Yesterday was the defense of my [ research prospectus ] . It was a two hour conversation with my committee, Ian Utting and Peter Linington. It was a spectacular demonstration of how not to manage your PhD process.

This post is about my failure to manage the PhD process adequately over the past three weeks (three months?). In subsequent posts I’ll dive into the content of the conversation, which is much more important to me. My mistakes should be of interest to PhD students who need models to avoid. The content of the conversation (and how my own confusion manifested itself therein) is primarily of interest to me, although other students may benefit from these reflections as well.

Mistake #1
The first mistake goes back to last May; I got too involved in the Bootstrapping CS Education project. Considering the timing, I shouldn’t have been hacking code to mine data for the project. Please, don’t get me wrong: It was a very rewarding experience, I met lots of great people, and am very glad I did it. That said, it was time I should have been spending on reading literature and exploring ideas related to my new thesis direction.

I followed that up with another mistake: I went to the States for… 6 weeks? During that time, I spent no more than 5 days in any one place. While it was wonderful to see all our friends and family, I got no work done (to speak of). Furthermore, I realized (in hindsight) that I had no downtime. It came to the point where I had to schedule a morning of golf with my father, because I was so over-committed in going to weddings, visiting people, etc. There wasn’t time to just breathe. (Our next to the States, while potentially a long one, will be spent at home, working. Anyone wishing to see us during this time may drive/fly/walk to Ohio.)

We returned to the UK from the States, and immediately spent two weeks traveling with Carrie’s family: up to Edinburgh, into London, and back down to the Southeast of England. Again, this was wonderful, but it wasn’t in any way relaxing. And, again, it was time spent not thinking about work.

So two months go by with no work, and furthermore, they aren’t restful. When Carrie and I got back, I was a wreck. I spent the month of August useless. No work. I was, simply put, depressed. This was a new experience for me. I coped by trying to remain interested/excited about anything. (This led to a series of posts about XML-RPC and Applescript on my weblog, for what it’s worth.)

Part of this was because the 30+ articles I asked Carrie to bring to the States for me were completely inappropriate for the new research direction. They were all cognitive in nature, and provided no insights into compilation behavior. I came back having no literature to support my new research direction, and the prospectus was supposed to be submitted back in July.

The month of September got better; I found my relevant literature, dug into it, and fortunately it was neither wide nor deep–this is a novel area for research, and there isn’t a large body of research directly related to the study of novice compilation behavior (or compilation behavior, or even the behavior of programmers). But this lead me into mistake number two.

Mistake #2
The second mistake was not spending enough time asking how and why, and discussing that with my supervisor. By this, I mean that we had a working hypothesis: compilation behavior is related to programming effectiveness (and no, for the moment, I won’t define or defend any of that… I mean for that statement to be, at this exact moment, to be over-broad and sloppy). We had a proposed mechanism for collecting interesting data, and some ways to go about analyzing that data.

In writing up the prospectus, I got caught up more than once on distractions, and those were (at some level) related to the nature of the data being collected and how we were going to analyze that data. It boiled down (for me) to questions of research methodology.

We had worked out what I was going to do, by-and-large, and I proceeded to start worrying about the approach I would take to get this work done, instead of focusing on the goals of the work and why I was doing it. As a result, I become concerned with the methodological approach I would be employing in my research. So, I started reading as much stuff about radical behaviorism as I could, focusing primarily on the works of B.F. Skinner. In particular, The Selection of Behaviour: The Operant Behaviour of B. F. Skinner: Comments and Consequences, edited by Catania and Harnad, and About Behaviorism, by B.F. Skinner. (Selection of Behavior is great because it takes a handful of papers, critics respond to them, and you get Skinner’s replies to each of those critiques. Fascinating reading.)

This is what I mean by not spending enough time on the how and why of my thesis. When I became lost/confused, I reached out for a methodology to ground myself in, without fully working out the question. This is like saying “I’m going to use a hammer!,” and then going in search of a problem. Yes, I can probably use the hammer for a lot of different tasks, but I might be better served by, say, a screwdriver.

The true breakdown of process occurred when I failed to discuss this with my supervisor before my defense. As a result, I walked into the defense, and laid out a (new) research approach, one that wasn’t reflected in the prospectus. I didn’t exactly throw away everything I had done previously, but I sure came close.

Why did this happen? I became confused as to what I was doing, searched for a tool I could cling to. After wrapping my head around the new tool, I looked back at what I had said I would do, and… well, I couldn’t use my new tool to accomplish that task–there was a mismatch. So, there was (to my feeble mind) only one thing to do: come up with something that fit the tool, and try and retain as much as possible of what I had done previously.

I did just this, proceeded to present a new research direction to my committee, and then try and explain what I had done to my supervisor… she wasn’t pleased. And, rightly so.

The next few posts will explore what the document proposed, what I presented to my committee, and where I need to land with these two proposals in mind, by going back to the goals of the research and what I hope to accomplish.

What are the lessons learned?

• Talk to your supervisor when you’re confused.
• Focus on what you want to accomplish.
• Keep asking how you’re going to get to your goal, and then ask why that’s a good way to do it.
• The how/why sequence will lead you to an appropriate methodology. Choose methods appropriate to the task. Do not let the method drive the research; the question/hypothesis drives the project, not the tool.
• Did I say talk to your supervisor?

Awful as this sounds, all is not lost for our expatriated protagonist! Tune in next time, when Matthew reveals how a tub of Chocolate Fudge Brownie, a glass of cheap port, and sleeping in the next day led to the revelation that he had made a critical error, and surely his perspective alone wasn’t enough to see what was the correct, next step!