Privacy and the Google Books Settlement

[now 15 min. late]
11:00–12:30
Chris Hoofnagle (m), Director of Information Privacy Programs, UCB (Boalt)
Tom Leonard, University Librarian, UCB
Angela Maycock, Office for Intellectual Freedom, American Library Assoc.
Jason Schultz, Samuelson Law, Tech & Public Policy Clinic, UCB (Boalt)
Michael Zimmer, Asst. Professor of Information Studies, U of Wisc.–Milwaukee

CH: Google often says that privacy is important and that they’ve thought about it + security; journalists ought to press them a bit, since the details of these thoughts are what really matter. Privacy by design requires early intervention—now is the time.

====

AM: Settlement as it exists now is silent about user privacy, in contrast to some other issues, e.g. security protections for digital books themselves. Let’s start with core questions: why do librarians care, how does loss of privacy affect readers, what are implications for digital books and GBooks Settlement in particular.

Privacy = right to engage in open inquiry w/o having object of your interest scrutinized by others.

“Libraries are cornerstones of democracy.” Think carefully to avoid chilling effects.

Some challenges to reader privacy: AM’s office provides confidential support to libraries where challenges have taken place, ergo few public records of these exist. Debunk the “one shouldn’t have anything to hide” myth.

Wise County, TX: DA request for phone #s of everyone who’d checked out books on childbirth for prev nine months, while investigating a child abandonment case. Library challenged subpoena, case fell through. Supreme Ct of CO said law officials couldn’t justify getting list of book purchases from Tattered Cover in Denver (drug mfg).

Running out of time! ALA hasn’t opposed settlement; filed comments prior to orig 4 May deadline to request vigorous oversight by court, re: reader privacy. Google have been great listeners, but statements so far re: privacy have been informal, and they need to be codified into policy (accountability).

====

TL: privacy pre-GBooks Settlement and post-. Research libraries are less exposed to concerns and public feeling than public libraries; issues overlap, but different. Public-lib users are assumed to have genuine personal interest that’s not subject even to the library user standing behind them in the checkout line, whereas research users are completing an assignment and tend more to make their reading choices public or semi-public. Core value shared = protecting anonymity; no monitoring w/o good reason and with user’s knowledge that it’s being done. (Policies are visible: networking. Rare materials are different, and justification = protecting the collection; cf. Smiley.)

Other cases: would we want there to be no records of map use when investigating the pillaging of Native American grave sites?

If a library of the future is to develop neat techniques for helping users discover things, we might want to collect personal info so that we could show you related works you might want: tradeoff.

Privacy post-GBS: where is Google commitment equivalent to what’s seen now in library policies? Minute monitoring is not practical at this point, perhaps even for Google. Thus TL believes Google—but would like to believe them more, once they’ve articulated clearer reassurance. CDT recommendation, frex.

Google reminds TL of the UCB–BP deal. When you have big pockets and can sued by any judicial system on the planet, you’re wary of saying too much. When you say too little, you enjoy BP’s reputation. TL encourages Google to continue on with its good start. [But basically, it’s back to DON’T BE EVIL. heh.]

====

JS: how many are familiar with Stanford marshmallow experiment? (Take some 4yo kids, put a marshmallow in front of them; I’ll go away for 20 min., and if you eat the marshmallow, the end, but if you don’t, I’ll give you a second one.) How do we wait, how good are we at waiting…. How good a marshmallow will this be, and do we believe in 20 minutes?

Previously this was all about copyright protections and fair use. [Outline of what GBooks offers: scanning gives access to book interior, and you get a snippet if it’s under copyright.] Also, decent way of addressing orphaned works: at least some access to at least some info within these works, even if we can’t determine copyrt status.

Now we’re in a different situation, which is how privacy came up. Four main concerns have emerged:

• size of deal (thus feels different from others)—largest copyrt deal in US history
• compulsory nature of settlement (class includes anyone who has a copyrt interest within the time period, and binding on everyone unless you opt out even if you’ve never heard of it)—thus different from other Google services. Grants Google a license to do something, and in return you get certain benefits; but shouldn’t one benefit be privacy? Can offer Gmail, e.g, w/o license or permission, but here they need permission in order to offer GBooks service. Should we give Google permission?
• implications of privacy—with this many people + books involved, how much privacy do we mean?
• via class-action mechanism, Google wants to use legal system to address some problems it has (incl. orphaned works), which makes this a legal hack. Not necessarily bad, but worth considering

Governments and law enforcement try to get info about suspicious individuals / entities. Will Google be better or worse about this? There’s an assessment re: whether to turn over the data.

Want to end by pointing out two models Google has. In Google Health there’s a commitment that your records will be kept separately from all your other Google records. Also, Google location-based products have a sort of amnesia built in: says officially that it doesn’t remember where you were, just your current location. Could that be applied to books (if info is kept, done only minimally)?

How will we know whether Google has held to its policies?

====

MZ: looks in his work at intersections of ethics/technology, privacy/surveillance, Web/new media, info/society…. Acknowledged that other panelists had outlined the issues well already (minimize repetition).

Vague general public concern that Google Is Watching You. We try to get them to make their policies and processes more transparent.

Will we carry the norms of libraries or the norms of web-searching into GBooks?

Competition in privacy policies beginning for search engines (Lance Cottrell), but given [effective] monopoly presented by GBooks, not clear that we’ll get similar competition here.

Consider Google Street View: arg that it’s already public, so there’s no big deal; cf. photos taken by a journalist walking down the street, w/o asking permission of anyone. In Canada and the EU there are stricter privacy provisions, so Street View had to be released with blurred faces. Then perhaps we should blur faces everywhere, ethically speaking….

One nearby idea: build in anonymity (can I have a login for GBooks that’s separate from regular Google personal access). Google Health policy mentioned again. Another: TrackMeNot Firefox extension, which pollutes your browsing trail; perhaps we need something similar for GBooks so it’s harder to sniff what a user was reading at a given moment.

====

audience (something rights lawyer): should we worry about watermarks?

MZ: anything that will go with me when I download a page (or book) is a potential concern.

JS: need to extrapolate out: if a Google service or product changes, it shouldn’t permit Google to change its policy to allow a greater privacy breach later.

TL: can’t think of a way that one user utilizing a particular electronic resource via campus could be traced lastingly.

AM: libraries work with a variety of vendors and need to protect info from each.

TL: core values versus core experience: unaware of chilling effect from the fact that one could probably retrieve a record of everyone who’d used Mark Twain Papers collection over decades. [This is true for MTP, at least: it doesn’t disincline readers from visiting us.]

audience: which paradigm—context of which level of commitment we’d like Google to make, and enforceability of commitment. How might we make things enforceable over time?

MZ: tradeoffs discussed earlier: collecting data in order to customize what’s offered. Do we need to loosen up on core values in order to remain viable and competitive?

AM: if I didn’t finish a book I checked out from Chicago Public six months ago and ask them what it was, they won’t/can’t tell me. What would “loosening” mean for scholarship, if people are directed to “more like this” links? Too obvious via algorithm, versus really interesting and groundbreaking research.

JS: Google privacy policy has an Enforcement heading, but it says to contact them and have a chat, basically. If they’re honest, that’s great; if not, what are options for individual? What happens if they start doing everything we don’t like? The enormity of the permissions required to float GBooks is an opportunity to get a more explicit policy that would bind them legally (contractually).

MH: what should constraints be? Has someone articulated a standard that would satisfy the civil-liberties community?

JS: Tattered Cover, Kenneth Starr wanting Monica Lewinsky’s reading list—there is a process, since can take request to court and either a warrant is granted or a judge denies. Need to show compelling need to get the information, and then judge would order Google to turn it over; but if there’s no challenge, then there’s no accountability or judicial review.

Dan Clancy (aud): re: privacy, is very cautious, but wanted to mention—one Street View issue is that its “product” segments can’t be discussed by public until after release. If they’d put privacy into the GBooks settlement, it would’ve been wrong because it wouldn’t have had review/discussion by others outside of Google yet. Public discussion improves policy. // You buy a book before getting married, and don’t want new spouse to be able to see it or record of its purchase; how do you throw away data from the cloud? Thus big issue w/ privacy has put spotlight on complex issue that’s larger than GBooks settlement.

Eli Edwards (aud): EU has different privacy policies; how might we devise one for the US? [not sure I have that right]

MZ, MH: we’ll see some behavioral targeting.

Hal Verian (aud): works at Google but in different area. Library access q: fixed IP address, delete cookies, no sign-in; how would anyone track?

DC (aud): not everyone travels to physical library to do research.

Berkeley Public Lib employee (aud): would GBooks access be on a single computer at a given library? If you ask for one traceable IP, that’s suspect.

DC: postpone to afternoon.

? Boyd, Media Bistro (aud): if given no more time and had to accept/reject settlement as is, what would you do?

JS: marshmallow 🙂

TL: we do have an alternative way to get to some of that content.

AM: as ALA rep, ALA and ARL and ACRL support settlement even w/o privacy firmness, and depend on a judge to use keen oversight.

JS: it’s not all or nothing. These things are modified all the time, often in context of lawyers asking for too much money: judge rejects to produce something more proportional, frex.

MZ: ditto Jason. Would want to find different way to frame what settlement does, focusing more on public interest side. Standards and principles that don’t rely on conventions of a major corporation’s practices. Would consider opting out, personally, as rights-holder.