Friday, May 26, 2006
NSA,CIA, and cocaine: 3 birds with 1 stone
We can assume the NSA's version of TIA will find a new home at the CIA. For the sake of argument, let's assume everyone involved in this program is committed to the rule of law and seriously interested in protecting America from threats.
Even if that is all true, we still have a problem. How do you demonstrate the program's value to the people who pay for it without sharing classified information? It turns out, Ronald Reagan had a solution to that sort of problem; he called it "trust but verify." We can apply that approach to solve this impasse while meeting the legitimate needs of all the stakeholders. Here's the proposed test:
Demonstrate the NSA program works by using it
to eradicate cocaine traffic into the United States.
In a previous posting, I calculated the probability of finding a terrorist plot by data mining massive data sets assuming no baseline information. Based on those numbers, the odds of locating a terrorist plot were about 1 in a billion. Let me put that in perspective. You are about 300 times more likely to be killed by lightning. Clearly we need a test with better odds of success.
Unlike future terrorist attacks, cocaine trafficking is a well-documented phenomenon. There is no mystery where it is grown. There is no mystery where it is processed. There is no mystery how the cocaine gets into the country. In other words, there are plenty of nodes to start with for any sort of network analysis. That dramatically increases the prospects of finding useful relationships.
Many people think of "mules" as a major source of cocaine trafficking. Actually, the overwhelming majority of cocaine entering the US arrives via container shipping and general aviation. So how many containers are coming into the country each year? About 10 Million. How many general aviation flights? Less than a hundred thousand. Those are big numbers, but they are peanuts compared to the 1 trillion phone records the NSA is supposedly data mining.
In addition to smaller numbers to search, cocaine trafficking also has a baseline history of interdiction data. The Coast Guard, by itself, seizes one half to one third of all the cocaine entering the country. We know where those ships came from. We know who the shipping agents were. We know where they were heading. We know a lot about the supply chain. The same is true for cocaine busts involving general aviation.
Here's the math: Assume you have 10 Million container shipments and 100 have cocaine on board. Now assume your algorithm for detecting a suspicious transaction gives you a false positive 1 in a thousand times. Assume your algorithm misses a suspicious transaction 1 in a thousand times.
The big difference between this test and the proposed terrorist surveillance program is we have cut our total number of events from 1 trillion to a mere 10 million. That lowers the odds of finding a drug shipment to 1 in 100. Let me put that in perspective. That is more likely than getting dealt a straight in a game of poker!
Based on the numbers, a system that can locate just one terrorist plot should be able to locate well over 50% of the cocaine shipped into the United States. Add that to the cocaine already being seized by law enforcement and you will wipe out the cocaine trade.
You can increase the odds even more by focusing geographically. State and local officials seize about one third of the cocaine that gets past federal authorities. When you break down those numbers by state, it turns out over 90% of those seizures occur in ten border states or ports of entry: Texas, Florida, Illinois, California, New York, Arizona, Missouri, New Jersey, Georgia, and Louisiana (in decreasing order).
The benefits of demonstrating this program's usefulness by eradicating cocaine traffic into the US would be clear to everyone. Successful conclusion of this demonstration won't require publishing classified information. It won't require any acts of faith on the part of the general public. It won't require assaulting anyone's civil liberties. The results would be obvious and independently verifiable. The benefits would be immediate and substantial. In addition to saving a lot of taxpayer money, we will also be denying a revenue stream to arms dealers and terrorists.
Finally, there is a residual benefit to this approach. Many people are concerned about the possibility of some WMD being shipped into the country. Anyone contemplating such an attack would have to think long and hard about their ability to evade detection if they knew they had to run a gauntlet even drug dealers could not beat. I'm not saying this would prove the system will protect us from terrorists, but you have to admit it would be a hell of a deterrent.
Mything The Point ©:
"Examining unexamined beliefs America accepts on faith value."