by Xinyi Zhou '10
Steve Chien’s (’93) job is to tweak the Google search algorithm you use daily to make the results more relevant. Previously, he worked at Microsoft Research Silicon Valley and focused on algebraic computation, game theory and algorithms for large data sets and search. Dr. Chien chats with us about his Magnet experience, his job and gives aspiring computer science researchers tips for landing a similar position.
Q. Tell us about your Magnet experience. What clubs and electives did you take? What teachers were memorable?
A. From what I remember, I don't think I could have been a more stereotypical Magnet nerd if I had tried. (Not that there's anything wrong with that.) I was on the math team, physics team, chess team, computer science team, and It's Academic team, and was also socially awkward beyond belief. One of my most formative experiences was being on the Takoma Park math team in 7th and 8th grade, and I'll always appreciate what our coach, Ms. Counihan, did for me. I thought I'd try giving some of that back this last year [by coaching a local middle school math team], and working with these very talented and dedicated students was inspirational and rewarding for me.
I took most of the Magnet electives that were offered-- I remember in particular things like Statistics, Linear Algebra, Optics, and Origins of Science. I remember a lot of the teachers with respect, including (but definitely not restricted to) Ms. Verona and Ms. Ragan in computer science, Ms. Dyas and Ms. Escatell in math, Ms. Steinkraus (who later became the Magnet director) and Mr. Donaldson in science, and Dr. Alperin, a NIST scientist who volunteered his time trying to teach us university-level physics out of the Feynman lectures. There were a lot of other good teachers, too. I remember when I got to college that Blair had really prepared me well-- a lot of my fellow classmates hadn't seen some of the things that we took for granted in our classes.
Q. Did you do a Senior Research Project that got you started in research?
A. I did a project looking into winning solutions for a variation of a game called Wythoff's Nim. That year was a good year for Magnet projects in various contests (the Westinghouse/Intel competition, the Montgomery County Science Fair, etc.). I remember being pleased with my project, and it probably gave me a lot of self-confidence in school later on, but it's hard to be more concrete than that about how it helped my later research career, given the large difference between high school research and what was expected later on.
Q. How does your research translate into the search algorithm? How long does it take to see changes integrated into the site?
A. Our typical work cycle involves coming up with an idea to improve our search results, prototyping it, evaluating its effectiveness, and then either launching it, refining it further, or abandoning it and starting over. As you can see, the job is somewhat like research in terms of devising ideas and testing them, with some software development mixed in to prototype or implement our changes.
Depending on how ambitious or influential a proposed idea is, it could take anywhere from several weeks to months or even longer to do all the experiments, get the approvals, and implement in production.
Q. Why did you decide to do research instead of software engineering?
A. It was a matter of what I was interested in and what came more naturally to me. Some people really like writing code and hacking things together, and they can quickly put together some impressive or beautiful applications. I was never particularly good or fast at that kind of thing-- for example, programming contests were not my forte. I was always drawn to the development of ideas (not that this is always different than building systems), and I tended to view programming as a chore that was necessary to make the idea concrete. I may be a reasonable computer science researcher, but I'm not sure I would have been that good a pure software engineer.
Q. You were in a fairly academia-like position at Microsoft and now a less academic, but still research oriented position at Google. What's it like doing that in an industry setting?
A. While I was at Microsoft Research, the environment combined a lot of the advantages of academic research while removing a lot of the burdens that faculty often have. We had a lot of freedom to choose what to work on and had the same opportunities to publish as in academia, but didn't have to spend time applying for grants or grading papers. One of my coworkers called it heaven, and I think he had good reason to. (Not every lab is like that, of course. At others there may be more top-down direction, or limits on how much money you have to do your work.)
The drawbacks to industrial research tend to be longer term. There's no equivalent to tenure, and circumstances often change, even at places that have been around for a while. It may have seemed that the golden eras at places like IBM Research, Bell Labs, Xerox PARC, DEC SRC, etc. would never end, but eventually things did. The great research position you find today may not be there in ten or twenty years.
Q. How do students interested in similar positions get a foot in the door?
A. Probably the best way to get a job in a research lab is to do such brilliant research as a graduate student that they have no choice but to hire you. Of course, if you don't manage to do that, internships are a highly effective way of getting your foot in the door. Research labs get way dozens of strong applicants for every position they have, so all things being equal, they'll feel more comfortable hiring someone they already know.