Hacker News Reader

Comment by cmdrfred | Hacker News Reader

There should be a way to do that automagicly no? Anytime you see a grouping of digits like ###-###-#### black it out? I guess it all depends on how good the OCR software is.

View on HN

I'm not convinced. All it would take is OCR to read one of the zeroes as the letter 'O' instead, and suddenly there's your SSN out in the open. That's at least a civil lawsuit, maybe a criminal one, depending on context. (Healthcare, banking, etc. have strict laws in the US regarding the release of personal information.)

In other words, the risks avoided by paying the best heuristic computer known to man -- a human -- far outweighs the cost of the resource.

Shouldn't the office releasing the SSNs (the police department) to a member of the general public (programmers in question) be held accountable for releasing sensitive information, not the person that bought it?

Unsure how you derived that interpretation. Yes, of course the releasing agency would be the ones liable.

Nope. First the dashes may not be there. It could also be just the last 4 of the SSN (which is still bad). Also other info could be present that is not easy to identify as private. For example, say the cop was caught using extreme racial slurs against a man who was being arrested after giving his child an STD via rape. It would be extremely difficult to redact enough so that you do not release any info that could identify the victim. Even as a person who knows a thing or two about looking stuff up on the internet it would be hard to redact just enough to protect the identity of the victim while not redacting too much about the police misconduct. Doing so automatically would be a strong AI sort of problem.

It's more than just series of numbers that need to be redacted. It's why those two guys themselves have not published this data publicly:

> COP considered dumping the raw data online for anyone to view, but decided against it because it would involve the publication of sensitive personal details—things like mental illness, drug addiction, or cases involving minors.

Reply on news.ycombinator.com