Lavabit founder: 'If You Knew What I Know About Email, You Might Not Use It'

[MineralMan]

38. Metadata.

If you've actually read the disclosures in the briefing documents Snowden released, it's easy to understand. The algorithms used by the NSA to examine communications metadata is about recognizing associations with known bad guys. Unless such associations appear, they're simply not interested in the content of the communication.

That's how it all works. Without algorithms designed to discard the huge majority of uninteresting communications, there are not enough people and computers on the planet to look at all of it. In actual fact only a tiny, tiny part of the communications are interesting at all to the NSA, FBI, etc. And all of those are discovered through associations revealed by analyzing the metadata, with specific targets in mind. Phone numbers. Email addresses. IP addresses. Other metadata.

What the NSA does is whiz through that metadata looking for references to actual targets. Did the target communicate with someone? Did someone communicate with the target? If so, then the NSA algorithms make note of that. One association isn't important, since the likelihood of any useful information would be extremely small. What the algorithms are looking for are multiple associations. Did an entity in the associated communications then communicate with another target? Were there multiple associations or communications over some period of time? Did two targets communicate with each other? How often?

The NSA algorithms analyze associations and provide some sort of ranking data that human analysts can look at. It flags stuff that the algorithm is designed to flag, and some actual human being then looks at those flags and makes a determination whether there's a reason to go further and find out what sort of communication is going on.

At that point, other people are involved and eventually an order is sought to actually go and collect content data associated with specific targets. FISA has to approve that collection. Then, and only then, is actual content collected for analysis.

That's what happens. It's all in the information provided by Snowden. But nobody's actually reading that information. They don't have time to bother. I read it. I'm still reading it as new information is available.

Information associated with specific targets and associations is what they're looking for. You're not one of those targets, and neither is anyone you communicate with. You're screened out in the first pass when metadata is looked at. Nobody cares about your communications. They don't matter.