March 2008

The fraudnest was a high-rise apartment rented by a friend. The mastermind was tracked down to that Baltimore neighborhood through his credit card dining. The building he frequented was identified through facial recognition software over security footage in the lobby, (after authorities sifted through tapes of the entire street). The exact apartment he visited was identified by crossing phone records of all the residents and apartment owners with the suspect's phone. Telescopic surveillance placed in the next building did the rest.

In this particular case, the police exploited what might be called the bit mirror of their suspect. But the very same buildup of individual "bit crumbs" we drop behind as we go about our normal business is of increasing interest to private detective agencies. Detectives, in real life as in movies, are rather effective at tracking down their targets, making a lot of money servicing jealous spouses and worrisome parents. But that is retail. What is new is the wholesale nature of this bit tracking. In recent years, a new field of mathematics and computer science has been nurtured to near perfection by eager info snoops. It's called data mining.

In this discipline, practitioners deal with ever-evolving algorithms that can provide answers to questions not clearly asked, or not asked at all. It is straightforward for a bank to dig out all transactions carried out on a particular day on accounts belonging to single women with a 6-year-old daughters residing in Manhattan. Or New Orleans. It's a well-defined data slice, instantly responded to by automation. What is trickier is to unleash an algorithm with an assignment to spot irregularities where the term is ill-defined.

The state of the art for these mission-impossible algorithms is quite impressive. Many instances of fraud have been spotted by the authorities once they armed themselves with this power. Police and security companies alike are singing the praises.of data mining But for these data dogs to be able to sniff around crowds, it is necessary to feed them with as many disjointed bits as possible. Hence, we see a market for electronic records of public and private parking lots, credit card usage, entrance records to public buildings, and street and mall survey tapes. No byte is too small, no bit too marginal. The data dogs digest them all.

But this new police power comes with a dark side that we as citizens should carefully consider. In data mining, each record in and of itself is of little consequence. An individual file is meaningless. But together they allow someone who is well-versed in data mining to document the personal, private history of ordinary Americans. While the national-security argument is very potent against objections to this intrusion, it must be said that the more popular this sort of processing becomes, and the more Americans it tracks, the greater the likelihood that it could falsely victimize you and me.

In a large enough crowd, someone will behave outside the norm by the very definition of norm. And those data dogs will gallop back to their geeky masters with the names of innocent (albeit weird) Americans clenched in their teeth. The dogs' masters are eager to vindicate their invention, making it quite difficult to prove innocence. For example, what do you do if an employer hired such a service to check you out, then ended up denying you that prized job because your teenage son used your credit card to buy smut? It looked bona fide when you examined your statement, but the data-mining sniffer put you in a zone none of us should find ourselves in, and wouldn't without this technology.

Right now, most Americans shrug their shoulders, expressing resignation about this bit stripping of their privacy. But as these algorithms grow more powerful all the time, we've got to think this through on a very profound, systematic level.

The bad guys do. Knowing the data miners are lurking, criminals find and use clean records belonging to legitimate consumers. They use their credit cards and get their faces on camera. By sticking to cash purchases and throwaway phones, these criminals evade the algorithms, leaving these hunting tools to chew on the innocent. ClearBIT and other firms developing closed-circuit currencies are attracting growing interest from those trying to ensure privacy in electronic transactions. But the field is wide open for a fair and equitable think tank to offer a solution that can be readily adjusted to balance security and privacy.