There's an excluded middle here.
The tiny state school might have the best machine learning class in the world.
My counter example would be the Algorithms course from Princeton with Robert Sedgewick. I would rate that as basically "just ok."
As a university teacher the most important thing you have to do is to pitch the level right. Anything that isn't a top-level school suffers from the tyranny of low expectations, and the cliff after the first 20-25 schools in the rankings is really quite steep. There are no public ivies.
Yet somehow the University of California at Berkeley has one of the 4 best computer science programs in the world, and it's better than all but 3 programs at private schools (Stanford, CMU and MIT).
The critical component is someone putting together the order in which you learn it. Perhaps that requires top educators, but I don't think it really does. What MOOCs do that really benefits me is removes the wasted time of discovery with an ordered process that makes sense.
My daughter does Khan Academy. The information she is learning isn't particularly novel, however the order and structure is. That's the key.
Google searching for the same topics will get you great information as well, but you need to know what to search for. You need to include that search time through each part. You need to filter out bad material.
That's my view at least.
Actually, I'm starting to get worried about this. I took the first version of Andrew Ng's ML class in late 2011, and I'm taking a different Coursera class now, and there hasn't been much improvement that I can see in the basic offering. Specifically, one area where online classes should be able to outperform traditional classes is in the handling of pre-requisite material (since in a MOOC it isn't necessary for all of the students to see the same material, some could start at different points), but there hasn't been any work or progress in that direction at all. That may explain why almost all online classes are intro-level.
Or, if we allow students from remixing music and images in art studies, then their education is going to get worse?
Or if we allow students to take apart technical diagrams and improve on them, then their education is surely going down the hill?
Please enlighten us how such restrictions would help a student who want to learn, experiment and improve their skills. It might help akbar501 businesses if he could take something under bsd/mit and slap proprietary restrictions on them, but I do not count that as "benefiting education". Thats like me asking people to send me money in order to "benefit the economy".
MOOCs like Udacity and Coursera have a fair amount of funding...
I should have been explicit with the open content I was referring to. Specifically, I meant the content from Khan Academy, CK12, and the like. Not the for profit companies.
On the positive side, the benefits are substantial. 1. Democratized access to the world's top educators, 2. Low cost distribution of education.
On the negative side, the MOOCs are kind of the large classroom problem taken to the extreme. It's generally agreed that a classroom with more students per teacher is not as good as a lower student/teacher ratio.
But to critique my own negativity, the people working on MOOCs are smart and motivated. It's a bit naive to think that the current MOOC is as good as they will ever get. Clearly, these are early stage products that have substantial evolution and improvement in their future.
The one area in open education that needs to improve is around content licensing. If you look at most open educational content, the licenses are restricted open source (GPL like) and note unrestricted open source (MIT/BSD/Apache like). I fail to understand how making the content unrestricted would not benefit everyone.