Hacker News Reader

Comment by xfitm3 | Hacker News Reader

> L.A. requires operators to send start and stop locations of individual vehicle trips back to City Hall, within five seconds, in addition to the routes they traveled within 24 hours

Wow, this is dystopian. I support Uber fighting back on this. It's an egregious privacy violation by the city. I trust Uber with the data a lot more than the city. Let's say you take a scooter to an area where a crime happened and suddenly you're being interviewed as a suspect. No thanks.

View on HN

This is true to a certain extent but nowhere in the specification does it state that providers are required to send customer information. Each trip is associated with a foreign UUID which is linked back to the provider (scooter company). If LEO use this data for that purpose, then they would still need a warrant for that information.

https://github.com/openmobilityfoundation/mobility-data-spec...

To say that a person is “suddenly” being interviewed as a suspect just for traveling in a certain area is somewhat alarmist.

Also this is Uber we are talking about. I wouldn’t be surprised if they are using the privacy concerns as a pretext to protect their own private data mining efforts.

It's more that they don't want to be regulated.

Is there any indication it contains identifying information? Send the start and end location fuzzed by a few hundred metres and the time and it would be very difficult to deanonymize within an urban environment.

Fuzzing the locations might be useful if it's allowed but if they have to send the data within 5 seconds fuzzing the time won't do much. It's probably also not that hard to correlate the fuzzed start/end points and the full route which would be harder to fuzz effectively.

A similar release of "anonymized" NYC taxi data from 2013 shows that trip information can be pretty revealing.

https://gawker.com/the-public-nyc-taxicab-database-that-acci...

That wasn't anonymized, strictly speaking, it was (badly) hashed/encrypted, and it didn't actually reveal anything, it was pairing the data with photographs of the celebrities (and photos that included the cab license number).

And after all, out of 173 million rides some 5 (five) were "identified".

As always, your adversary isn't subject to your rules, man. I can defuzz that easily with the time granularity required by the report combined with other information sources - none of which individually break the privacy budget but all of which together can.

Yeah, agreed on time granularity. Without that I think it’d be fine.

Reply on news.ycombinator.com