[yummyfajitas]

This is a very promising result - apparently Khan is roughly as good as a traditional teacher provided a teacher is in the room. Now the important question is to figure out how much of the teacher can be done away with.

For example, can 1 teacher + Khan educate 80 students as well as 1 teacher without Khan educate 40? Or can we replace the (well paid) teacher with a lower paid day care worker for at least some of that time?

If Khan can reduce the number of teachers required to educate students, that would go a long way towards reducing our out of control spending on education.

[capnrefsmmat]

I don't think that's the way to approach this. Replacing teachers with videos isn't going to make education better -- in fact, I'd argue it would make it worse.

There's a great deal of research in the effectiveness of different kinds of physics teaching, for example, and it all shows that the most effective method is interactive. Students often form misconceptions about concepts, and the only way to break those misconceptions is to engage them, have them think about problems, and tailor your explanations to address their confusion.

A video can't do this. Instead, a video could provide the groundwork so that a teacher could spend all his time working with students interactively.

So if you decide to cut down on teachers because of the videos, you're giving up the potential advantages in teaching that the videos would bring.

I can link to a few papers if you're interested.

[yummyfajitas]

Replacing teachers with videos isn't going to make education better -- in fact, I'd argue it would make it worse.

Why argue? Why not simply test it?

There's a great deal of research in the effectiveness of different kinds of physics teaching, for example, and it all shows that the most effective method is interactive.

A video can't do this. But perhaps 2 hours of video + 1 hour of interaction might be just as good as the 2 hours of lecture + 1 hour of interaction provided by the current system. If so we can cut our teaching expenses by 2/3.

Khan provides a virtually free substitute for some pieces of our current educational system. The trick is to figure which pieces.

This is a big problem - the US spent $864B on government-sponsored education in 2009, more than it spent on the military.

Students often form misconceptions about concepts, and the only way to break those misconceptions is to engage them, have them think about problems, and tailor your explanations to address their confusion.

Khan is attempting to build automated systems that do exactly this.

[capnrefsmmat]

> A video can't do this. But perhaps 2 hours of video + 1 hour of interaction might be just as good as the 2 hours of lecture + 1 hour of interaction provided by the current system. If so we can cut our teaching expenses by 2/3.

Okay, but that doesn't improve education, it just makes it cheaper. It would certainly be a useful start; one could invest that money in hiring better-qualified teachers, or even a program where college students come in to do interactive tutoring.

> This is a big problem - the US spent $864B on government-sponsored education in 2009, more than it spent on the military.

How much of this is personnel and salary costs for teachers?