Affidavit on the Power of Metadata

We’ve referred to the matter of governments spying on citizens a number of times here on Slaw, particularly in relation to the feebleness of the “it’s only metadata” excuse. If you’d like more confirmation that metadata has the power to breach privacy in a serious way, have a look at the affidavit — the “declaration,” to be precise — by Princeton computer science prof, Edward W. Felten filed in an American Civil Liberties Union lawsuit. Professor Felten takes you through the process step by step, intending, of course, to educate the court about these technological matters.

There’s perhaps a natural inclination to think that even though capturing metadata about phone calls is rather invasive, at least it’s “second worst” to capturing the content of the calls. And while in a friction-free world that might be true — all thoughts expressed over the phone: known and recorded — in this world the “friction” involved in coping with the content of phone calls makes metadata clearly the way to go, if you’re a spy. Felten points out that “[t]elephony metadata is, by its nature, structured data” and, so, easy to aggregate and analyse. By contrast, spoken conversations are hard:

Some people speak English, others Spanish, French, Mandarin, or Arabic. Some people speak using street slang or in a pidgin dialect, which can be difficult for others to understand. Conversations also lack a common structure: Some people get straight to the point, others engage in lengthy small talk. Speakers have different accents, exhibit verbal stutters and disfluencies. Although automated transcription of speech has advanced, it is still a difficult and error-prone process.

Another point that struck me as I read the declaration was the problem of very large numbers. Felten makes a series of conservative assumptions in order to arrive at some estimates of the size of the collected data — the size of the problem, if you will. Three billion US phone calls a day yield 140 gigabytes of data a day, or 50 terabytes of data a year, which, in terms of “pages” comes to “25 billion pages of information every year.” The program has been in place for seven years.

I realize that this appears to be the flip side of the point just made above, yet the fact is that NSA technology can cope with this volume. And with “big data” come big results so far as intrusion into individual privacy is concerned. But more problematic, for me at least, is the fact that these numbers are beyond my imagining. Terabytes have for me “slipped the surly bonds” of understanding and simply sail off into the incomprehensible. This is dangerous because we cannot grasp the enormousness — and enormity — of the problem and, so, might tend to file it where we file other unimaginable things, such as the beginning of time or the edge of the universe, which is to say in the “I can’t handle this” top shelf of some mental closet. All in all, this sort of “big” intrusion might be the latter day equivalent of the “big lie,” successful because of its size.

There’s not a great deal new in the declaration, but the facts are stated clearly and calmly — and by an expert. Worth reading.