Judging by posts and comments by graduate students in math.stackexchange.com and mathoverflow.net, I have come to the conclusion that a large part of their knowledge comes from SE sites and the Wikipedia, as opposed to traditional publications such as books and papers.

I was deeply shocked the first time I heard of a student researching the literature through Google, which spits out at least as many SE and Wikipedia answers as references to books and journals. But, of course, nowadays I do the same, certainly more often than going to the library!

Could it be that the often touted resilience of conventional publication methods is finally giving way to new forms? Should an entire generation go in a new direction, the old ways will not be able to keep up regardless of their advantages and their virtues. We've seen this happening elsewhere, such as in social interactions, so why not in Science as well?