Hacker News Reader

Comment by makomk | Hacker News Reader

This effectively provided carriers with aggregated, deidentified information about people's locations. It's easy to see from the problems with things like Strava's heatmaps just how much of a privacy headache this could be and how hard it would be to ensure that nothing sensitive leaked out. Also, even if they did people are sensitive enough about privacy that with the right spin it could be turned into a muck-raking story and readers would believe it.

View on HN

I was about to suggest this, because anonymised data is often not-so-anonymous in aggregate. AT&T's recent disclosure of selling this information to bounty hunters[0] shows that sharing across companies isn't always done with the end goal of user privacy. Examples of over sharing can set precedents in court cases showing that such sensitive data can be treated as a liability rather than an appreciating asset in the wrong hands.

[0] https://www.eff.org/press/releases/eff-sues-att-data-aggrega...

If you're a carrier then you can do coarse-grained detection of everyone's location by timing-based triangulation - modern cell phone standards require knowing your distance from the cell tower (see https://en.wikipedia.org/wiki/Timing_advance for example) so it's measured (and thus can be recorded if the carrier wants) as part of every keepalive connection and if you're in range for more than one tower, which is almost always, you can do triangulation.

Of course, getting GPS data will be more accurate, but you should assume that it's unavoidable that the carrier can know roughly where all the phones are right now.

I wager that coarse-grain location data is not so coarse these days. My carrier seems to know the instant I crossed a border for instance, sending me a text message about it within seconds. Years ago people used to get nailed with roaming charges simply because they were near a border and happened to accidentally get connected to a tower across the border, but that's not happening anymore. Their system knows enough about your location to know if you are across the border or just near it.

Coarse-grain location data becomes very revealing when you have enough timestamped data points. The NSA's CO-TRAVELER[1] program only needs to watch which towers cell phones use to build a very accurate map of each phone's pattern-of-life (where does a phone regularly travel/idle) and social interaction network (e.g. which phones travel/idle together).

[1] https://www.washingtonpost.com/apps/g/page/world/how-the-nsa...

By course-grained I mean 0.5-1 km; which is very inaccurate compared to GPS accuracy (and isn't sufficient to tell which shop you're visiting, which has commercial value in sale of location data) but is accurate enough to tell which side of the border you're on, or which mall you're visiting, or when you're home or at work and when somewhere else.

> Distance

Don’t modern cell towers use beamforming? In which case they get direction as well as RTT, so triangulation is not even needed (although it makes location more precise)

I may be wrong, but I think that at least the vast majority current widely depoyed systems don't use beamforming, it seems more of a "here's a tech demo that we're going to roll out soon with 5G" thing.

However, it's a bit the opposite way around - it's not that they can use beamforming to get direction, but it's that they need to know the location before they can use beamforming - so the protocol (5g?) needs to require the device to know or obtain its location, and send it to the cell tower so that it can apply beamforming; similar to how in 3g the protocol includes measurements and adjustments of the timing advance so that the distance can be used when actually transmitting data. So it would be essentially like "the phone can't turn off GPS or transmitting it to everyone, because it needs it for the radio to work".

If they're so concerned about peoples' privacy, perhaps they could disable their own non-aggregated, individually recorded, indefinitely-retained, fine-grained, timestamped GPS data collection?

Of course, their definition of 'privacy' is 'making sure nobody but Google gets your data' so they wouldn't see any need to do so.

One could also argue that it also provided law enforcement with another potential target for blanket, large scale search warrants.

Strava heatmaps? I seem to recall the problem that came up there was exposure of military bases?

Google could totally do this properly if it really wanted to.

Reply on news.ycombinator.com