Refining A Research Question: Fill & Download for Free

This is a difficult question. I think every faculty colleague I know does something a little different, because how we approach research is really a byproduct of our personality, our work style, our area of research, and a lot of luck.I will say that it seems like some of my colleagues have it figured out, where they are pursuing a long term goal that is grand in scope and impact, while also clear in focus, allowing them to keep pushing one step at a time towards that goal. Some of my friends in security and others in Operating Systems research fall into this category. Whether you're pursuing a truly secure browser, or a bug-free OS, these are goals that you can chip at one meaningful piece at a time, and there is always more to do.I personally took a long windy road to finding out the backbone of my research agenda. I did programming languages research as an undergrad, switched to OS/systems when I first got to Berkeley, slowly drifted into and became a networking person by the time I left. By the time I started to figure out what my research style was, I had basically tried everything and narrowed things down by process of elimination. I finally bumped into data-driven systems/measurements almost as an accident in 2009, when I started mining data on Facebook and other online systems. That was the first year I submitted a paper to IMC, a conference that I never thought fit me particularly well. Turns out I didn't know myself as well as I thought. I submitted 3 papers that year, and all 3 got in. I recall the exact moment when it finally dawned on me that empirical, data-driven work was my natural style. Having figured that out, and knowing that I am a wanderer by nature, I am content to drift from area to area, working in each for about 3–4 years before finding the next focus that captures my attention, but always with a data-driven perspective, and often with an adversarial/security perspective.My suggestion is to not force it. Trying to intentionally "define" or "develop" a research agenda is very hard, because what you consciously think you want to do may not be what you're best at. I would suggest just working on interesting problems that you want to work on. Once you find a few of those that you're good at, your natural inclinations will probably become more obvious. Ultimately, it takes time and experience to understand your own style and what problems are of interest and tractable within your skill set.Good luck!

I think you have a terrific advisor who is willing to teach you the most difficult part of being a researcher: coming up with ideas that are novel, interesting and tractable.If you're having problems coming up with ideas, the first question I'd ask is how many papers you're reading per week. Ten is adequate, twenty is probably too much. For papers you find particularly compelling, read a few of the citations as well as later works that cite it. Start getting a sense of how ideas are developed over time, and how changing circumstances provide space for new ideas.Second question: what are your tools? If you're working at the OS and architecture level, you should be very familiar with the architecture documentation of your processor. It gets updated a couple of times a year, and going back to see what features have been added is a really fruitful method of generating new ideas. If you're working at the language level, you should be familiar with the language specification documents and how they've evolved. At the network level, what happens when a specification turns into code?Third question: how many research ideas are you generating per day? Keep a (physical) notebook and jot down two or three ideas daily. They can be completely infeasible. They can turn out to be done already. They can be uninteresting. The point is to get into the habit of writing down something. After six months you're going to have a massive pile of terrible ideas with six or eight really cool ideas sticking out at odd angles. You can make a career on two good ideas a year.Fourth question: have you treated this problem as a research question? You're not the first person to have trouble coming up with research ideas, and there's been a lot of research done on the problem. Go read it.Fifth question: have you asked your advisor for advice? How does she come up with ideas? Is she expecting you to come up with general topics that the two of you will refine together? Does she want a formal two-page research proposal? Other than complaining how hard this is, have you discussed this with her at all?Grad school is supposed to be hard. The best students are able to come up with their own research projects. You have an opportunity here to do that. Don't waste it.

When I started grad school, I was very ambitious and wanted to discover all the deepest theory and solve all of the hardest problems.After completely failing at this for many years, I finally learned that trying to directly tackle the hardest problems is usually not a productive strategy. It’s like trying to climb Mount Everest without climbing smaller mountains first to build up your skills, experience, and self-confidence.I am no longer in academia but this is a lesson that’s served me well beyond my PhD: There is no free lunch. Deep theory is usually developed and refined through lots of hard calculations with small, concrete examples. It doesn’t come to you fully formed in some “eureka!” moment.