The New York Times publishes an article by Eric Lichtblau this morning which reveals the spread of surveillance by the FBI along lines of association. If Person X was a designated target of surveillance, that surveillance would spread to a new set of people who were in contact with Person X. Why? Not necessarily because they had done anything suspicious, but just that they knew someone suspicious. Lichtblau refers to this practice as part of “link analysis.” Another name for it is “network analysis,” and the technique involves what’s called “snowball sampling.” Start with just a bit of snow, roll it, and you end up with an awfully hefty big snowman-sized thing that can tear up the sod when you move it.
One problem with using snowball sampling to expand surveillance is principled — it offends the idea of freedom of association, turning the all-seeing eye of Big Brother onto someone just because of who they know. Police and federal agents have always been inclined to follow lines of association to some extent when pursuing an investigation, but until recently they were constrained by the 4th Amendment need to show probable cause before invading their privacy. But under the Bush administration, the 4th Amendment is being ignored, and there was no need for a warrant under the FBI’s practice. As Lichtblau notes, the FBI would simply deliver requests for the information to telecommunications companies, and then receive it. That’s not constitutional.
Another problem, a problem that especially bothers me, is one of scale. Each time the government expands surveillance according to targets’ associations, the size of the population under surveillance is multiplied. If the size of the initial surveillance target population is 1,000 and the number of a typical person’s contacts is 10, all of a sudden you’ve got a surveillance population of 10,000 — ish. Actually, that’s not precisely true, because some of the original 1,000 people’s contacts overlap. Arnold and Olivia may both be in contact with Olysses, and you wouldn’t want to count Olysses twice since Olysses is just one person. To correct for this Bernard et al have studied extant networks and from their experience recommend a division of that 10,000 by 1.6. But even with this in mind you still move from 1,000 surveillance targets to 6,250 surveillance targets.
That doesn’t sound so bad, does it? Well, consider that the goal of this surveillance strategy is to identify new suspects. Once those new suspects are found, the strategy indicates one would look at their contacts too. And indeed, Lichtblau almost off-handedly indicates this is being done:
Officials at other American intelligence agencies, like the National Security Agency and the Central Intelligence Agency, have explored using link analysis to trace patterns of communications sometimes two, three or four people removed from the original targets, current and former intelligence officials said.
Yes, the NSA and CIA are doing it too. And at four removes, you really get an idea of how many people can get caught up in the surveillance snowball. Again, from an initial target population of 1,000:
first remove: 1,000 * (10/1.6) = 6,250
second remove: 6,250 * (10/1.6) = 39,062
third remove: 39,062 * (10/1.6) = 244,137
fourth remove: 244,137 * (10/1.6) = 1,525,856
Yeah, that’s a lot of people. This is the kind of operation Kansas Senator Sam Brownback was referring to when he said in a presidential debate last week that:
We need this because we’re not talking about just a needle in a haystack. We’re talking about a needle in a hayfield of millions of people that we have to, we need to, watch, and we need to find with good intelligence who it is that seeks to do us harm.
This is also why the Bush administration and the spooks under Bush’s command don’t want to have to pay attention to the 4th Amendment to the Constitution any more. Not just because they don’t have probable cause for their invasive surveillance, but also because they want to put so many people under their Orwellian eye. Writing out millions of warrants would lead to a lot of papercuts.
Is that why they say any given person has 5 degrees of separation from any other person?
I seem to remember something else like this. It was before my time, but it was called the McCarthy Era. People were questioned about their associations and they were hauled in to testify before a congressional committee about their friends. People who were merely socialisticky were shunned and could not get jobs, to say nothing of the real card-carrying Bolshevik communists. If you studied child development in a psych class, you will remember someone named Erickson who postulated eight developmental “tasks”–he left his position at Berkeley very publicly rather than go along with the witch hunts.