Re:Oakland releases months’ worth of license plate reader data
by bunderbunder on Saturday November 28, 2015 @06:38PM
(#10643017)
They could have captured basically all the potential benefit of this dataset while more-or-less curtailing the potential for abuse with very little effort: Just give salted hashes of the license plate numbers rather than sharing them in the clear.
[ Parent
| on HN ]
Re:Oakland releases months’ worth of license plate reader data
by morisy on Saturday November 28, 2015 @08:04PM
(#10643283)
Even if the hashes are salted, if you see XXXX ends up parking outside address 123 Hickory nightly, it's really easy to figured out who it is. It's incredibly hard to release this kind of data responsibly and safely.
Re:Oakland releases months’ worth of license plate reader data
by vijayr on Sunday November 29, 2015 @08:38AM
(#10644536)
Safely or not, what is the reasoning behind releasing this data to the public at all? What can we build with this data that is useful without violating anyone's privacy?
Re:Oakland releases months’ worth of license plate reader data
by x1798DE on Saturday November 28, 2015 @07:24PM
(#10643157)
Why use a salted hash at all? Why not just randomly generated unique tokens?
That said, even with randomly generated unique tokens, I suspect knowing where someone was at two or three specific times would be enough to deanonymize them.
Re:Oakland releases months’ worth of license plate reader data
by jacquesm on Saturday November 28, 2015 @11:39PM
(#10643747)
You don't even have to know where. Given enough samples all you need to know is that they weren't in certain places (for instance, at home). Eventually the constraints will line up in such a way that only one car matches the pattern, and then you know all the other places and times too. A sort of reverse sodoku.