If you haven’t already, check out Siva Vaidhyanathan‘s excellent Chronicle of Higher Ed piece on privacy and surveillance: a review of several new books treating various aspects of the topic, but a great all-around thought piece. A taste:

Certainly the Stasi in East Germany exploited the controlling power generated from public knowledge of constant surveillance and the potential for brutal punishment for thought crimes. But that is not our environment in the United States. Basically, the Panopticon must be visible and ubiquitous, or it cannot influence behavior as Bentham and Foucault assumed it would.
…what we have at work in America today is the opposite of a Panopticon: what has been called a “Nonopticon” (for lack of a better word). The Nonopticon describes a state of being watched without knowing it, or at least the extent of it. The most pervasive surveillance does not reveal itself or remains completely clandestine (barring leaks to The New York Times). We don’t know all the ways we are being recorded or profiled. We are not supposed to understand that we are the product of marketers as much as we are the market. And we are not supposed to consider the extent to which the state tracks our behavior and considers us all suspects in crimes yet to be imagined, let alone committed.
In fact, companies like ChoicePoint, Facebook, Google, and Amazon.com want us to relax and be ourselves. They have an interest in exploiting niches that our consumer choices generate. They are devoted to tracking our eccentricities because they understand that the ways we set ourselves apart from the mass are the things about which we are most passionate. Our passions, predilections, fancies, and fetishes are what we are likely to spend our surplus cash on.

And so these concerns extend to the realm of online reading. With networked texts, a book (or whatever other document form) may be reading you while you’re reading it. This creates a major ethical quandary for libraries of course, who, to take advantage of social networking, collaborative filtering and other powerful affordances of digital technologies must radically revise their traditional stance on privacy: i.e. retain as little user data as possible.