:::ahem:::

Do the police need a warrant to watch you when you leave your house and go to the market/the mall/a friend’s home/school/etc?

No. They don’t.

Understand that police work entails a lot of boring information gathering. From a pattern that citizen A has lots of different cars coming and going at all hours of the day/night and with the visitors only staying a few minutes a visit, the police - legally having said house under surveillance (no warrant needed) - THEN develope enough probable cause to seek a search warrant for the inside of the house.

Geez, I’m beginning to think that some people believe suspects are identified by calling the Psychic Hotline.

We have to distinguish between “traffic” information– the phone-numbers-time-and-duration data the NSA got from the telcos– and “wiretap” (call content) information (which clearly requires court authorization when it involves purely domestic communication). The law concerning privacy of “traffic” data is much foggier. (For example, traffic data routinely passes between telcos, for billing purposes and to investigate fraud. Also recall the flap this past January when it was revealed that average citizens were buying cell phone traffic data from brokers and using it to stalk people.) While so-called “pen register” laws limit the use of this information by domestic law enforcement, I’m not enough of a lawyer to know what restrictions apply to espionage/terrorism investigations.

To attempt to answer Andrew’s cost/benefit question, once you get the traffic data into machine-readable form (which it is already), finding the patterns is relatively easy. This post by Kim DuToit (who’s in favor of it) gives the procedure. Pull quote:

[W]hen you establish that (357) 243-3006 belong[s] to Abdul El-Bomba, who received a call from his brother Aziz, a known member of Hezbollah in Syria, you now have the ability to focus only on all the calls Abdul made and received… That would be a couple hundred calls, out of the (literally) tens of billions of records you’ve collected. Most… will be innocent… [b]ut out of the couple hundred calls, you may find five that are [suspicious]. [I]t’s NOW when you, as the investigator, [] get a warrant for a wiretap so you can start listening to actual content…

The initial traffic analysis can easily be done by computer (you could do it on your laptop, except there’s too much data). But consider that, at this stage, computer analysis means greater privacy for innocent bystanders than hand-correlation, because computers only pull out the relevant data, and don’t get idly curious when they run across records involving somebody they know.

Of course, the concern is that, now that you’ve assembled all that data, it’s easy for someone who wants to “get” Andrew Tyndall to start with Andrew’s number and go on a fishing expedition. All it takes is a couple of keystrokes.

Right now the bottleneck is listening to the wiretaps, once you have them (unless the NSA’s voice-analysis technology is a lot more effective than we know). Which means that the traffic analysis is a good thing, because it helps to concentrate our investigative resources on likely suspects.

Ed Felten has begun a series of posts examining the technical and privacy issues of all this in light of Moore’s law/increased computing power. [First] [Second] Here’s his prediction:

“Before too much longer, Moore’s Law will enable government to record every email and phone call it knows about, and to keep the recordings forever. The cost of storage will no longer be a factor. Indeed, if storage is free but analysts’ time is costly, then the cost-minimizing strategy is to record everything and sort it out later, rather than spending analyst time figuring out what to record.”

Add even rudimentary speech-recognition, and that’s what’s scary to me, which is why this discussion is a good thing.

Besides the point that Jeff raises about whether the NSA project is a wise one, a second, different, question is whether it was an accountable one. Both tests have to be passed before it warrants our approval. The ABC/WaPo poll question was lazy because it acted as if the second question does not even exist.

As for the New York Daily News, when it said this–”But when the call is part of a pattern that suggests a security risk, this is worth red-flagging and perhaps eavesdropping on - with a warrant and court supervision, as all right up to the commander in chief agree would be necessary.”–it seems to have wilfully ignored the very point in dispute, namely that the Commander in Chief had ordered the eavesdropping to proceed without the required warrant and court supervision.

On the first question–whether the anomaly crunching of such a vast database is wise–a relevant consideration would be a cost/benefit analysis. There are trillions of calls being studied. Can any BuzzMachine reader with a knowledge of computer costs enlighten us about the order of magnitude in costs to run this model? Is it thousands or millions or billions? The answer is relevant. People might consider such a far-fetched scheme reasonable if it costs pennies; if it is a vast expensive permanent bureaucracy, then not so much.

The real issue is whether the executive branch can do whatever it wants with no oversight, simply because the President wants to do it. If there are no limits, then the checks and balances are gone and we are left a few arbitrary steps from dictatorship. I know, that sounds like hyperbole, but if there are no limits, then why can’t the President order arrests of dissidents? Call out the military for domestic use (oh, hold it, he’s about to do that one).

If there are limits, then we need to raise this issue loudly and debate what those limits are, what needs oversight and legal checks such as warrants - and what doesn’t.