There is nothing new here. At 13:33 she says that a student testimonial shows that this online, a/synchronous format offers something deeper than a classroom experience. So, someone is awake somewhere else in the world and can respond. What is the quality of the response? Does Koller bother to research that? She seems to think that when and where a click happens constitutes meaningful data or insights into human learning.

Maybe. Maybe it’s just noise. Her bias seems to be that this is all deeply meaningful–because technology is involved and how cool is that. But some of its saliency is only evident when large group learning is the venue. In other words, if one has preselected for online, large groups to engage in a common learning task, operations that identify mistakes or common errors will of course generate what one would consider to be meaningful data, or the identification of common errors. Try this approach with say, a group of six or ten students and the resultant matrix of cognitive or learning errors will not be amenable to the same treatment.

Koller at 15:48 says that when one has a large sample of the same wrong answer, the fix can be the same for all via an error message. Again, I submit that Koller is offering a teaching model relevant only within a narrow confine of delivery modes, nothing more. This research offers nothing not previously known about human learning but would support an effort to lower costs of educational delivery through the elimination of human interaction in favor of algorithm-mediated behavioristic instruction.