Pieces of a Puzzle

Gellman made it a practice always to try to deduce a source’s situation and motivation: where did the source stand in the hierarchy of the organization, and at what stage in a career? Was a source an embittered, late-career individual seeking to settles scores, or a loser in some policy debate, or someone who wanted to undermine a hated boss? But the first step in this case was to verify whether the source had any credibility at all. So Gellman devised a system. He suggested questions for Poitras to ask to which Gellman already knew 80 percent of the answer. That way, Gellman could tell whether the person knew what he or she was talking about and might even glean a little more information in the process.

The first responses were encouraging. For one thing, the source used covert program names correctly. “I noticed that when [the person] used covert terms that I knew, [the person] used them right,” says Gellman. For example, STELLARWIND was a series of warrantless surveillance programs conducted during the Bush administration. The NSA typically combined two words into one and capitalized all letters. The new word was then abbreviated into three or four letters for internal use: thus, STELLARWIND became STLW.

The source mentioned STLW, not the full word. Gellman had not used the shorthand in his book about Cheney, and he could find no reference to the abbreviation when he checked the Web. “One of the things about delving into a subject for a while is you know things that you don’t ever publish, but they can become meaningful at a time like this where you say: this person actually does seem to be displaying some inside knowledge,” says Gellman.

Poitras began passing verbatim questions and answers between Gellman and the source. To each question, the source’s answers were methodical and articulate and demonstrated a breadth of knowledge. All the while, Gellman was getting a better, bigger picture. The insider continued to use NSA terminology and, more importantly, admitted to not knowing certain things. “That’s always encouraging in a source—a source who doesn’t claim to know everything,” says Gellman. In a sign of confidence in Gellman, and even demonstrating a sense of humor, the source nicknamed him BRASSBANNER. The source in turn adopted the pseudonym “Verax,” Latin for truth teller.

Gellman and Poitras had not yet received any documents. By April, however, Gellman believed they would come and that the documents would be a “big deal.” They were most likely classified “special compartmented information,” a category above “top secret.” Under that classification, even someone with the highest security clearance could see compartmented material only on a “need to know” basis. Meanwhile, the subject was slowly coming into focus. Verax was suggesting that large private American communications companies were cooperating with the NSA, and hinted at their names. “Make a list of the top five or top 10 Internet companies, and they’ll probably be on the list,” Verax wrote.

Gellman became more specific in his questions about the documents themselves: “Who’s writing it to whom? For what purpose? What’s the subtext? What’s the context? What’s the motivation of the document writer? Is this a pitch for more budget?… Is it bragging to Congress? And how do you imagine that I could possibly authenticate that what’s said in the document is true?” Verax suggested that Gellman could find a couple of sources cleared for special compartmented information in the Federal Bureau of Investigation’s (FBI) Digital Intelligence Technology Unit. Even for a veteran national security journalist like Gellman, that would not be easy.

He also started to ask more probing personal questions: “What’s the worst thing people are going to say about you once they know what you are doing? Why are you doing this? Aren’t you worried that you’re going to be blowing secrets that could do harm to American national security?” Gellman still had no idea who the source was. “At this point, I still don’t know who [Verax] is,” Gellman says.

I want to understand more than I can tell my readers about who the source is so that I can find a way to triangulate the truth. I don’t care what anybody’s motives are. I don’t care whether they’re a good person. I don’t whether they have absolutely ill intent…. If you’re telling me something that’s true and important, I just need to figure out which part of what you’re saying is true and why it’s important.


PRISM logo from classified documents

PRISM . Finally, in mid-May, Verax sent Poitras and Gellman a single document, and soon thereafter others (which they had not expected). [18] The two reporters decided to scrutinize the first one first, a PowerPoint of 41 slides titled PRISM. It was very recent—dated April 2013. They were amazed by its classification category. By executive order, there were three classification levels: confidential, secret and top secret. This was above top secret. “I knew that this was going to be the most sensitive story I’d ever had,” recalls Gellman.

There have a been a few times in my career when I have had information that was probably as sensitive or as highly classified, but never a whole document with those stamps on it. The Washington Post had never had in its possession, I don’t believe, in its entire history a document with stamps like that on it. The Pentagon Papers were classified secret, which is actually what they used for Navy laundry instructions.

Poitras and Gellman met to read the document together. Then they read it twice more. After several hours, Poitras said she understood 10 percent of it at most. “I understood about half of it by the end of that night,” recalls Gellman, “but getting the rest of it would be hard.” Many of the slides were filled with bureaucratic jargon. While it took Gellman and Poitras time to understand the details of what they were reading, the presentation seemed to be a briefing for senior analysts in the NSA’s Signals Intelligence Directorate. The slides described how the NSA, along with the FBI, had built systems to pull information about tens of thousands of people from nine major US Internet companies. By obtaining user data from the servers where it was stored, rather than trying to catch it on the fly as it crossed the Internet, the government could extract all manner of foreign communications data, including chats, photos, email and documents, that crossed those servers.

PRISM, Gellman recognized, was unprecedented in its scope. Intelligence agencies could never before have obtained so much data so easily for the simple reason that the companies collecting it, such as Google and Facebook, had not existed until recently. The technology that had swept the world into the digital age was as useful for government surveillance as it was for businesses and individuals.


[18] Gellman declines to be specific about the dates.