Archive for October, 2014

Observations about Coursera’s “Fundamentals of EE,” by Don Johnson

October 12, 2014 Leave a comment

Around January of 2014, I finished watching all the lectures for Don Johnson’s Fundamentals of Electrical Engineering on Coursera. While sorting through e-mail from around that time, I came across some observations I sent to a few colleagues, which I refined to share here.

1) As far as I can tell, Don’s course is unique in the set of ECE intro courses. There’s no class at another university that I can directly compare it to. It’s sort of a cross between UIUC’s Analog Signal Processing class (a circuits/signals-and-systems hybrid) and Georgia Tech’s Introduction to Signal Processing class (which focuses on discrete-time signals-and-systems), with information theory thrown into the mix. Don compares analog and digital communication schemes in the context of channel capacity. The course covers both analog and digital processing, with the focus on signals as carriers of information. The scope of the class is breathtaking; the last time I looked at a class and got a similar mind-blowing impression of its depth was MIT’s old Structure and Interpretation of Computer Programs.

It’s utterly brilliant. (But, I have to be careful not to equate “something a professor thinks is really interesting” with “something that will gel with undergraduates.” These are not always overlapping sets.)

Don includes a lot of material on Fourier series and transforms and frequency responses, but he doesn’t include anything on Laplace transforms. The Laplace omission makes sense since there’s not a lot of emphasis on generic switched-voltage-source, capacitor charging/discharging examples that typically eat up a lot of the time in most traditional sophomore circuits courses. Don goes into the frequency domain quite early and generally remains there through the rest of the course.

The uniqueness of the course, to me, provides one of the strongest arguments in favor of MOOCs. If you want a standard sophomore circuits course, nearly every EE department offers one, and frankly, there won’t be much difference between the one at Georgia Tech (ECE2040) and its equivalents at Southern Poly. But before this MOOC, the only way to see this material assembled in this particular fashion with this particular vision would be to move to Houston.

2) The sound quality is utterly horrible, and it’s a revelation to me how much that effects my overall perception of the class. The fan noise from Don’s computer is quite evident, and although I don’t know for sure, it sounds like he’s using the built in mic on his computer. From lecture to lecture, or sometimes in the middle of the same lecture, the sound quality will suddenly change, as if someone was experimenting with different levels and noise reduction settings in some audio editing software package. I was alternating between Don’s course and some of Udacity’s courses, and the higher production quality of Udacity’s products is striking.

Magnus Egerstedt’s Coursera course on Control of Mobile Robots (basically a graduate linear systems theory course like Georgia Tech’s ECE6550, with neat material on robots added in) is on Coursera, but it was taped in the fancy recording studios at Tech, so the audio for his course is much better. That said, although the audio in Magnus’s course was mostly noise free, it was encoded at a relatively low level. I like to watch lecture videos while riding on the stationary bike, and with Don’s and Magnus’s courses, that was hard to do, even with my laptop volume control maxed out; but with Udacity’s courses, the sound was loud enough (perhaps they used professional limiting program like the Waves L1?) that it overpowered the sound of the bike.

I can’t remember who said this, but someone once noted that the main key to professional looking video is professional sounding audio.

3) The way the in-lecture quiz questions were handled was absolutely maddening. On Coursera, an introduction to the upcoming quiz question is embedded in part of the preceding lecture segment. In Don’s course they were clearly thrown in after the fact; sometimes, it almost feels like they interrupt what he’s saying mid-sentence. They’re jarring. It would sometimes take me a second or two to realize I was being quizzed, and the audio hadn’t stopped because of an internet slow down. I didn’t feel like they kept me engaged, the way the Coursera quizzes did; I felt like they interrupted the flow, and it was hard to get the vibe back after the sudden interruption. The most bothersome questions were the ones that were thrown in to correct errors made in the main presentation. They’d start with a phrase like “The instructor made an error when writing the node-voltage equations. What should the second equation have been…” Sometimes I wanted to yell at the screen. It sounds like a small, petty detail, but it’s interesting how many small details add up to create a perception, good or bad, of the experience.

4) While Don put together this course brilliantly, I’m not sure Don is the best person to present it in this format. He sometimes tries really hard to sound really excited, and he’s clearly putting in a massive amount of effort, but his voice could send a dozen kittens in Consumer Report’s laser pointer test center into the deepest slumber. He tends to trail off at the end of sentences, so sometimes the start of the sentence is above the threshold of the fan noise while the end of the sentence starts to dip below it. I realize some of that awkwardness probably stems from the unnaturalness of having to talk to an empty room, which I find to be tremendously difficult.

I should note that my comments above apply to the first Coursera offering of the class; they may have made improvements in newer offerings.

Categories: Uncategorized

The tyranny of semesters & the trouble with Coursera

October 3, 2014 1 comment

Throughout numerous news reports and blog posts, comments on those news reports and blog posts, and e-mail discussions prompted by them, many legitimate criticisms of methods of teaching and learning outside of the usual on-campus class structure have been raised. But those usualĀ on-campus classes have their own limitations; we are just so accustomed to working around those limitations that they’re barely noticed. Why should these same limitations be mapped into the online space? Why do we keep putting horseshoes on our automobiles?

This leads us to the problem with Coursera: it has the word “course” in its name. I’m not saying the courses themselves are bad — the quality on Coursera varies wildly, but many of them are quite good, and I hope to do a “course” on Coursera at some point — it’s just that the very concept of a “course” is artificial.

Courses, along with time units like semesters and trimesters and quarters, are organizational artifacts borne of the practical need to allocate chunks of time associated with chunks of physical space and chunks of biology called “students” and “professors” and get them to line up in some way those chunks of biology can readily remember, like meeting TuTh at 2-3:30 or MWF 1-2, staring on a certain date and ending at a certain date. Are any of those optimal in any way? Can anyone tell me if there’s any research on whether three days a week for 50 minutes is better or worse for learning than two days a week TuTh? Maybe there’s some material that’s best learned MTuW of WThF. Can anyone tell me? Has anyone even asked?

What about time of day? I have heard rumors of the existence of “morning people,” and there are probably some faculty and students who fit in that category. But for students who are not in that category, I conjecture that an 2:30 PM class is going to be a hell of a lot better for learning than an 8:30 AM class. Has anyone studied that? In all the discussion about problem-based learning and flipped classrooms and clickers or whatever, if someone could show that simply not having class at 8:30 AM resulted in massive improvement in learning outcomes, would we change our scheduling to accommodate that finding? Or would we not even bother to ask the question, given how limited we are on classroom space, and just implicitly state — whether we mean to or not — that we believe that the material taught at 8:30 AM is less import than the material taught at 2:30 PM.

This connects to Lanterman’s Temporal Maxim of Education: Any method of online course delivery is superior to an in-person class that meets at eight-assclock in the morning.

Dear university professors reading this post: Did you stop learning after you finished your PhD? If not, how many things have you learned in the past decade that had specific start and end dates?

Categories: Uncategorized