The NSA: Too big to not fail

Lest anyone get the impression that the National Security Agency has time to do anything but aggressively violate Americans' rights, the Guardian fronted a story about a draft memorandum of understanding between the NSA and Israel's signals intelligence agency. Notwithstanding the context, or a close reading of what the MOU actually permits, which you can read about here, the story is useful because it points to one of the reasons why the NSA has a lot of trouble figuring out just what the hell it is doing with all of its nodes and devices and satellites and fiber lines and servers.

With every NSA document dump, as Benjamin Wittes points out, you can read it as a case of an agency struggling with technological problems, identifying its own mistakes, conceding them, and rectifying them — or evidence of a continued, deliberate, unquenchable thirst to do what the mega-surveillance leviathans of fiction and philosophy are supposed to do. The NSA's history allows for both readings. To read is to interpret, is what Peter Gomes always said about the bible.

I hate to bring September 11 into subjects unnecessarily, but it is crucially important to try and understand why the NSA is doing what it is doing and why it seems to the lay person to make a heck of a lot of rather large mistakes, like not understanding what it is actually collecting.

Subscribe to The Week

Escape your echo chamber. Get the facts behind the news, plus analysis from multiple perspectives.

SUBSCRIBE & SAVE

Sign up for The Week's Free Newsletters

From our morning news briefing to a weekly Good News Newsletter, get the best of The Week delivered directly to your inbox.

From our morning news briefing to a weekly Good News Newsletter, get the best of The Week delivered directly to your inbox.

Sign up

Before 9/11, the NSA collected on foreign targets. Virtually every system they built was set up to collect on foreign targets. The technology of collection advanced at such a quick pace that the NSA's minimization technology began to fail in such a way that the people who manage the NSA noticed.

But little heed was paid because nothing was intentional and over-collection was simply not a big deal. NSA simply could not care, and did not care, about communication that either emanated inside the U.S. or transited it. They gathered intelligence on foreign governments in foreign lands, on terrorist entities overseas, on nuclear proliferated wherever they hid. When it was necessary to connect a dot inside the U.S., NSA asked for a FISA order and that was that.

Then comes 9/11. NSA is caught off guard. The demand for SIGINT about terrorist cells inside the United States goes from priority 54 to priority 1 overnight. The NSA, unlawfully, but with the president's consent, scrambled to fix this national security emergency. It twice reorganized itself. It purchased tens of billions of dollars worth of new equipment. Congress threw money at NSA, which means that a lot of untested technology became a part of the NSA's infrastructure. The NSA built stuff quickly. Think of those rapacious zombies climbing over the wall in World War Z. It is, like an large organization, unable to be fully transparent to itself. Unlike other large organizations, it revels, it thrives, on a culture of secrecy.

Then come the leaks to the New York Times, and suddenly, more court scrutiny, and congressional scrutiny. The FISA court's judges had to become SIGINT experts overnight. That's impossible. They are experts on the law. They can't become anything else overnight. So they relied on NSA's own representations of what was going on. And NSA has its own language. "Selectors." "SCTs." "TOPI." NSA lawyers understand it, but others outside the government have to be educated about it. Since the process is secret, judges cannot consult other technical experts to see if something the NSA says makes sense. Congress is a whole other story. Some in Congress knew about SIGINT, but they didn't know SIGINT, because there was never a reason to exercise oversight to closely before. Lots of actors in this play are speaking different dialects and trying to understand each other.

Back to the Israel story. The NSA's main counter-terrorism analysis is conducted by the fine folks who work in an office known internally as S2I. S2I, according to the MOU, is the main provider of SIGINT; it decides what Israel gets to see, and asks the NSA's technical folks to aim that beam correctly. WHILE THIS IS HAPPENING, a whole other NSA branch, the folks who focus on "counter foreign intelligence," are collecting significant amounts of raw SIGINT on Israel. Why? Well, for one thing, allies don't always tell the U.S. the whole truth. But Israel has a reputation for spying on the U.S., for political and economic gain. The NSA helps the FBI and CIA counter that spying. So — JUST WITHIN the S2I division of the Signals Intelligence Directorate, there are analysts who are giving Israel secrets and preventing Israel from getting other secrets. Working this out is really complex. The two groups are segregated to prevent conflicts of interest, and the U.S. of course classifies as TOP SECRET//SI//NOFORN the fact that it labels Israel a counter-intelligence threat. NOFORN means that Israel won't see that unless it steals it or Edward Snowden leaks it. On top of these segregated units, a whole other group of NSA analysts decide which selectors to give to Israel, and which to withhold.

With this in mind, think about the technological challenges inherent in creating an unprecedented architecture for collecting intelligence on targets inside the United States while adhering to the letter of the law, which knows SIGINT from a whole in the ground, and constantly shifting expectations from the courts and Congress about what the law means.

It is not surprising at all, really, that the NSA's biggest institutional problem, the one that is responsible for virtually every single one of the violations that we know about it, is that the NSA simply was not able to understand what it was doing. It could not build, and did not build, these systems carefully enough because, well, there was a national emergency and it was all secret. I don't like to overuse this term, but think of signals intelligence collection after 9/11 as an emergent property, greater than the sum of its parts. When you try to figure out it by taking it apart, it kind of falls apart. And NSA simply was not prepared for this type of existential problem. People in charge of things simply did not have enough knowledge to understand what was actually happening because it was not possible for them to do so.

The leaked documents show that NSA has gotten much, much better at being humble about its own inability to understanding everything fully. And Americans have every right to expect that the learning curve needs to be much steeper, because even the hardiest of NSA defenders recognize that tangible values are at stake.

One good thing to come out of all of this might be that NSA has to pause a bit. Slow down. Understand everything better. Then continue, with new restraints, and with a public and Congress that understand more fully what's going on.