MOOCs Suffer from Some of the Same Problems as Traditional Courses

I took a few online courses to enhance my computational photography skills for my startup. I was looking forward to trying out this new way of learning, after years of hype. I ended up completing one course fully, the second one mostly, the third one only specific parts, and the fourth one hardly at all before giving up.

Why? It turns out MOOCs have some of the same fundamental flaws as traditional courses:

First, they teach theory for days on end, without telling me what exactly I’ll learn if I go through it. A college student might be okay with “Spend 40 hours on this and something will come out of it”. As a working professional and a founder, I’ll spend time only if I understand the ROI, which means exactly what I’ll be able to do at the end of it, and how what I’m learning will lead to that result. Because there are tons of other useful skills I can acquire in those 40 hours: I could learn to market our product better. Or negotiation. Or Android programming. Or how to write a business plan. Or interview dozens of candidates to hire. The image processing course needs to justify why it’s a good use of those 40 hours.

This is a new way of thinking for professors, who’re used to sitting in their ivory towers and aren’t used to competing for listener time. They think that’s below them, they’re not selling biscuits, that their wisdom sells itself. It doesn’t.

The solution is to start from a practical problem and then work backwards and teach me the minimum theory needed to solve that problem. For example, an image processing course can start with a question: how do you sharpen a blurry image? This gives me a clear, motivating goal to work towards. When I achieve that goal, then lay out another question. And so on. The course would be organised as a series of problems you solve, each with a sense of satisfaction that you can now do something that you wanted to do, not as a series of topics. People care about the former; it’s why they’re taking the course to begin with.

As the teacher designs the course, starting from the problem of sharpening a blurry photo, for example, he may realise that it’s easier to start with how to blur a photo. He’s then refactor to make that the first question. Just as an engineer refactors his code or a startup refactors its product.

Second, one of these courses spent an hour outlining a technique and then said it’s effectively the same thing as another technique it already taught. Then, why did you waste my time? I want to learn a technique only if it lets me do something I couldn’t do before. These people should learn to respect their audience’s time.

Third, these courses were designed by university professors and seemed targeted to academics / researchers. They scored poorly on: “Having finished the course, what can I do with it in the real world? What app or feature can I build with this?” There was a lot of discussion on concepts that may be interesting intellectually, but are not feasible yet in the real world, or not useful enough by themselves. It’s fine to have a course targeted to researchers, but it should clearly be identified as such, so that practitioners like me don’t waste time. This requires the professors to first recognise that there’s a tradeoff to be made.

Fourth, they have a linear format, like a book. This defeats the point of a MOOC. I might as well buy a book or attend a lecture in person. When we have a new, interactive medium like the Internet, the content needs to be rethought from scratch for that medium. Not copy-pasted from another medium: you don’t make a decent TV show by taking a radio program and slapping some visuals on it. You don’t make a news channel on TV by having someone read the newspaper aloud. Likewise, you shouldn’t make a MOOC by recording some lectures and throwing it on the Internet. Adding some interactive quizzes as an afterthought doesn’t count. The core material needs be rethought for the medium. Here’s how we’d do that:

In the real world, different learners have different goals in mind for a course, and it would be good to let each learner achieve their personal goals. The course could outline a series of problems, and the student would click on one that he wants to learn. Like: Do you want to blur a photo? Create a panorama? Create a HDR? Align hand-held photos? Stabilise a video that was shot handheld? You’d click a problem you’re interested in learning to solve, and the course would then teach you the minimum you need to know for that problem. If a basic technique like frequency domain is required for two of the problems, it would be refactored into a common module, which is then included in those two solutions. But not in all solutions. Not like “Learn frequency domain, because it’s generally useful”. But like “Learn frequency domain because it’s useful for this particular problem you said you want to solve.” Again, think of how a programmer would refactor common code that occurs in multiple functions into a helper function and invoke it where needed. But only where needed, and not from every place.

If so many paths are hard to create in practice, having even two paths would be a step from the current linear format. And it would force the professors to recognise that one size doesn't’ fit all.

Fifth, these courses have a rigid, academic tone. Academics often seem to write for fellow researchers. Or use unnecessary jargon. Or try to come up with the most precise and concise definition for something, rather than the most understandable. They sometimes hesitate to approximate and simplify because the simplified explanation isn’t “accurate”. But you shouldn’t write to defend yourself against nitpickers; you should write to explain lucidly. Taking an approximation, for example, is fine if it helps understand. I don’t have time to waste to decode someone’s communication.

All these flaws make MOOCs not as much of a step forward as I thought. They’re old wine in a new bottle. The world will make real progress in learning when we let go of some of these old habits, write content from scratch for the medium, and write to optimise learning and real-world things people could do after finishing the course.

Consulting CTO. Earlier: Google | Founder | CTO | Advisor