That said I do think agencies like NIST should define anonymization standards.
That said I do think agencies like NIST should define anonymization standards.
Huh that is pretty solid point, so anonymization is useless to those who are the most interested in privacy?
turns out that in this regard, everyone is special
https://arstechnica.com/tech-policy/2009/09/your-secrets-liv...
"At the time GIC released the data, William Weld, then Governor of Massachusetts, assured the public that GIC had protected patient privacy by deleting identifiers. In response, then-graduate student Sweeney started hunting for the Governor’s hospital records in the GIC data. She knew that Governor Weld resided in Cambridge, Massachusetts, a city of 54,000 residents and seven ZIP codes. For twenty dollars, she purchased the complete voter rolls from the city of Cambridge, a database containing, among other things, the name, address, ZIP code, birth date, and sex of every voter. By combining this data with the GIC records, Sweeney found Governor Weld with ease. Only six people in Cambridge shared his birth date, only three of them men, and of them, only he lived in his ZIP code. In a theatrical flourish, Dr. Sweeney sent the Governor’s health records (which included diagnoses and prescriptions) to his office."
This same article also mentions one of her more famous findings too: "in 2000, she showed that 87 percent of all Americans could be uniquely identified using only three bits of information: ZIP code, birthdate, and sex."
And at that point you’re starting to destroy the value of the data.
Assuming you are looking at someone between the ages of 1-80, knowing birthdate further filters in just 1÷(80 * 365) of the sample space.
Since they're 42000 ZIP codes in the US, knowing the ZIP code lets you filter in just 1÷42000 of the sample space. Together, 1÷2 x 1÷(80*365) x 1÷42000 = 0.00000041%
With these three datapoints, you can identify roughly 1.3 US persons (assuming a US population of 330M). Not too bad, imho.
Which they aren’t. At all.
Given this, the only real issue is ZIP codes. If we assume that we know nothing about how populations are distributed across ZIP codes, given just the gender and date of birth, we can narrow down the cohort to just 5650 US persons (330M x 1/2 x 1/(80x365)).
According to this link - https://www.johndcook.com/blog/2019/08/21/zip-code-populatio... - 80% of the US population lives in 27% of her ZIP codes.
Assuming your target individual is in the 80%, given the gender, birthdate, and ZIP code, you can narrow down to the following - 0.8x330M x 1/2 x 1/(80x365) x 1/(0.27x42000) = 0.4 US persons per ZIP code.
Basically, these three data points can almost certainly uniquely identify specific individuals - the only remaining thing is to connect a name/phone number to each individual.
Im just nitpicking your weirdly precise results of your fermi math. It would be easier to grab this from the census data, right? 40 year old males with a given birthdate, no zip code, narrows to ~7,154.
that basically means not trusting… anyone?