Elizabeth A. Jones and Joseph W. Janes of the Information School at the University of Washington have just published Anonymity in a World of Digital Books: Google Books, Privacy, and the Freedom to Read in Policy & Internet. It is the most careful and sustained analysis to date of the privacy issues surrounding the proposed settlement, as well as being an absolutely crackerjack example of how to apply Helen Nissenbaum’s contextual integrity theory of privacy to a specific problem. Here is the abstract:

With its Books project, Google has made an unprecedented effort to aggregate a comprehensive public-access collection of the world’s books. If successful, Google’s collection would become the world’s largest and most broadly accessible public book collection—indeed, project leaders have frequently spoken of their desire to create a “universal library” (Toobin 2007). Still, the Google “library” would differ from established contexts for the provision of free, public access to reading materials—like public libraries—along several policy-related dimensions, of which perhaps the most glaring is its treatment of reader privacy. This paper teases out the specific differences in reader privacy protections between the American public library and Google Books, and what those differences might mean for the values and goals that such contexts have historically embodied. Our analysis is structured by Helen Nissenbaum’s “contextual integrity decision heuristic” (2009), which focuses on revealing changes in informational norms and transmission principles between prevailing and novel settings and practices. Based on this analysis, we recommend a two-pronged approach to alleviating the threats to reader privacy posed by Google Books: both data policy modifications within Google itself and inscription of privacy protections for online reading into federal or international law.

The article is available for free download with registration or at any institution with a site license.

Privacy as contextual integrity has troubled me since Nissenbaum proposed it, and its application to Google Books in this paper helps me put a finger on why.

What contextual integrity requires is a monologic theory of social acts, forever vexed by life’s fluidity, like a word whose utterance demands that its meaning no longer be subject to negotiation or drift. A social act under this lens would be, in Bakhtin’s words, “obliged to exhaust itself in its own single hermetic context.”

This paper carries out its obligation well. Were it to conceive of those who interact with book contents on the web as anything but public book borrowers, and Google anything but a circulation terminal, its ethical ground would expressly fall away.

Meanwhile there’s a world of actual lived context unfolding, spilling over boundaries, and opening to renegotiation, as it always does.

Let’s apply this paper’s alarmism to Google itself. Beginning with a corpus of partly new, partly old text that had already dwarfed the combined holdings of all libraries, Google violated many librarian norms: it operates for profit; it requires no membership; its front desk is a blank text field; it relies more on brute statistical signals than credentials or categories; it subordinates preservation of source works to the demands of speed and scale; at the same time it captures many small behaviours that libraries would have no conceivable use for.

Google, not libraries, has been the world’s dominant information retrieval system for about a decade now, as has the web for longer than that. So, do un-librarian digital norms result in the chilling effects described? Has Google’s popularity tracked with decreases in anonymous/pseudonymous browsing, inquiry, speech, association, production? These are questions the authors are free to measure right now rather than theorize about.

Registering another zero point in history agrees better with the framework, though. That Google Books instantly brings into focus at least two prior contexts – Google and books – is irreconcilable with contextual integrity. There is no integrity from which to begin the analysis, and so all but one of those contexts needs to go.

The Open Book Alliance on this paper: Google Book Settlement: Serving Ads, Serving Google, Not Serving Privacy