Wednesday, 28 August 2013

Meet the students - An interview with Feynman Liang

What does it take to complete more than 36 Massive Open Online Courses while studying for a double major? Are MOOCs a good way to land a job at Google?
The Good MOOC interviews Google intern and "MOOCaholic" +Feynman Liang.

  • Can you introduce yourself?

I'm +Feynman. 21 years old. From Bellevue, WA. I'm currently majoring in biophysics at Amherst College and electrical engineering at Dartmouth College.

  • Can you describe what led you to take so many MOOCs?

I first heard about Udacity when I was waiting outside of lecture one afternoon. Curious, I checked out "Programming a Robotic Car" by +Sebastian Thrun on Udacity. At the time, I had just began playing with Python and had no real computer science background. By the end of the course, I had learned about Kalman filters, particle-based localization techniques, A* search, and a whole lot of other exotic computational techniques used in robotics. Hungry for more, I heard about Coursera through the Udacity forums. One class led to another, and here I am.

A big reason why I'm able to have taken so many MOOCs is because I'm fortunate to be in an environment which enables it. College frees me from many everyday obligations, giving me a lot more free time than the average person. In addition, professors and other students provide me with an intellectual community I can go to whenever I have questions about things being covered in MOOCs.

  • How do you compare your brick and mortar classes to MOOCs?

Both have their own pros and cons. I find MOOCs to excel particularly when it comes to lectures and assignments requiring little creativity. For example, professors in college lectures try to talk at the median speed preferred by the class. This leaves the upper tail of the distribution of students bored, and the lower tail lost. MOOCs allow for students to watch lectures at their own speeds and to pause / rewind as many times as necessary (without any annoyance from the professor or other student). Straightforward assignments are also effective when delivered over MOOCs because immediate feedback is provided and oftentimes there is an opportunity to retry. Unlike college, where I turn in a homework assignment and then only pick it up again when reviewing for an exam, I oftentimes find myself submitting the same assignment 3 or 4 times until I see the 100%.

Traditional classrooms are superior to MOOCs when it comes to personalized mentoring and uniform standards, which make assigning creative assignments particularly difficult. Not having someone help you figure your way through an open-ended assignment oftentimes leads to hours reading documentation just because you forgot some silly details. It's also more difficult to flesh out why you're having trouble in a subject, which is why I oftentimes take my MOOC materials to my actual professors and review with them. Additionally, the diversity of students taking a MOOC make it difficult to expect consistent grading of anything without a definitive right or wrong answer. Others taking the course may be a foreign student who just learned English recently, or a Ph.D. in the subject you're studying. This makes it hard to expect consistent feedback on your work and difficult to establish a sense of improvement.

  • Where do see higher education going in the medium to long-term future?

I don't think online resources will be replacing real classrooms anytime soon, but I do think MOOCs will disrupt our traditional conception of a classroom. While there are already electronic tools which supplement traditional classes (Blackboard, Moodle), they have typically been used as a repository for class resources. As mentioned in the previous response, lectures and assignments delivered over MOOCs provide many benefits over traditional assignments, and require almost no effort from the professor once the content is created. By shifting the lecture and homework part of the classroom to an online platform, professors can focus on adding value through personalized mentoring and open-ended projects.

  • There's a fair bit of MOOC skepticism out there (particularly when it comes to retention rates)... Do you have anything to say about that?

It's unsurprising, and I believe there are multiple causes. One of the causes is that students oftentimes sign up for classes simply to explore and get a feel for the subject/professor. I personally have dropped about 20-30% of the classes I sign up for. Measuring retention from the initial group of students who signed up just isn't fair. At a talk Professor +Daphne Koller (co-founder of Coursera) gave at Dartmouth, she presented much more reasonable retention rates when we look at student who completed the first three assignments in the course.

Another reason why retention rates are so low is because the student has to invest almost nothing to be a part of an online class. In contrast, my colleges cost $60,000 USD / year and the opportunity cost of my time spent on online courses is also non-negligible. Thus, loss aversion oftentimes has me (and perhaps other participants) prioritizing other activities over MOOCs (which are free and always available).

  • What is your college professors' reaction to you taking MOOCs?

The professors I've talked to about MOOCs are generally surprised by the depth the online courses cover. Most are fairly indifferent; neither overly excited or antagonized by the competition.

  • Which are you favourite MOOCs so far and why?

I've really loved:

  • Design of Computer Programs (+Peter Norvig, Udacity)

    This class was surprisingly difficult for being labelled CS212 (though I think Udacity has now moved it into the advanced section). The class taught me how to decompose complex problems into a procedural solution, and then how to write that procedural solution using minimal and concise Python. I've never had the problem of not being able to complete a programming assignment after taking this class.

  • Probablistic Graphical Models (+Daphne Koller, Coursera)

    Again, a really tough class. This one starts with an assumption that you understand basic probability theory and calculus. Through the 12 weeks of the class, you develop a deep understanding of PGMs, which are used throughout artificial intelligence and machine learning.

  • MITx - 6.002x Circuits and Electronics (Anant Agarwal, EdX)

    I really enjoy Agarwal's lecture style, and how he uses analogies and intuition to teach not just electronics, but LTI system theory. I took this class right before I started my electrical engineering classes at Dartmouth. This class covers many concepts core to engineering, and is perhaps one of the best produced MOOCs available (after all, it is the debut MOOC for edX).

  • Are there any courses you're looking for but don't exist?

I wish there was more formal mathematics classes available in a MOOC format. Harvard Extension School and MIT's OCW have some good content, but the community surrounding a MOOC really make learning much more fun. I'm especially excited for Wesleyan's Complex Analysis MOOC.

  • Did you take your MOOCs all on your own or did you create informal study groups with some of your friends from university?

I started completely on my own. Over the last two years, I've tried numerous times to get my friends to take them with me. Only one friend from back home, Andrew Tsai, has actually taken a serious commitment to it. Perhaps it's because everyone else I ask are people I go to college with and are already busy with real classes.

  • Keeping in mind your best MOOC experience so far. How would you improve on the course?

Find some way to centralize all my classes and keep track of deadlines. Neither Coursera, edX, nor Udacity currently offer a centralized location to keep track of your due dates. What would be awesome would be a tool which aggregates due dates across the three sites. I smell a possible hackathon project...

  • How has your MOOC experience been received at Google?

I put it on my CV, but when I talked to my recruiter she told me that it's about as impressive as "Relevant Coursework". For technical interviews, they really don't care about your background or CV, just that you can write working code in a limited amount of time. My certificates of accomplishment didn't impress during the recruiting phase; the most useful impressive part was the skills I developed from taking MOOCs.

However, my co-workers and managers have a very different response. Oftentimes, they are familiar with the professors teaching the online classes and are impressed with my grasp of the material. I distinctly remember getting into a discussion about Buttersworth vs Chebyshev filters one day at lunch; something I would have never done had I not taken Digital Signal Processing with Paolo Prandoni and Martin Vetterli. MOOCs give me something to connect over with other people at Google, which is actually a quite valuable thing to have!

  • Are some of your colleagues at Google also finding time to take MOOCs?

As far as I know, it's still largely unheard of. Most people I talk to are surprised that colleges are releasing all these classes for free and are generally very excited about the opportunity. My manager told me he would check Coursera out after I told him about it. So far, I've only met one other person who's actually taken classes on Coursera. Coincidentally (or perhaps not), they are an intern as well!

Write a comment below or share this post!
Post a Comment