One specific example he provides is developing better digital textbooks. Specifically, digital content that is more than just a PDF version of a paper textbook. Some options:
* Use multimedia effective (video, animation, sound could all be integrated effectively with the text)
* A/B testing to determine the most effective way to teach a concept
* Mine data from interactive problem sets to see where understanding tends to break down, then iterative on the text until you teach those concept effectively
Another specific example he touches on: better video content. Consider that a lot of MITx's video content was just a recording of a professor lecturing in a monotone at a blackboard. This is the least possible effort they could have made. For a specific example of a better way to lecture online, he references Khan Academy's video. I fully agree. And he points out that not only is this approach effective, it was also inexpensive to produce.
* Use multimedia effective (video, animation, sound could all be integrated effectively with the text)
* Interactive examples (see this HN-featured page about Markov models for some nice examples: http://setosa.io/blog/2014/07/26/markov-chains/index.html)
* A/B testing to determine the most effective way to teach a concept
* Mine data from interactive problem sets to see where understanding tends to break down, then iterative on the text until you teach those concept effectively
Another specific example he touches on: better video content. Consider that a lot of MITx's video content was just a recording of a professor lecturing in a monotone at a blackboard. This is the least possible effort they could have made. For a specific example of a better way to lecture online, he references Khan Academy's video. I fully agree. And he points out that not only is this approach effective, it was also inexpensive to produce.