, the methods, and even the failures. I was keen to tout my failures because they were just doors that I felt I should caution other PhD students about.
The beauty of spending the first three years of my PhD solving one problem was that it took far less time to solve a different one. Within six months, I had opened a new door successfully and was racing towards the finish, towards my own PhD thesis. The door that had closed on me made it easier for me to spot an open door and figure out how to open it.
In turn, I made sure to write in several open doors in my own PhD thesis. And, of course, I made sure to describe the door that closed. I was proud of it. It made me. It made my thesis.
How this changed my research writing
Since this experience, I tend not to bury our research’s open doors in a “Future Work” section when my students and I write our papers. I don’t want it to be glossed over or make it difficult to find. I had rather be upfront.
I am fanatical about writing these four sections as early as possible in our research papers.
- Assumptions. All of the constraints that we are solving the problem under, so that the reader can understand the context for the solution. This section answers the question: Under which conditions will our research work? Under which conditions will our research not work?
- Goals. The precise problems we are solving. Zero embellishment. Zero claims to having a generic solution. Zero claims of broader impact that we are not entitled to. This section answers the question: What is the specific, narrow problem that we are focused on solving?
- Non-Goals. The problems we are not solving. This answers the question: What is out of scope in our research? What does our research leave unresolved or open-ended?
- System Model. The applicability of our solution. This answers the question: Which kinds of applications and systems can use our research?What kinds of applications and systems cannot use our research?
I don’t view these sections as limiting the impact of our work, constraining our thinking, or narrowing the focus of our research. I view them as necessary and clarifying. I also view them as our being honest.
I also view them as a service to PhD students out there who might think — as I did — that their PhDs are hopeless after reading our publication. I absolutely want those up-and-coming PhD students to know how awesome my PhD students are, and how awesome their work is. But, I also want those readers to know what we didn’t do, what we chose not to do, and where’s room at the table for all of us.
The best PhDs solve one problem, and create new ones for others to solve.
The best PhDs open doors for more PhDs.