Monday, February 06, 2006
ECHELON is data-mining you
While people argue about the possible illegality of President Bush’s ordering of wiretapping of US citizens by the NSA - flatly ignoring the NSA’s own internal guidance forbidding listening in on American citizens’ private communications, not to mention FISA - there has also been some debate about what the NSA is capable of doing, and whether or not the eavesdropping might actually be practical.
Interestingly, the NSA is more than happy to share a great deal of information about its tasked mission with the general public. This seems reasonable to me, as any modern hacker is probably as up to speed on eavesdropping capabilities as any government employee, and in many cases, probably more so. It’s only that most civilian hackers don’t have access to the government funding to duplicate the NSA’s capabilities.
In a nutshell, here’s how it works:
All electrical communications devices emit electromagnetic radiation, which can be sent over wires or broadcast through the air. We’re talking telephones, fax, the Internet, TV, radio, everything. It’s no big deal for anybody with sufficiently sensitive equipment to pick up these signals, which is all that the NSA is doing on the mechanical side.
Their program for the harvesting of signals is called ECHELON. It is basically a world-spanning, ground- and space-based network of antennas, picking electronic signals out of the ether, amplifying them and sending them to Ft. George Meade, Virginia, where the signals are processed by a program called DICTIONARY. The FBI has had a similar program for years, code named CARNIVORE, which has its own troubled history.
Some nice graphics illustrating the various mechanical devices used and their locations are available here, here, and here.
DICTIONARY is a search algorithm, kind of like what Google, Yahoo, Ask Jeeves, and other search engine sites use. In fact, the NSA has applied to the US Patent Office for a patent on the kernel of DICTIONARY’s algorithm.
The only real secret about NSA’s search capability is the fact that they even exist, and, until recently, the code names. Once we know that they do exist, we can extrapolate the sophistication of the enterprise just by comparing it to what Google does in matching Google Ads or AdSense ads to your search results. There is no practical difference. In fact, Google probably has the better search engine. Free enterprise at work, don’t you see.
So here is the real nut of the problem: as anyone who has actually looked at the ads that Google presents to you as a reward for your search for “widgets” knows, the results are, to be fair, mixed.
The NSA has the same problem. If they intercept a person talking on the phone about exploding piñatas, however innocent the conversation may be, as a conversation about explosions it gets tossed into the bin along with conversations of members of the 9-11 survivors’ group, say, or Fourth of July celebrations, as well as the targeted Golden Gate Bridge explosion plot, and about ten thousand other conversations to boot.
Because they have one more problem that Google or Yahoo doesn’t.
Bear in mind that all modern electronic signals are multiplexed. In order to first get even one of those conversations, unless they are actually tracking a particular cell phone number, or have a physical tag on a landline, they willy-nilly pick up every other signal that is piggybacked with that targeted signal. So, in order to get the one fishy signal, they have to listen to one thousand, and in order to get a thousand, they have to listen to two hundred thousand. In a weird way, we are all on a planet-wide telephone multiple-party line.
To illustrate: the Feds have a tip that an al-Qaeda cell is located somewhere in a city block. They want to intercept the telephones calls of that cell. Unless they know the apartment number of that cell, they have to eavesdrop on every telephone call on that block, which could be upwards of ten thousand people in a place like New York City, for example.
That’s information overload folks, with a vengeance. It’s also why we call it data-mining, or dragnetting. And to make it even worse, it’s probably useless and a huge waste of the taxpayers’ money.
Not only are they listening to the bad guys, they’re listening to you, too, and that’s illegal, and that’s why they need a warrant, no matter how loudly they scream “national security.”