Yes, Virginia, Data Mining Can Catch Terrorists

A recent article from Information Week, “Can Data Mining Catch Terrorists?,” asks a question I first thought was rhetorical…until I read further. The author, J. Nicholas Hoover, covers all the recent activity of the National Security Agency (NSA) spying issue, touching on USA Today’s article, “NSA has massive database of Americans’ phone calls,” by Leslie Cauley. They’re both worth reading.

Hoover is right about the technology for assembling humongous databases, not the least of which is Microsoft’s Access. Data, of course, is a necessary part of the equation, and the jury is still out on whether that has been given up. If the NSA does not have the algorithms that allow them to conduct the predictive modeling/data mining that identifies patterns of terrorism, it is simply a matter of incompetence. It is expensive—when has that ever bothered this administration?—but readily available.

But there is one part of the article that I take issue with, and it is where Hoover indicates the NSA could learn from retailers (junk mailers) who mine customer data “without invading customer privacy.” When CitiBank buys your name and financial data from TransUnion (one of the big three credit bureaus that uses data mining techniques for selection), to send you a credit card offer that could be intercepted by an ID thief, that is invading your privacy.

My other major concern is the attention given to the “accuracy of data” in the piece. Hoover indicates its integrity “is different” for each data broker. What isn’t mentioned is the fact that much of your private information is also incorrect. In the article, “Privacy Activism study finds new problems for ChoicePoint, Acxion,” two major data brokers, there are alarming facts that should make every American consumer rise up and insist on control over their names and personal data.

The error rates for private information by Acxiom and ChoicePoint were 67 percent and 73 percent, respectively. 100 percent of the eleven participants in the survey had mistakes in their background check reports. This included the most basic information of name and address, but also involved Social Security numbers and phone numbers. The latter, of course, the basis of NSA’s database.

ChoicePoint has 19 billion records, including information on most U.S. residents. Acxiom’s warehouse is similar and includes over two hundred demographic and lifestyle items the last time I checked. Each has sufficient data to insure that the NSA could easily determine your household’s daily habits.

In the Hoover piece, the Government Accountability Office found in a 2004 survey that federal agencies were already involved in or planning 199 data mining projects, including 122 involving personal data. That’s your private stuff, folks. And, many of these will use ChoicePoint and/or Acxiom data, because of the contract these companies already have with the government.

So yes, Virginia, data mining can catch terrorists. But only if it is done correctly and with accurate data. It must not in the process, however, randomly access the personal records of innocent U.S. citizens, who give up this data for purposes that have absolutely no relationship to terrorism.

In my next post, I’m going to reveal the technology and antics of the predictive modeler/data miner, so that readers can understand the process their names and private information go through to forecast their next move. It’s an event that occurs hundreds of times daily, creating an identity crisis for your household each time.