Drowning in Information: NSA Revelations from 262 Spy Documents | Hacker News Reader

Drowning in Information: NSA Revelations from 262 Spy Documents

theintercept.com

71 pointsseverine12 comments

The 262 "Spy Documents" are actually from an internal newsletter called "SID Today".

While the newsletter itself is classified, my understanding is it's distributed to everyone in the agency, from HR to the security guards.

The articles that proclaim the NSA is "drowning in information" are not specific. They're not from an Inspector General or other internal review.

This is the reason why I don't fear government surveillance as much as corporate surveillance. The government simply can't process all the data they collect. And the value of obtaining information from the data can be highly variable.

Corporate data collection and analysis efforts are optimized for least cost/highest return. They really don't care about individuals as much as they do classes/tranches of people. However, the most concerning part is when corporations will pass on, willingly or unknowingly, their trade secrets to the government to improve the latter's techniques.

> The government simply can't process all the data they collect.

They don't need to. They can focus their attention on high value targets. You and I might be low value targets and safe from surveillance (for now), but that doesn't mean the surveillance of high value targets doesn't affect us. As a thought experiment, imagine a world where an unscrupulous president uses the NSA to snoop on the opposition...

Now imagine an unscrupulous NSA Director mining through years of accumulated data to retroactively snoop on US Presidential candidates' past behavior...

Heck, why stop at the top? I'm sure digging up dirt on current Senators and Congressional representatives would be useful too.

Didn't that already happen back in the 70s??

> The government simply can't process all the data they collect.

I'm not comfortable with this conclusion for two reasons:

1. If it's actually true now, it's a high-stakes gamble that this will remain true. Say someone sets up CCTV and starts recording everything – how many years can you say “no big deal, they can't watch everyone!” before computer-vision improvements means that they can, and can retroactively mine their archives as well? A new President changes policies so you're suddenly higher up the priority list than you thought and suddenly you're trying to explain why your cell-phone position data never showed your position at a church or did show you near a protest, family planning clinic, etc. If the data isn't collected in the first place, you don't have to worry about any of that.

2. Maybe the government isn't so great at building data-mining operations — the federal procurement and hiring process are definitely huge obstacles there — but they can also outsource it to someone like https://en.wikipedia.org/wiki/Palantir_Technologies who might be more effective.

> [...] or did show you near a protest, family planning clinic, etc.

I agree that people relying on the argument from gov't ineptitude are making a gamble-- and one that history of technological development tells us is incredibly risky.

But doesn't the fundamental danger depend on those activities or locations being viewed by society as suspicious in the first place? Also, doesn't the contingent danger of wide-surveillance thrive mainly off of the asymmetry of access to its inferences? Isn't an activist who speaks publicly about their struggle with bipolar disorder and builds a strong support group inherently safer than one who only tells a wide-net surveillance database?

The problem is that in the cases where you'd most want to, you never know what could be a problem until it's too late. Say in 2016 you went to a gay friend's wedding. No problem if that end up in some vast anti-terrorism archive of social media images since it's perfectly normal but … what about a decade later when President Pence of the Republic of Gilead declares that homosexuality is a crime against God and suspected sympathisers should have rights like employment curtailed?

That's an intentionally unlikely example but it's not like there aren't plenty of historical examples of this – e.g. imagine how much more dangerous HUAC[1] would have been if they'd been able to mine every picture on Facebook using facial recognition software to build a list of everyone who'd ever been in the scene with a targeted person.

1. https://en.m.wikipedia.org/wiki/House_Un-American_Activities...

> The government simply can't process all the data they collect.

This is only a matter of time.

Only if computing power catches up to handle the rapidly increasing rate at which data is produced. But as computing power increases, that also affects the rate at which data is produced. So, I firmly believe most data will sit around unprocessed and eventually decay into uselessness.

The problem is when someone else gets ahold of that data who has the time and patience to mine it.

Yeah, and what happens when someone does steal unanalyzed data? You don't even know the damage. And then you have to decide, do we analyze what was stolen to see the effect, and lose real time analysis, or keep going with the real time and just wonder about what's coming from the breach.

Reply on news.ycombinator.com