Today is the Sensys’09 programming committee meeting, going on just down the street in our spaceship home, Maxwell-Dworin. So fairly soon we’ll hear about the fate of two papers our group sent in, one on system called Mercury — a paper I was peripherally involved with — and another on a system called Draft, which I lead authored along with my advisor and the fantastic Jason Waterman.
The review process is always fraught with tension, particularly around decision time, but I’m finding myself struggling a bit harder that usual to wait quietly by the internets for a verdict. I think a combination of various factors has led me to feel more invested that usual. Enumeration follows the break…
First, the Draft paper is really the culmination of a line of thought and development that began during the unsuccessful first and second attempts (Sensys’07 and NSDI’08) to publish what eventually became the Lance paper published at Sensys’08. The winter following the Sensys’07 rejection and resubmission was hard for me. Due to a string of good luck and good timing and with a lot of help from others I hadn’t had a paper rejected from a conference previously. The process of writing the Sensys’07 paper was as difficult as any I had been through, with several concurrent submissions from my group leaving me pushing that stone uphill largely on my own. And although I tried to be positive about the outcome (which found me walking around the Capitol Hill neighborhood of Seattle on a summer stint with MSR that was difficult for other reasons[1. Funny but I have entirely forgotten where I was at the time of notification for the next two big deadlines.]), the reviews were fairly negative and I didn’t get the feeling that the paper ever had a real shot. Strengthened by our 2007 deployment at Tungurahua the paper faired better review-wise at NSDI’08, but still missed the cut. But at that point I was really back at the drawing board.
Those series of rejections really led me to question the process I had been going through in generating research, which resulted in a period of multiple-months during which I read a great deal of research papers and spent at least an hour (frequently more) writing in a research journal daily. The speculations that resulted paved the way for the changes that eventually strengthened Lance for a Sensys’08 appearance, charted the course that led our Draft paper under submission, but most importantly convinced me that I was capable of doing original research and started me thinking about the academic job market in a serious way. I’ve been telling people lately it feels like there are three stages to the career of a successful computer science researcher[2. I won't try and generalize any further.]:
- You learn how to do research, by which I mean testing ideas, participating in development, running experiments and writing papers. The ideas, however, are not really yours. This is the stage that most early-career Ph.D. students find themselves in during their first couple of years (supervised my mentors and more senior graduate students) and it’s really the only stage at which you can fail Ph.D.-wise in many programs.
- You learn how to produce research, by which I mean a mixture of developing ideas to their logical conclusions and inventing new ways of thinking about things. Ph.D. candidates probably learn those things in that order, with incremental work being naturally easier to pick up than the fresh broad strokes that signal something truly new. The challenging thing is that making it to this stage requires genuine creative development, which is hard to teach. (I’d like to write more about this at some point.)
- You learn how to transfer your research intuition into another sub-field. This stage is something that, it seems to me, a lot of potential faculty members don’t really get a chance to try until post-tenure — with switching areas before tenure probably seen as fairly risky — unless they take advantage of a gap year between finishing their degrees and starting as faculty to do so[3. As my advisor did.] Not having got here yet I don’t pretend to know what this is like, but it seems to me hard to imagine a 30 or 40 year career in research without doing this at least once or twice.
So to end this fairly elliptical set of observations, that series of rejections set up a process during which I was able to convince myself that I could get over hurdle #2, at which point a whole host of interesting possibilities — including faculty positions — started to look less frightening and more feasible. So I’m a bit partial to the outcome of those brainstorming sessions, currently being tossed over by a bunch of colleagues in a building not far away.
Second, this is a paper that I think makes a substantial algorithmic contribution, which is not always the case with systems papers. We have a lot of other things to work on in our research purview, things like system architectures, frameworks, aesthetics, etc., but given the history of computer science being firmly rooted in algorithms and their development I like the conceit of thinking that I contributed in some small way towards finding a new way of doing things. Obviously we’ll see what others think.
Finally, I just couldn’t have been much happier with the process of getting this paper together and out the door. Was it hard, fast, stressful at many moments? Yes. Did I sleep on my sofa at least once? Yes, many more times than once. However, whereas the Sensys’07 deadline push felt like a death march this one felt like a good day in the mountains: stressful, but exciting and fairly (if I do say so myself) well-planned. Matt gave me a lot of rope, and I loved working with Jason, who was brilliant all the way in, fantastic as we talked over the scope, shape and contributions of the paper, and absolutely instrumental in the development and testing of the system. It’s natural for things to get loud angry and tense towards the final moments of a big deadline, with everyone overtired and feeling the stress, but there was none of that this time and that made it, for me, a huge pleasure.
So does a good way produce a good thing? I don’t know yet, but writing about this now as the tensions rise has relaxed things a bit more me, since I’m remembering that I’m extremely proud of this paper. I do hope we get a chance to talk more about it at Berkeley this fall, but I’ve learned so much in the process of doing this that, one way or another I know we’ll be back.
With that said I’m going to go out into a glorious day, probably spin down to Boston Common where my wife is waiting for lunch. Notification will find me I’m sure, and no matter the outcome I’m going to want to be close to her and close to a beer.