Comment by kccqzy
1 day ago
An earlier article on the same blog had some very useful information on how easily the aggregated census data from 2010 (before differential privacy) could be used to reconstruct real data for individuals: https://desfontain.es/blog/us-census-reconstruction-attack.h...
I am personally convinced that the reason noise infusion was banned was because powerful people were already reconstructing individual data from census for the purpose of gerrymandering, and they wanted to continue gerrymandering.
I went down a bit of a rabbit hole on re-indentification a while back related to a research journal I was helping with and some presentations I was giving. One of the things I ran across was how easy it was to de-anonymize healthcare records from way less information than one might assume.
I really feel like the first step here should be to make deanonymization illegal. Obviously it wouldn't fix everything, but there's a bit of an implicit breach of contract if people are promised their data is anonymous, but then it's sold to someone else who breaks that, but as far as I can tell there's no law against what's pretty clearly a violation of the premise under which the data was allowed to be collected.
Or worse. It allows anyone to build really targeted data sets. Insurance companies would love such data, and many of those will use them without scruples.
Why do you need individual data for gerrymandering? Don't you only need area level?
To optimally crack a district to bias in one parties favour it is often required to literally run a boundary down a street to separate one side (close to a university, say) from the other.
Once you've table voter preferences to actual street addresses you are no longer in the realm of "broad area cumulative averages and medians".
My favorite gerrymandering story was when I learned an ~800 student local college had been split down the middle, so students were in different congressional districts depending on their dorm building.
1 reply →