Earlier quoted context omitted.
I think both you and the grandparent are making a mistake. He assumes children dislike math. You assume children will like karel. I think the truth is that children have different interests and that you can only bring them to programming by latching on to those interests. Many children like math. Many other children will not like karel. The result: to engage as many children as possible, you need several introductory…
Nobody assumes that children inherently dislike maths. They can't dislike something they don't even know. The point is that children start to dislike maths once they go to school. And they do - all kids in my class either already knew basic maths before going to school or were completely bored by it. The same will happen with programming once it will be "taught" in schools as well. Think of it this way - if you want…
You can get a bigger n, not by increasing class sizes alone, but also by letting teachers teach for longer per day than children learn. I.e. use you math per hour taught, not for salary.
What this means in practice is, use teachers for higher value added work, but let computers (or somebody less qualified) oversee other parts of the learning process.
What seems to work fairly well (I remember a study about it), is to let students read material at home, but solve problems in class. That's the opposite of the traditional getting lectured-at in class / exercises as homework approach.
Khan academy is another vector for leveraging teachers. It does not replace a good teacher, but it can augment a teacher.