[xiaoma]

Out of curiosity what sorts of things do you watch on Coursera? I was a big fan in 2013-2014 and did a bunch of courses. Some of them, like the Scala course, were very self-contained and great, especially with the automated graders. Mostly, I found the platform frustrating, though.

The artificial pacing of the courses made it essentially impossible to go through any faster than you would in an an in person class. Also the honor code prevents people from actually helping each other when they're stumped in the forums. There's clearly some tension between my goal of learning and the platform's focus on credentialing (via credentials that carry almost no weight).

Many courses were watered down versions and since different institutions break up their coursework slightly differently, there was a constant problem of repeating material or missing parts of prerequisites. Now it seems nearly all the courses are watered down and most don't seem to have any prerequisites at all.

Is it possible to do an entire technical degree's worth of material on Coursera, even at the undergrad level? Regardless of the value of their own certificates and capstone projects, if a student could learn enough to pass a GRE subject test, that credential that would actually carry some weight.

[icc97]

Andrew Ng's Machine Learning 18 months ago and now his deep learning course. I do them because he's the best lecturer I'll ever have. So I'm not so bothered about credentials because I know I'm learning valuable information. The ML course was a bit limited as it was all Matlab, but the deep learning is all python and tensorflow. I also started the Scala course, but stopped as life was too busy, and I'm half way through Geoff Hinton's NN but he's just not as good at explaining things as Andrew Ng.