Hacker News Reader

Comment by r00fus | Hacker News Reader

This says "Telcos" and then lists specific names. Do we know for sure that any telcos (my concern is US-based) who do not do this?

e.g.: T-mobile? I have less hopes for AT&T...

View on HN

I wrote this story and spent months trying to get these guys to talk. As I note in the story, "Verizon and Sprint declined to comment for this story. AT&T and T-Mobile said they don’t share consumer or location data with SAP, Sybase, AirSage or Vistar."

The data cannot for all intents and purposes be anonymized and still be useful. I have seen airsage data, they give the user a new unique identifier but it is consistent over a 30 day window. Very easy to deanonymize, the unique identifiers are only opaque to people who can't do math.

Yeah, data anonymization doesn't really work, except perhaps for highly aggregated data (eg: percentage of people out of work in a nation). It does provide a small barrier in the form that "if you don't look, you won't see" -- so it might prevent some inadvertent leaks of private information.

But it should rarely be considered a "security boundary" so to speak.

Why not? It's not hard to set up streaming filters that count how many people visit a certain mall in realtime without ever persisting the raw location stream.

Right. But when the data you have is detailed logs of individual movements, second by second, meter by meter - stepping down to "number of unique IMEIs in a 100 square km area" is moving to "highly aggregated data". If that is what is seen as anonymization, the yeah. But then you've pretty much reduced a dataset to information that answers a specific question. Typically anonymization is (AFIK/IMNHO) taken to mean taka a dataset that can answer some questions, and transform it to a dataset that can still answer some questions we haven't come up with, but avoids answering questions that lead to identifying individuals. Eg: the problem with the NYC taxi driver data:

http://arstechnica.com/tech-policy/2014/06/poorly-anonymized...

What other companies are utilizing this data? Any beyond SAP, Airsage, etc... I'm trying to do some research in this area

> I wrote this story ...

Thanks for joining us here. I'd love to hear more.

That you can get an audience on Hacker News isn't surprising, but how much interest is there from the general public? My very limited experience with non-technical people I talk to is that they don't know anything about it, don't understand the implications, and don't want to bother to figure it out. I'm hoping your much broader experience is more encouraging!

By the way, if you post something to the top level of the discussion identifying yourself, you'll probably get plenty of feedback and interest (unless it's too late for this discussion). Where you posted is buried too deep to be found by most HN readers.

Reply on news.ycombinator.com