I don't know if I'm misinterpreting Derek Muller, but I didn't get the impression that people had to be confused/overwhelmed to learn. I think it's only necessary when there's incorrect prior knowledge.
If interested, here's my take/rant about it:
I believe that prior knowledge and misconceptions can change how people deal and interpret new information. This can help or it can hurt the learning process.
It can help because it often makes new knowledge more likely to "stick." For example, a soccer fan is more likely to remember soccer scores [1]. I know there's other research that indicate more prior knowledge leads to quicker learning, but I don't remember them off the top of my head.
It can hurt in much of the same ways Derek Muller describes. When hearing new information it can easily be misinterpreted to match current knowledge. For example, anecdotally, those only familiar with the words "negative" (bad) and "reinforcement" (adding/reaffirming) are more likely to misinterpret "negative reinforcement" to mean "punishing behavior by adding something bad" instead of "rewarding behavior by taking something bad away." I believe there were similar findings/studies within statistics. Another common one is F=mv as shown in Muller's videos. I'm sure there are plenty more examples.
I believe Derek Muller's problem with Khan's videos is that they might not challenge incorrect previous knowledge. I agree with Muller and personally feel that Khan Academy will likely encounter some of these same problems when students watch the videos.
But, I also believe that the other half of Khan Academy (exercises) provide feedback that videos alone otherwise do not have. I believe that the requirement for mastery through the exercises can and will challenge at least some of these misconceptions. Particularly if the questions are good questions.
I also believe that Khan Academy is going to find themselves in a position where they can potentially have a lot of interesting data. I'd love to be able to see whether they can use this data to find some of these common misconceptions and then improve videos and/or questions. In the very least, I believe that their exercises will serve as a good metric to see where misconceptions are being poorly addressed.
We also have to keep in mind that, at least in the pilot in Los Altos school district, these videos are not the only source where students are learning. They still have time with teachers where these misconceptions can be addressed. In addition to that, they'll have the opportunity to review the videos to contrast, reflect on what lead to the misunderstanding, and hopefully reaffirm the new knowledge.
tldr - I believe Khan Academy's exercises and class/teacher time might help deal with the issue of misconceptions.
[1] Morris, 1981 - http://onlinelibrary.wiley.com/doi/10.1111/j.2044-8295.1981....