Brad Templeton from the EFF organized this talk, and he’s currently speaking:
Brad: The fourth amendment does not apply to documents held by a third party. Important not just in criminal but in civil cases. The process of discovery allows your lawyers to see what’s being given out–not so if documents are taken directly from the third party. Not “data portability” but “easy export of my data”.
Time-travelling robots from the future: Improved AI, visual and textual recognition, capable of reaching into the stored past and interpreting, indexing, and making it available.
Paradox: Ease of use can be a bug. Who doesn’t want ease of use? Well, when it’s exporting everything about you, maybe you don’t want it to be easy. Suppose you were asked to enter much information about yourself in order to join a site. You’d say, “No way.” But suppose it could be done through your portable data–much easier to do, much more common to ask for.
Data portability and single sign-on move the bar for what is much easier to ask from the user. How many of you use Facebook? Consider their “applications”: Essentially a sub-website run by a third party. There is a check box when you install, which says, “Let this application access my data.” Have you every tried to uncheck that box? A lesson in one-click data export.
Second paradox: These applications allow the user to control what happens to their data. Now, data portability arose from Microsoft Passport, then the Liberty Alliance, now OpenID. When you click to join a website, you either accept the term or go away. No negotiation as in a contract agreement. The process of user control precludes the ability to negotiate.
Perversely, a Microsoft as the broker in this transaction would’ve allowed negotiation on their part in the interests as users. Some organizations, such as AAA, can negotiate in this manner. Perhaps Consumer Reports, something of that nature could.
Audience: British government data loss: When information requested on an individual, it used to be unitary. Now being reworked for more granularity.
Brad: How to solve this problem. It is not going away, so it must be dealt with. Many useful things which can be done with data portability. Development of a layer–trusted partner–for hosting data (identity). An application wants to email all our friends in the Bay Area. That application gets to send the email without seeing any more data about them. Difficult–RSVPs complicate that.
Guy from Disney: We don’t allow our data out. Used internally. Do our own mailings.
How to have all the flexibility in that layer to give the functionality people want? Possibly use of code sandbox to see what is done with your data. You could, perhaps, be your own trusted host for an interaction with (for instance) Disney.
Me: Is this self-hosted, identity providers, or identity service through a third party such as an ISP or a hosting company. Answer: Yes, all three.
Don’t really want people to do their own hosting–some form of group negotiation required.
Privacy policies: All I’ve seen (that I didn’t work on) have said, “If we wish to change this at any time, we can do so without notifying you.” Not just policy, but also in terms of service.
Me: Why this not a subject for regulation? HIPAA and FERPA have been very effective. (No answer.)
Q: Why not make it easy to get data, but not effective to use it? For instance, watermarking for e-mail. A: I do that–use a different e-mail address for everything.
And now the answer for regulation: If that data gets out, regulation doesn’t stop that. It’s better to keep it from getting out. (I don’t find this convincing. If data gets out with or without regulation, it’s hard to get back either way, and data which is or is not regulated can still get out either way. The question is whether regulation makes it harder or easier to keep the data in the first place.)
Some discussion on the relative disastrousness of the original British privacy release, a bulk release of about twenty million names, versus the twenty percent of that original release whose information was sent to an old address. Who is living at that address? Possibly an ex-spouse–this database was about children with single parents. No clear answer to this–good arguments on both sides.
Again, an unconvincing answer on how mixing data on which current state is all that’s held and data in which history is kept and how that can be done with change data capture.
More concerned about people being evil on accident, rather than on purpose.
Q: How do you keep someone (like Facebook) from doing this? A: Competition or social pressure. Gives Google as an example, but notes they also had legal concerns. Also a balance between ease of entry (small companies want it, established companies don’t) and user needs.
I hesitate to generalize from this tiny sample, but I’m continuing to find EFF’s disdain for regulatory action unproductive. It’s a valuable session and I’m enjoying it, but there’s the point on which I’m in disagreement with what they’re doing–valuable work, but lacking on this point.
I did not expect to get into a discussion of Oliver Sipple and Dan White during this discussion, but it’s an interesting ending.