HOW WIDE A NET?….Here’s a bit of thinking out loud about the NSA spying case. Don’t take it too seriously; it’s more barroom conversation than anything else.

First, Sifu Tweety argues that the NSA program wasn’t quite as “targeted and focused” as General Hayden suggested on Monday:

Data mining WAS used, but it was used for target selection. People have been talking about ?data mining? like it?s a be-all-end-all surveillance technique. It?s not. All data mining (or pattern analysis, or whatever) is going to give you is a list of potential targets.

OK, but what kind of data mining? Surely even the NSA’s supercomputers can’t literally track and transcribe every voice call coming in and out of the United States? A reader emails an alternate possibility:

The phone network really is two parallel systems: one that switches voice signals from user to user and a second “common channel” that passes control information between the switching systems. Once the number you dial reaches your local central office, it leaves the voice system and passes from switch to switch via the common channel.

You can see where I’m headed with this. What about monitoring the common
channel?….Why do this? Let’s posit that we have a list of known bad-guy telephone numbers. In the first tier around them, we collect the phone numbers of people who have called the bad-guy telephone numbers. Probably most of these folks are bad guys: cell leaders, fund-raisers, sleepers, etc. In the second tier, we find people who have called the probable bad guys. Most of these probably aren’t bad, but some small fraction are the operatives who fly planes, strap on explosive belts, and so on. Et cetera as far down as you have time to go.

This strikes me as plausible, and it also explains why NSA couldn’t get wiretap warrants: this kind of analysis isn’t even within yelling distance of “probable cause.” If you capture Osama’s cell phone, it’s one thing to ask for a wiretap on all the numbers you find in his speed dial, but it’s quite another thing to ask for wiretaps based strictly on a once or twice-removed traffic analysis of the phone numbers dialed by anyone who is “a member of al Qaeda, affiliated with al Qaeda, or a member of an organization affiliated with al Qaeda, or working in support of al Qaeda.” That’s a very wide net.

Which brings up another knotty question: just how wide is this net? Who’s likely to have called someone who’s called someone who’s suspected of being affiliated with al-Qaeda? I’d guess that this description applies to vast numbers of U.S. Muslims. In other words, the NSA program might have been nothing more than a thinly veiled excuse to identify a huge pool of American Muslims from which it could then pick and choose suspects it had wanted to track all along but otherwise had no justification for tracking. And this might have been so transparent that no judge would ever approve it.

As I said at the beginning, this is just thinking out loud and might be way off base. Comments are welcome.