undefined | Better HN

0 pointsrmrfstar5y ago0 comments

Airnb and the cell provides are good examples [1],[2].

The cell provider location data is the most insidious. They add noise to it, but the central limit theorem is a real thing and people who buy the data are aware of that.

[1] https://www.nytimes.com/2020/01/15/technology/data-privacy-l...

[2] https://www.vice.com/en_us/article/nepxbz/i-gave-a-bounty-hu...

0 comments

derivativethrow5y ago

I’m aware of what the central limit theorem is, but I’m confused as to what you mean here. Do you mind clarifying? It sounds like you’re saying the CLT leads to deanonymization. I don’t see how the two are related.

rmrfstarOP5y ago

Take several measurements and average them. The variance scales like 1/n_samples, which can take you from city-block resolution to building resolution quite rapidly.

unishark5y ago

There are conditions for the central limit theorem to hold though. One can generate random noise which violates them. It's certainly hard to guarantee privacy, but it's not trivially easy to hack if they're halfway smart about it.

j / k navigate · click thread line to collapse