NSA leaker "trained" to be Special Forces
Silent Warriors vs. the Warrior Princess

A Grim Reading of the Recent Intelligence Scandals

What was the worst thing you learned about American intelligence this week?  Here's the worst thing for me.

Not the top-level finding, that the CIA's analysis didn't always give specific categories to the people killed in allegedly-CIA-led drone strikes.  Nor the secondary claim, that the CIA is lying about the number of civilians killed.

No, the worst news was that the CIA couldn't just ask.  

For a decade, these tribal regions where these drone strikes have been conducted have been one of the very top priorities for US intelligence collection.  The most rudimentary of human intelligence networks could have come up with a definitive list of who was killed.  Almost no risks would have been run in collecting on this topic, as it would have been the subject of common conversation among everyone in the area -- everyone whose family members might have been killed, for example.  No one would have thought it was odd to ask who died in yesterday's drone strike.  You could collect on this kind of thing without breaking a sweat, if you had a HUMINT network at all.

What this means is that the CIA has completely failed at its main function, in one of its highest-priority areas, for more than a decade.  The reason we're turning to all this fancy "collect-everyting-anyone-says-anywhere-at-any-time" technology is that we've failed at traditional tradecraft.  The bueraucracy isn't doing its job.

As the CIA case shows, there are a lot of disadvantages to relying on SIGINT.  It's simply nowhere near as reliable as human intelligence.  You can't ask questions:  you have to infer from what you are told, or what you can happen to see in the signal.  The reason we don't know isn't that we aren't collecting everything SIGINT can show us:  as we see here, Pakistan was one of the NSA's most intensely-collected states.

I personally would like to see a lot of this SIGINT capacity dismantled, on the theory that we ought to be as secure in our electronic communications as in those we write on paper and seal in a thin envelope.  But whether you agree with that or not, the fact is that it's less reliable than the traditional capacities we no longer develop.  That failure -- a failure, I believe, of will -- is driving these scandals.  Because these SIGINT techniques are less effective, that failure is also putting America at risk.