The Guardian's NSA files have awakened many a curiosity about the actual technological capacity of the government. What it does do is a most important question; what it can do is only slightly less germane.
Still, there's a lot we don't know about some basic questions. For example, is it true, as Edward Snowden boasted, that an analyst can "wiretap" anyone simply because he or she chooses to do so?
Here's the basic gist of an answer:
The NSA has the capability to wiretap anyone it targets. It does not have the immediate capability to target Americans at will, but it does have the capability to change capabilities — to a point — to allow it to actually wiretap any American at will.
BUT — individual NSA analysts can wiretap anyone they want if the data they're accessing in real-time includes the content of everyone's phone calls or emails.
Individual NSA analysts cannot wiretap anyone in the United States — that is, actively listen in on calls — by typing in a phone number and simply pressing a button. When I say "cannot," I don't mean "would not" — I mean they technologically cannot do this.
To square the circle: If NSA, as an organization, wanted to give its analysts this capability, it could — but it would require a significant technological renovation and, obviously, willful total lawbreakery.
Right now, the digital content of my phone fixes on a cell tower, which transmits its metadata to a central base station close by; once the base station determines that the incoming number is legitimate, the base station routes the digital signal to the tower of the intended target. That target's phone, if turned on, will be interrogated for its location, which the carrier of the incoming call can ask for. Most of the communication takes place underground; the cell towers accept your call and route it using fiber wires to the local exchange. Federal law forces service providers to make it easy for law enforcement agencies to intercept anything or everything that flows through the service.
That said, the phone company won't give anything to the police without at least a request; they can get call records pretty easily for any number, but a warrant is needed for the phone company to use their "intercept access points" to transmit all the data that includes the identifying information coming from my phone.
If the FBI wants to tap my phone calls, it must obtain a warrant; it must send the phone company the warrant; the phone company will then open the data point to the FBI. Different companies use different systems to do this; most of them cannot handle thousands of content streams simultaneously.
So: If you're sitting at a desk in NSA Hawaii, how would be able to wiretap any phone, instantly?
You would have access to the raw data that the NSA gets from major telecom switches, except that that data is already filtered by the time it hits any server you'd look at; most domestic-to-domestic calls would be filtered out before it even reached the NSA, while others would be identified automatically by the NSA back end system and discarded.
As an analyst, you have live access only to the subscribers or selectors that NSA was already targeting. (Again, I'm writing about live content of calls, not metadata, which as we now know, NSA can query.)
So you could come up with a fake reason and fabricate some pretext to get your supervisors at NSA to pass along a tip to the FBI, which could ask the phone company to send all calls for 48 hours (subject to an emergency certification) without a warrant. But the FBI would have the lead. And a least a dozen people, probably more, would have to at least observe your request, and some would have to certify it and act on it, before you'd get to listen to the data. This process is not instantaneous, it is subject to real-time review because your superior would be aware, as would the section chief, the product line chief, the NSA's legal counsel, the FBI's legal counsel, the National Security Division of the Justice Department, and others.
Without a doubt, a large number of American phone calls do pass through NSA servers. But it is a large number of a fraction of the total. The chances of your live real-time radio waves being among those passing through the NSA servers and subject to the type of filtering that would allow an analyst to instantaneously intercept them is very small. If you receive a call from cell sites in Pakistan that militants tend to use, I have no doubt that the NSA analyst could listen to your end, the domestic end, of a conversation.
Most likely, however, the call would be technologically and automatically minimized, or should be; the NSA would have to establish the "reasonable articulatable suspicion" standard for your end of the call, at which point, yes, the analyst could listen to what was recorded.
I assume that many innocent Pakistan-Americans have been monitored in this way, at least for a minute or so. We know that innocent American soldiers, using the cell phone network that the NSA basically built for Afghanistan, found their calls accidentally monitored by NSA analysts in Ft. Gordon Georgia because NSA could not filter the Afghanistan telephony stuff well. This is disturbing, and I hope we learn a lot more about the NSA's over-collection.
In order for an analyst in Hawaii to have fingertip access to the next telephone call that I make, the NSA would have to reconfigure dozens of switches, servers, software, and data hubs. It would have to get rid of a lot of the foreign content it monitors; it would have to find a way to tap into the FBI's direct access to the carriers, assuming that the carriers could provide the FBI with everything passing through its entire system in real-time. The NSA has not done this yet, and so the answer to the question is, at this point, a "no."
I don't see how it would be possible for one person to have eyes-on access to everything, always, or even a tiny fraction of it. Because the data is in its raw form here, it is unprocessed and probably more difficult to audit access to, although in light of Snowden's own actions, someone should get to inventing a solution real quickly.
Still: How would Snowden query the raw data, assuming he could broaden the aperture? How could he translate the digital signals into stuff he could listen to or record? How would he get stuff that the NSA does not collect? Would he have access to the software and hardware that ran all the switches in the U.S.? Could he write a program that would allow him to somehow bring all of this together?
If the answer is yes, then his claim is still frivolous in this sense: Any police officer could spend months concocting a case against someone, making it up out of whole cloth. Are there federal agents with unsupervised access to DNA and fingerprint databases? Probably. Could some FBI tech working on those databases insert malicious code to switch someone's records? Probably. But there will always be, in these scenarios, a person who must have total and complete access to the entire system, someone who could disable all of the auditing systems that kick in sequentially. How do you audit the auditors? How do you audit the auditors auditing the auditors?
So in the grand scheme of the NSA files' revelations, I would suggest that we not obsess so much about this question. At-will wiretapping is not the problem.