What a thoughtful post. Thank you very much for sharing it. You emphasize some very interesting points and lay them out clearly.
To return to an earlier question I posed in another thread: what is (will be) the task of selection in this new environment? Every user with whom I have discussed these issues, from young person to emeritus scholar and researcher, wants information that is "reliable". That is not surprising but of course, it raises another question: what does "reliable" mean? One professor I spoke with last night told me that he *still* tells his students never to use Wikipedia. That is simply unfair to the students (who will use it anyway of course--or at least I hope they will!) and ignores reality. My own information literacy courses take another tact: I tend to say that all information resources are OK, but you must use each one wisely. One may be a great example of a primary source, another: of biased information, another: of obsolete, scholarly information, another: of time-sensitive newspaper information, and so on. I don't have time to do much, but at least I try to treat students as adults and I don't tell them they should bow to the local censor.
But it's also clear that "easy to access" must be considered in this because of the results of the Northwestern study and the quote from Marcia Bates. One additional complexity that I have met with when I discuss these matters with former students who have finished school: we should also prepare our users to be able to handle information when they are out of school and no longer have access to our fabulously expensive databases. We should prepare them not for some ideal world, but for the real world they will encounter when they leave our institutions because otherwise, we must confess that we leave them almost helpless and completely reliant on things such as Google, when in reality, they will still have many options.
When you mention screening out the "fraudulent materials", this is what I had in mind. I think this is what people want, but "fraudulent" means different things to different people. And when you write:
... how politically aware most undergraduate students in the U.S. are, and whether they have enough philosophical or political insight to adequately detect (increasing?) themes of political bias on the web.
I would only add: it is just as important for them to see political bias in printed items and this unfortunate situation is not limited only to U.S. undergraduates. I have found graduates and full professors from around the world fall victim to this political bias. It seems that detecting different types of bias is becoming more important by the day (I think), but in my experience, they are not taught it. The idea of labelling sites as: conservative, liberal, etc. repels me as a professional, and although librarians may be pushed in that direction, I would try to leave that to crowdsourcing as much as possible. (I know that's a problem, too)
I still believe that Ross Atkinson's idea of the Control Zone could somehow be the foundation: http://tinyurl.com/33k2qtv. This paper is rather complicated and I think the basic idea can be simplified: it would mean creating a "library space" where most of the traditional library functions could be found. With today's technology, something like this could be achieved technically in a bunch of different ways, but the most difficult part is getting agreement by all the libraries involved, to coordinate selection, metadata creation and organization, etc.
This means change, and I don't know if libraries can change that much.