Hacker News Reader

Comment by tboyd47 | Hacker News Reader

Agreed. The courses seem to lack depth. And the video-only method of instruction makes it feel like an infomercial. It's a shame, because the Coursera platform has a lot of potential. Andrew Ng's machine learning course was as good as a real university course.

To be fair, the specialization I was on included a final project which could have been cool.

View on HN

Yeah, the video-only thing is problematic. But I think they're stuck there. If you're having a hard time convincing students to pay $50 for the class, can you really ask them to buy a $200 textbook?

Fair point. I just don't think they've found product-market fit yet. Before, they were giving too little value to the schools, and now they're giving too little to the students.

Maybe if they switched to selling textbooks and bundle the video course with the textbook. I would never go out and buy a $200 textbook to something I have no foundation in, but if I'm getting a video course with it, maybe.

I think the "only" part isn't referring so much to the exclusion of a textbook as the exclusion of projects and exercises. Videos and textbooks are both too passive. The biggest issue with MOOCs in 2015 is the relative lack of active engagement with the material through checked exercises and with teachers, tutors, mentors, and peers through one-on-one engagement (whether in person or via phone or chat) rather than one-to-many forums. Both issues are more expensive to solve, but I'm disappointed that Coursera hasn't used their paid courses to attack them. I suspect that the price of those courses isn't high enough to justify it.

I think that while we've added a couple educational tiers above reading books and articles on one's own, we're still missing one or more between paid MOOCs and "real" higher education.

At Tuva (tuvalabs.com), we are creating an environment to learn foundational statistics and data analysis concepts and skills that go well beyond just video instruction. To get a feel for experience, check out a 30 sec demo here: https://vimeo.com/137256993

That looks nice, but people want problems to solve - gradual problems, that have a nice gradient of increasing difficulty, and are sufficient in number, until they can bring up the student to a feeling of mastery over the field.

I was looking for such a set of problems to gradually learn functional programming by exercises but all I could find was a video course that also had a set of 10-20 mini-problems to solve as it progressed. That was by far too shallow for this learning task. And to think that FP is the love child of Hacker News, with tons of posts in the last few years, yet, there is no comprehensive resource of exercises.

What about project Euler and the like?

I think the passivity of videos and lack of exercises is just half the problem. The other half is that there is no human supervision of each student. There is no feeling of accountability to a professor. In real life you compare with your colleagues and are evaluated by a teacher. That tends to make all this activity much more involved.

I think the problem can be solved by organizing study groups and paying for a tutor to see you through your online learning career. Regular meet-ups in real life with a person, to check the state of your studies and discuss strategies and future courses to take up would be better than just creating a free account and starting to learn on your own.

MOOCs could be complemented by in-person educational coaching. That would reduce the dropout rate and make it seem more "real". MOOCs don't need more teachers, they need more coaches - people with experience in the psychology of learning.

We're in violent agreement, and I meant to allude to the things you spelled out more clearly. I'm disappointed that Coursera hasn't (yet?) started using their treasure chest to push this sort of thing, but maybe that just means it's a market opportunity for someone else.

Reply on news.ycombinator.com