The video lectures at the Khan Academy are terrible.

That's a really shallow criticism of what might just turn out to be the most significant revolution in teaching in the last century.

Khan Academy is significant because of its scale (they want to systematically cover all of the basic educational material and they've already covered a lot), because it's free and because it builds the on-line system to provide a combination of self-learning and tutoring.

The important thing: do those videos work for students? KA is in very early stages but the early results show that yes, as imperfect as some guy on the internet might think they are, they do seem to work. Children learn from them.

Could they be better? Maybe, but existing videos are a solid baseline. When there was nothing like that, anything was better, so kudos to Sal for doing the videos. But now that we have a baseline, the standard for making an improved version is higher: we can't just throw (and implement) random suggestions like "less repetition", "prepare better for the lectures" or "do it faster". Given that human psychology is often counter-intuitive, those "improvements" can actually turn out to be worse at teaching so new ways of doing things should be tested against current way to see how they affect the desired final outcome.

In fact, Sal Khan says that the style of his videos (the lecturer is not visible, you just see a whiteboard and his voice) was incidental and at some point he thought that it would be better to make them in a more traditional lecture style i.e. where the lecturer is visible in video.

This sounds like a reasonable idea given that most on-line lecture videos work that way but it turns out that having lecturer in the video detracts from learning. It puts the focus on the lecturer and not on the material.

So when people criticize Khan Academy videos and propose various ways of improving them, we shouldn't just assume that just because an idea for improvement seems plausible, it'll improve the effectiveness of videos.

Finally, just like in open-source "code speaks better than e-mails", the criticism would be much more credible if he re-did at least one of the videos in a style he thinks is better and let us judge based on that.

I don't think they're all bad, but they are slow. I do have gripes with some of the VC and investments vids. Having built and sold a startup, I know first hand that some of these concepts aren't just over simplify but some are just dead wrong or severely lacking information that could have easily been included. Of course, Khan was a hedge fund manager but never actually built a scalable startup or was an investor so I can't really fault him entirely since books aren't the greatest coverage in this area.

I disagree. The lectures are aimed at a demographic that probably does not have the background that hacker news readers likely have.

He's given every child in the world with internet access an MIT-educated tutor who cares enough to make these videos available for free.

The video lectures at the Khan Academy are terrible. Slow, not to the point, lecturer repeating himself because it was taking an age to actually write anything on the ridiculous blackboard they're using, and he's trying to fill dead air.

No, they're great, and the very things Ducker criticizes are what makes them great. Most people that need help working problems find Kahn's "ridiculous blackboard" very helpful (myself included). His repetition is also useful if you're not an expert. And the pace? It's perfect if you're seeing the material for the first time.

Not everyone is Andrew Ducker. What's obvious to him is not necessarily obvious to someone else.