2 ms·
The author doesn't care about that and says it pretty clearly. He doesn't (and I don't) think that it makes sense to think in terms of "deanonymization attacks"
by SpicyLemonZest 5y ago
The author doesn't care about that and says it pretty clearly. He doesn't (and I don't) think that it makes sense to think in terms of "deanonymization attacks" here, since your legal name, address, and basic demographic information are already semi-publicly linked together in a large variety of ways. Is it deanonymization if I go to the county registrar and pull the deed for someone's house?
- kodah 5y agoI fixed the link, sorry about that. You can deanonymize block level data and the study shows as much. You just need to have some amount of information that is either unique or low enough in distribution. There's a tutorial on Twitter if you'd like to follow: https://mobile.twitter.com/john_abowd/status/1114942180278272000 https://mobile.twitter.com/john_abowd/status/111494218027827...
- SpicyLemonZest 5y agoI see. You're right, I misunderstood what the attack vector is supposed to be here.
- giaour 5y agoThe census contains some data that commercial databases are legally prohibited from holding (e.g., info about children within a household). The deanonymization/reidentification attacks rely on joining census data against commercial datasets to produce something beyond what either contains. It's not accurate to say this information is already semi-public, because a good deal of it is not.