A startup opportunity? Video cameras are incredibly cheap, a PC with eight webcams, software to do initial editing of a lecture with final edit done by a human?
I can imagine a lectures could be put on video for much less than $2000 each with the right technology.
Webcams are really horrible though. If you're working with an eye to posterity you'd be better off using a few regular cameras, and people to operate them. Editing is a mixed bag. You could show the blackboard, but depending on what's up there you might be better off doing titling in software and having it float beside the lecturer.
Certainly you can do it for much less...$2000 was a ballpark price for doing something of pretty decent quality that's going to be visually engaging - the best lecturer in the world is a lot less compelling when s/he's been reduced to the size of a postcard or less. I think auto-editing would be next to useless except for the most basic things like:
while(wide_cam.frame.rightside.movement = true)
wait 10 seconds
edit_cut_to (medium_closeup.cam[1])
end while
It'd be a lot simpler to just hire a bunch of people and pay them to do the work - trying to automate what is essentially an aesthetic process is a recipe for wasting money. Bear in mind that you do get some economies of scale as you do more 'episodes'.
On the other hand, I do think there's a great startup opportunity here, mixing some basic production techniques with some visual FX techniques (let's animate that polynomial as the lecturer explains it to us) and some decent interactive controls (pause and rewind are sloppy relics of linear information access - what we want are the kind of contextual controls at which game UIs excel).
Mturk might be cool. "Here are 3 30 second video clips. At any given time, hover over the one with the clearest view of the subject matter. Here's 5 cents.
I can imagine a lectures could be put on video for much less than $2000 each with the right technology.