A while back I wrote about Stanford’s online DB course. Many of my friends who expressed interest unfortunately couldn’t afford the time to invest in an online course. Luckily, that wasn’t their last chance. But before I get to the upcoming courses, let me reflect on online courses in general and why Stanford, deservedly, got a wide coverage and following.
Online Lectures
I’ve been watching (and listening) to university lectures since circa 2004. Back then there weren’t too many available. In 2005 I discovered UC Berkeley’s WebCasts. These were RealVideo lectures from early 2000s on many hard sciences and some humanities. The quality of the courses were as expected high. The main issues were that RealVideo only allowed streaming, so there was no way of downloading. Besides that, the video was in rather low-resolution, low-quality and recorded from a single angle and reading the blackboard was unnecessary strain on the eye. Back in 2006 I wrote a python script using MPlayer to parse the pages, play the video with MPlayer and dump the resultant stream to files. The main issue with this was that a single hiccup and the lecture had to be downloaded from start again. Using this slow and painful method I downloaded dozens of courses and filled about 250GB worth of media and shared to colleagues and friends. Berkeley improved this by introducing Mp4 downloadable links in higher resolution starting in 2009 (I think) and now moved completely to iTunes and YouTube as platforms.
During that time I discovered MIT and Yale but neither could equal Berkeley in the number of courses or the topics. Berkeley’s Physics for Future Presidents is perhaps the best example that I can think of to show off what Berkeley had to offer. In 2008/2009 Stanford’s Leonard Susskind’s Modern Physics was available online, which is another top-notch lecture series.
As MIT, Yale, Harvard and Princeton made available more courses, I discovered Academic Earth, which I think at this point is perhaps the single best site for high-quality, diverse and highly usable video lectures.
Why University Lectures?
The process of learning isn’t linear nor comes in one flavor. Different people have different preferences on how they rather best learn or study something. Indeed, different topics might be best learnt in dissimilar methods. From books, tutorials, hands-on examples to demonstrations, labs and homework assignments. But university lectures place one’s mind into the classroom state. With all the students, rigor in subject treatment, questions from students, they all contribute to the state of mind that is very important in taking the information seriously. The lectures are also divided in such a manner that’s expected to be reasonably-paced. There are also review sessions and sometimes quizzes.
At this time I’m enjoying Justice on political philosophy and morality (by Michael Sandel of Harvard) and Physics I: Classical Mechanics (by Walter Lewin of MIT) on Academic Earth, besides Stanford’s DB, ML and AI courses, of course.
With tutorials, books and other forms of teaching most of the above is lost. Not to mention the caliber of the teachers in these universities are expectedly quite high.
The Stanford Model
Stanford introduced 3 “experimental” online courses in Database, Machine Learning and Artificial Intelligence in the Fall of 2011 (it’s mid-term exam week as I write this). Besides the fact that these courses are highly sought-after, the Artificial Intelligence course is taught by Google rock-star Sebastian Thrun who’s behind the Driverless Google car (video featuring Prof. Thrun).
Unlike all other online courses, Stanford’s approach was more course-like and less of a video recording of a lecture, as all others are. MIT might offer notes, slides and transcripts, but Stanford’s courses have forums where TA’s and teachers participate, sidecasts where teachers do online video chat with students, online and interactive quizzes that pop-up during the lectures, assignments and exams with automated grading. In addition, at the end of a course, the teacher will give a signed statement of accomplishment to the students who participated. This is in addition to prerequisite and preparatory lectures, external resource links, books and reading materials, transcripts and translations, downloadable video and lecture slides.
So there is not only much more interactivity with thousands of other students with study groups gathering in person and online, but also there is interactivity with the teachers and with online tests with immediate feed-back. These features make the courses much more than the sum of its parts. They exploit the internet and multimedia to their true potential and deliver a remarkable package for a globally available learning experience.
For those who can’t, or don’t want to, take up the assignments and exams, they can choose the Basic track. The course material is promised to be available to all during and after the end of the courses.
More Courses!
With the success that the DB, ML and AI classes saw, with well over 300,000 students enrolled in total, it’s no surprise that they are expanding this to other subjects as well.
- CS 101 by Nick Parlante.
- Natural Language Processing by Dan Jurafsky and Chris Manning.
- Software Engineering for SAAS by Armando Fox and David Patterson.
- Human-Computer Interfaces by Scott Klemmer.
- Game Theory by Matthew Jackson and Yoav Shoham.
- Probabilistic Graphical Models by Daphne Koller.
- Machine Learning by Andrew Ng.
- The Lean Launchpad by Steve Blank.
- Technology Entrepreneurship by Chuck Eesley.
- Design and Analysis of Algorithms I by Tim Roughgarden.
- Cryptography by Dan Boneh.
These courses will start in January or February 2012. I expect they will announce others, most probably Database and (prof. Widom announced that the course will be available next fall) Artificial Intelligence, or so I hope.
Final Thoughts
Needless to say, Stanford’s online courses is a very welcome project and one that will change the face of education, e-learning and especially distance learning and autodidactism by raising the bar and setting new standards. US universities aren’t the only player here. Perhaps the best example is India’s National Programme on Technology Enhanced Learning (NPTEL) which already has hundreds of courses and plans to expand to over a thousand. Their Artificial Intelligence course, by prof. P. Dasgupta, is highly acclaimed. From the other end of the spectrum there is Khan Academy, which is a not-for-profit educational organization led by a very enthusiastic and charismatic figure, Salman Khan, who’s behind the 2700+ videos on most all topics. Khan Academy also has practice problems with scoring method and graphs tracking progress over time and in each subject.
Education and learning in general has never been this accessible before. With the internet, Wikipedia, free books and video tutorials and university lectures available to anyone with an internet connection. We no longer have an excuse for ignorance but our lack of will. Not looking up things we don’t know, suspect to know correctly or completely or want to learn more about, is practically inexcusable. Of course with all that information and availability also comes a sizable amount of chaff that one must weed-out from wheat. The signal-to-noise ration can be quite disappointing on some topics. But at least that can be done faster and much easier than it was only a decade or two ago, and not searching for criticism or opponents to get a more balanced picture is as inexcusable as not looking up in the first place. At least now one can also reach experts and hear what they have to say. We can even attend the best universities from the comfort of our armchairs.
I can’t help but wonder if the next generation will look at physically attending lectures as this generation does to putting pen to paper.