1) Testing the hypothesis that university is bad
2) Testing the hypothesis that university is not worth the cost for the vast majority of people
What the Fellowship might plausibly be doing:
1) Testing the hypothesis that university is not worth the cost for people like the Thiel fellows
The actual motivation of the Fellowship, which is all over the press and has been related to me by Thiel Foundation workers, is that students with great potential accumulate tons of student loan debt in college and subsequently, instead of taking on projects that might actually help the world, are financially pressured into working at hedge funds and the like. The Fellowship is offered to these students as a two year break to attempt these projects supported by 100K for living expenses and, crucially, extensive mentorship.
Again, the point is not to discredit universities, but to keep talented kids away from hedge funds (ironic coming from Thiel but there you have it).
Edit: Also note that a failure of the Thiel Fellowship project does not disprove the education bubble in general, since it only applies to a very small group of people.
Surely kids as brilliant as these ones sound are getting full scholarships to college?
If you really want to test the hypothesis: that university is not worth the cost for people like the Thiel fellows then I reckon you'd need two groups. One drops out of university and gets $100,000 and extensive mentoring. The other group stays in university and still gets $100,000 and extensive mentoring.
I see it's been a while since you went to college. Let me introduce you to a concept called "need-based aid:" it's virtually impossible to get a full-tuition scholarship at a top school based on academic achievement.
See also, market segmentation -- http://www.joelonsoftware.com/articles/CamelsandRubberDuckie...
I agree. It's not really a valid experiment, which is why I said "might plausibly be" instead of "is", just to acknowledge some connection with the education bubble debate. Though two things to keep in mind for a valid experiment are that from the 100K for the second group would have to be subtracted the costs of university over two years (for me [mostly my parents actually] that's more than 100K), and that at good universities at least extensive mentoring is readily accessible. So it's a reasonable approximation at least.