We are all Angela Merkel.
Today, the National Security Agency can tap the German chancellor's phone, monitor her electronic activity and use computer analysis to attempt to figure out whether she's plotting to undermine American monetary policy at the next EU Summit or just planning to go home to make Mettwurst and cabbage.
Soon, Walmart or Amazon, or another large entity is going to be able to vacuum up large volumes of data about you, spin it through predictive analytics, and know with great certainty that because you just bought Pop-Tarts, light bulbs, and paper towels you are about to have a baby - before you know. They'll send you an offer for Pampers before the dot on the pregnancy test turns blue.
We're all so focused on the NSA's data collection that we might miss what comes next. Scientists and engineers are developing technologies that can analyze the huge flows of data from our digital lives to predict individual behavior. The goal is to predict your behavior before you even decide what you're going to do.
This is a very powerful, very attractive idea to spy agencies - as well as marketers, banks, police, and even professional sports coaches. It's a capability no one has had before. I mean, no man can even know his wife well enough to predict with reasonable accuracy what she will do in any given situation. (Wait - is that just me?) But the NSA has to be drooling over the promise of knowing the next move of every head of state, and marketers are sure they can sell more stuff by accurately anticipating customers' needs.
As is often the case with technology, behavior modeling is not intrinsically good or bad. On the positive side, it might be nice to have the doorbell ring just as you get hungry, and find the Domino's guy on your porch carrying your favorite pizza. But the technology could easily be abused to ransack our privacy like never before. It's a conversation policy-makers need to have - as soon as they finish fulminating about the NSA's omnivorous data grab.
For the NSA, behavior modeling adds perspective to the now-embarrassing program to snatch and save our allies' private communications. It is the likely answer to the question: Why is the NSA tapping cell phones and Google and Yahoo traffic and barely using much of it? "They park stuff in storage in the hopes that they will eventually have time to get to it," said James Lewis of the Center for Strategic and International Studies. "Most of it sits and is never looked at by anyone."
In anticipation of refinements in predictive technology, it makes sense for the NSA to suck up every morsel of attainable data, which is why it has turned into the data equivalent of a hoarder. To hold all this data, the NSA built a facility in Utah that can store something like 100 years' worth of worldwide communications. The place uses enough electricity to power 65,000 homes.
"There's no such thing as too much data," says Michael Campbell, CEO of International Decision Systems, who has long worked on predictive software. "The more data there is, the more apparent the patterns become. And the more 'normal' data you have, the more the weird stuff pops out at you."
For companies, early versions of the technology are already here. A credit card issuer, for instance, can take the data from all your transactions and model your behavior and then look for anomalies. If your card is suddenly used to buy $2 worth of gas and then an iPad, you get a call.
The next step is to use data to build more complex models of individuals, and then keep an eye on activity that might signal a change in behavior. Harrah's, for instance, tracks activities of its loyalty card holders while they're in the casino and tries to predict when they will get fed up with gambling losses and leave. When that moment has arrived, a Harrah's rep will appear and offer a coupon for dinner or a show to get the big-spender to stay.
Predictive capabilities will quickly get more sophisticated, anticipating subtler, more subjective decisions. However, "this modeling is really hard to do," says Irving Wladowsky-Berger, a Massachusetts Institute of Technology lecturer and former top IBM executive. The meshing of computer science and behavioral science still has a way to go.
U.S. security agencies are working on it and are no doubt leading the way. The CIA has invested in companies such as Palantir Technologies that can look for patterns and make predictions. It has invested in D-Wave Systems, a company trying to develop a quantum computer, which would work exponentially faster than today's computers and could presumably calculate wildly complex predictive models in a second. Who knows what else the NSA is developing behind those walls in Utah?
All in all, it's a good bet that before long, the NSA will have the computing power and enough data to model a Merkel.
As always happens, that kind of technology trickles down to the private sector - and companies will direct it at us. Today it's Merkel. Tomorrow, it's all of us who don't own a closetful of red suits.