I question your teaching comment. Sitting there and receiving a lecture is a rather small part of learning (varies per student). For the rest of it you're going to find that class size is by far the biggest determiner of success.
I wonder how the best teacher in the world teaching 5 classes of 40 would stack up against 200 people chosen at random from a pool who had already passed that class working 1:1 for 1 hr per day. Probably the ideal case is somewhere in the middle, but the point is that it's allocation of the teacher's time, and not production quality, that's the bottleneck when it comes to scaling teaching.
I meant to expand that to clarify that other aspects (supporting students as they work to apply the lectured material) would/must be done 1:1 in person, but the core lecture can/should be delivered by those with particular skills in that area and that we could apply extra effort to editing/production value.
My son's calc class is taught via watch the lectures online/at-home and then apply them in class and at-home via exercises. The only difference from what I proposed is that the same teacher pre-recorded the lectures, which seems to a practical solution but not an inherently necessary constraint (and there are likely thousands of calc teachers making average quality such videos).
But absolutely a personalized teacher must be available 1:1 for at least part of the overall process.
I wonder how the best teacher in the world teaching 5 classes of 40 would stack up against 200 people chosen at random from a pool who had already passed that class working 1:1 for 1 hr per day. Probably the ideal case is somewhere in the middle, but the point is that it's allocation of the teacher's time, and not production quality, that's the bottleneck when it comes to scaling teaching.