Facebook URLs Dataset Now Available for Academic Research | Hacker News Reader

Facebook URLs Dataset Now Available for Academic Research

socialscience.one

95 pointsmetahost25 comments

So the researchers have to sign NDAs. Great, that certainly protects us from leaks.

But what happens when the researchers are hacked or are negligent? That seems to have a nonzero possibility.

I don't think academics are necessarily experts in security.

The dataset is anonymized using differential privacy. In theory, someone looking at the data would not be able to tell if any particular individual were included.

Research on small communities might be worthless.

Seems like the worst idea from Facebook to have allowed third-parties outside of Facebook to process the raw data (1 exabyte of data!)

Why?

Privacy.

Also: https://techcrunch.com/2019/07/24/researchers-spotlight-the-...

The dataset is presumably going only to academic researchers under NDA. It's not likely that any of them are going to leak the dataset.

Why should I trust people working at universities with my private data?

Because it's necessary for academics to have real datasets in order to do quantitative research on social systems.

That's not sufficient reason to trust them.

I don’t think it’s necessary. In fact, I’d prefer if these researchers didn’t even exist.

Can you make a good case for legitimate researchers of any kind not to exist? Just because its a potentially dangerous/dismal subject matter doesn't mean it should be discarded as forbidden knowledge; if the truth destroys us then perhaps we deserve to be destroyed, better that than to live in ignorance.

Let them exist; However, their existence shouldn't be a free license to data.

A better question: why would you trust a company like Facebook with your private data?

To my mind, if you post something to a social network, and twice as much to Facebook, you should only post stuff which you won't mind posted openly on every wall in your town.

Because if done properly, employees interests are aligned with the interests of the employer.

If you own significant stock options in Facebook, it's not very wise to to cut the branch you are sitting on.

Additionally, external researchers have much higher risk, because they may seek fame, or need money (employee are generally in a safe position, the intern at a research group is not). An exclusive data leak can bring both advantages.

Also, they may not have the same security practices as internal ones (deep packet inspection, etc).

You shouldn't. But let's not make the problem worse?

Facebook collects tons of private data across the whole of the web that we don’t generally consent to giving them. They are an advertising company, after all.

lol. the risk of it becoming public has increased now that it's accessible by more people. NDAs are a deterrent, but it doesn't deter everyone.

NDA is only a legal protection, not a technical one.

It's not even a protection, I'd call it a "liability indicator," or even a detector.

You undercut your own argument by using the phrase “not likely”. You acknowledge that the probability of the data being leaked has increased as a result of more people having access.

There could be several reasons why it is "not likely" that a researcher would leak. But is the existence of an NDA really one of them?

Putting aside the likelihood of an intentional leak, how would enforcement of an NDA work in the event of an "accidental" leak.

Has anyone ever seen Facebook file lawsuits against academic researchers or their employers. When leaks have happened, has anyone seen Facebook file lawsuits to enforce an NDA in order to protect or provide compensation to Facebook users.

Of course we have all seen Facebook file endless lawsuits to protect Facebook.

What sort of "relief" are users going to get under an NDA between Facebook and an academic researcher if their data is leaked.

> What sort of "relief" are users going to get under an NDA between Facebook and an academic researcher if their data is leaked.

Keep in mind that this dataset being released to academics is literally the relief that Facebook is being forced to provide for the negative externality their users are imposing on non-users.

Even NDAs can't put cats back in bags.

Notice that they are not actually providing the raw data; they are providing access to a dataset that has been anonymized using differential privacy. This is a way of adding noise to the data that makes it impossible to tell (much better than random guessing) whether a person is in the dataset, even by combining the dataset with external data.

Reply on news.ycombinator.com