The Numerati
Page 7
With this type of snooping technology, managers can start scrutinizing our movements. In these early days, they focus more on overall patterns of traffic than on individuals. That’s because today’s cameras have foggy vision. They see us as little more than moving blurs, Ghani says. They’d be hard-pressed to identify our faces, even if we stood perfectly still, gazed up at them, and mouthed, ever so slowly, our names. Most automatic surveillance systems, which seem to recognize faces so well in the movies, don’t yet work such magic in the real world. Douglas Arnold, director at the Institute for Mathematics and Its Applications at the University of Minnesota, says that facial recognition was oversold as far back as the 1960s. Researchers are making progress, Arnold says, but “if people start relying on facial recognition systems today, they’re going to be burned.”
So how will Accenture’s cameras pick out individual workers and shoppers? Ghani introduces me to what he calls massive redundancy. This involves getting lots of cameras to work together as a team. Each one provides a bit of detail. It’s a little like a group of witnesses who see a thief dash by. One might remember his red hat, another the bandage on his hand. A third points to the alley he ran down. In Accenture’s case, the system can stitch together these smidgens and come up with a guess as to who each blur is most likely to be. They can be pretty sure, for example, about the identity of a short, dark-haired figure wearing a blue shirt who emerges from Ghani’s office accompanied by a taller stranger with an oddly stiff neck (me, bad hotel pillows). Stature, colors, and the patterns of movement all indicate that the person is Ghani. The system makes similar calculations about the other Accenture employees on the floor. This produces truckloads of visual data. Accenture’s computers use that information to feed all sorts of analysis. They can create charts showing each person’s migratory patterns, social hubs, and yes, even bathroom visits. Similar analysis could be focused on us as customers. In time, perhaps a store will recognize us by our movements in the aisles as likely butterflies or barnacles, or even potential shoplifters. And as the facial recognition systems improve, they may spot the barnacles among us the moment we enter the store.
If cameras don’t pick us up, a radio technology known as RFID just might. These are little computer chips fastened to a piece of merchandise, a shopping cart, or even a customer loyalty card. Each chip has a unique number, identifying the item or the shopper. But unlike a bar code, which has to be passed under a scanner, these chips can be read by radio signals sent by an automatic reader in the area. It’s great for logistics. Open a huge cargo truck, and instead of piling through it and scanning each bar code, the chips all transmit their data at once. The detailed contents of the shipment pop up in a split second.
These same chips can track us in stores and at conventions. AllianceTech, a company in Austin, Texas, puts these radio tags into the ID badges people wear around their necks or clipped to their lapels at trade shows. The company also puts receivers into the booths they visit. Then, when IBM or Texas Instruments wishes to know who visited its booth, AllianceTech can give them the names of the people (at least those who agreed to participate), their companies and industries, and the amount of time they spent in the booth. They can even see how much time these people spent visiting a rival’s booth. If you look at the flow of customer data, it’s as if the whole trade show is taking place on the Internet.
Imagine what would happen if retail stores used the same technology. Some are moving in that direction. Germany’s Metro, the world’s fifth-largest retailer, is equipping smart carts with radio transmitters in several of its stores. The technology, says Albrecht von Truchsess, a spokesman for Metro in Düsseldorf, is meant to provide shoppers with enhanced service but not to compile their shopping-related data or to profile them. (Data privacy is a far more explosive issue in Europe than in the United States.) As shoppers push the smart cart, they scan the bar code of each item they toss in. This information is sent by wireless connection to the computer, and, much like a driver cruising through an automatic toll-booth, the shopper can roll the cart out of the store without stopping to pay. The technology has taken care of that.
By mapping the trail of those scans, Metro can trace the minute-by-minute wanderings of each shopper. Even without building personal profiles, Metro’s analysts will be able to study patterns. They may discover that many of the most carefree, sky’s-the-limit shoppers never encounter the display in aisle three promoting sinfully rich (and expensive) Belgian chocolate. Just like a website, the store has plenty of options to entice the consumer: It can flag the chocolates on the smart cart screens. Or it can tweak the layout of the store, placing the chocolates along the pathways most popular among spendthrifts. Pity the dieters who dare to shop in the markets of the Numerati.
Chapter 3
* * *
Voter
QUICK. Who did you vote for in 1996?
Does that question inspire just a trace of panic? Do you worry that you might not remember, or perhaps that you’ll blurt the name of a candidate who wasn’t even running? Maybe you remember perfectly well, but you’re afraid that if you tell me, I will hound you with aggressive follow-up questions. You voted for HIM? What in the world was going through your mind?
If you experience these fears and anxieties, welcome to the club. You’re in the majority. Those who actually enjoy politics, who naturally think about the world the way politicians do, are a minority in just about every country. But because members of this cozy club run the political realm, they tend to analyze politics as if everyone else viewed it with equal fascination and zeal, and the same focus on issues. That’s why they fail to connect with voters, says Joshua Gotbaum, or even to understand them.
Josh Gotbaum is a card-carrying member of that minority. He served, as a young man, in the Carter administration, and he returned in middle age to work for President Clinton. He held senior posts at the Pentagon, the Treasury Department, and the Office of Management and Budget. When the Republicans run the show, he makes money and works for charity. At one time, he was a partner at Lazard Frères, the investment bank. After the terror attacks of 2001, he headed the September 11 Fund. Later, he moved to Honolulu and led the effort to pull Hawaiian Airlines out of bankruptcy. Now, as I talk to him, he’s attempting to resuscitate an education start-up in New York, from offices high above Wall Street. But he’s eager to get back to working in government. Naturally, the key is to find more people in the U.S. population who could conceivably vote for a Democratic candidate and to come up with just the right pitch for each voter. The trouble, he believes, is that millions of potential Democrats are camouflaged. For one reason or another, they pass for Republican voters. Some live in McMansions with fairway views and drive Hummers. Some no doubt carry weapons, revere the military, or spend much of their free time praying. Others are staying hidden because they haven’t been much impressed to date by Democratic candidates. To turn these people into Democratic voters, Gotbaum says, his party needs to surpass the Republicans in pinpointing potential supporters from within massive databases.
The Republicans set the standard for political microtargeting in the 2004 elections. First, they framed the issues in a fresh way, avoiding much of the policy verbiage that bores or bothers most voters to distraction. They focused on simple desires closer to the heart than the head. Things like feeling safe, loving their country, and being surrounded by people of faith. They spent millions on polling, and they used what they learned to unearth their target voters. Matthew Dowd, an aide to President Bush’s longtime operative Karl Rove, joined with two coauthors to detail this triumph in the 2006 book Applebee’s America.
Now Gotbaum, his auburn hair and mustache tinged with gray, has set up his own political firm, Spotlight Analysis, and raised $1.5 million in venture funding. He views it as the Democrats’ riposte. Success with Spotlight could be his ticket back to power. The way he sees it, the party that figures out how to harness the power of data, and the brains of the Numerati, is going to win at
the polls. It’s not that microtargeting will corral the 50 million or 60 million voters that a presidential candidate needs to win. The traditional approach—blizzards of TV ads and immense phone and door-knocking campaigns—bring in the lion’s share. But in races decided by one or two percentage points, or less, the party that pinpoints a few thousand individual voters here and there could come out on top. If the data we emit gives off even the slightest whiff of “swing voter,” the political Numerati will be hot on our tracks.
This has implications for the “decideds” too, as politicians start using their data to deploy limited resources more efficiently. If you come across as a safe vote, the candidates you endorse can direct their promises and stump speeches toward wavering supporters. This has long been the case. But traditionally, politicos have appealed to entire fence-sitting clans, crafting one appeal for retirees in Florida, for example, and another for worried autoworkers in Michigan. Now, though, the Numerati are collecting far more data on these swing voters and can study them with greater precision. This means they can place them into ever-smaller groups. As they create and test messages aimed at each sliver of the electorate, the science of the Numerati is supplanting the folk wisdom of the precinct chiefs.
THIS IS THE CHALLENGE Gotbaum faces as he tries to find votes: how do you calculate what moves people politically if, for many of them, the entire subject is toxic? Compared to politics, shopping’s a cinch. People long for things. They pick them up. They buy them. They leave clear records. When Rayid Ghani and his team at Accenture dissect your supermarket receipt, they can see that every two weeks you buy, say, a bag of bright green Granny Smith apples. In a sense, you vote for those apples with your credit card. Based on the patterns of other apple buyers, it doesn’t take a big leap on their part to predict other foods you might like. They’re working well within the realm of the data you provide. The comparison is what we might call “apples to apples.”
Now consider politics. Many people don’t like to think or talk about it. They switch the channel or turn the page. With the explosion of new media, they have literally thousands of news and entertainment choices—many of them, in their opinion, a lot more fun than politics. What’s more, U.S. citizens resist formal efforts to measure their political sentiments. If you think I’m kidding, consider this. Pollsters say that as few as 12 percent of Americans bother to respond to their phone calls. What’s worse, from a data hound’s perspective, those moved to leave their house on a rainy day in November vote behind a curtain. It’s secret.
This means that political operatives must dig into other data to find potential voters. Traditionally, they’ve settled for proxies for our behavior. Our neighborhoods used to be a good predictor of voting behavior. The same with race. But most of those big categories are breaking down. We walk less and less in lockstep, and we have more choices. For the kind of analysis Gotbaum wants, he must delve deeper. He has to find out not just where we live and work but what we love, what we fear, what we feel deep down about squishy subjects like community and country. The mathematics of politics, strange as it might sound, has to extend beyond the grasping fingers of our inner shopper and plunge into terrain that lies closer to the soul. To get there, researchers must move far past mouse clicks and Google queries, those commodity piles of personal data that course through the Internet. This involves asking lots of questions.
Here’s one, just to give you an idea. What is your community? Conjure it up in your mind. If you could draw a picture of it, what would it look like? Would it feature people waving from windows and porches in the group of houses up and down your street or the apartments on your block? Are these physical neighbors your community? Or is it a group with shared values, perhaps based in a church? Maybe your community is a widespread Internet group gathered around a blog that focuses on Dostoevsky’s novels or Chianti. Maybe you have an even more expansive view of community, which includes all of us who live and die on this blue lump of rock circling the sun. For some, our community extends to animals. Yet just imagine the blank looks a politician will receive in certain quarters if she evokes a community that includes pigeons and pilot whales. Some voters will write her off as an alien. Yet for others, she’s a kindred soul. It’s second nature for skilled politicians to tweak their message, or even their accent, for different audiences. But how can you read the crowds when they’re dispersed across networks?
Think of the hazy words and concepts that pop up in politics. Freedom. Democracy. Justice. Security. Opportunity. Human rights. Wealth. Politicians reflexively drop them into stump speeches and TV ads. Yet they all spark wildly different responses. For some, the word justice means executing murderers. For others, it’s giving poor children equal opportunities in school. If politicians could group people according to how they view these concepts, they would better understand the role that each person sees for government in society. That’s what we express when we vote. And with that knowledge, politicians could craft messages that speak to our values and appeal to our concerns.
This was a simpler job for past generations of politicos, because we Americans organized ourselves in clear groups and worked hard to assimilate. Take my parents. In 1954, they moved to a leafy suburb of Philadelphia with their three girls. (I wasn’t around yet.) They promptly went about the business of fitting in. They bought a big red Plymouth station wagon and put their names on the waiting lists for two of the snootiest clubs on the eastern seaboard, one for tennis, the other for golf. They began attending a venerable Episcopalian church favored by the established order. They subscribed to the (now defunct) Republican paper, the Evening Bulletin. And like most of their new neighbors, they registered locally as Republicans.
If one of the political Numerati at the time (such as they were) had been told to create a mathematical model of my parents, he probably would have asked, “Why bother?” There was little need to customize anything for them. They were doing all the work themselves by adjusting to the local norms. They were squeezing into an off-the-shelf algorithm. It defined a large and politically moderate group known as Rockefeller Republicans. Many of these families had been Republican since the days of Abraham Lincoln, when Republicans in the North led the fight to defeat slavery and save the Union. From oil paintings on the walls of our living room, a couple of our nineteenth-century ancestors stared down sternly at us. I always assumed they were Republicans. Wasn’t that what we were? In the early 1960s, my father ventured into local politics and won a spot on the town board. He could tell you, block by block, which houses could be counted on for Republican votes—and which doors to skip on election day. (Professors at nearby colleges tended to vote the other way.) Things were pretty simple. But with the growth of the civil rights movement and the Vietnam War, changes started to percolate in our Republican household. The party’s nominee for president in 1964, Senator Barry Goldwater, stood far to the right of my parents, especially on civil rights. My mother feared he’d push us into nuclear war. So she became a “Republican for [Lyndon] Johnson,” and she broadcast her support for the Democrat by placing a bumper sticker on our station wagon. My father, hounded by a local conservative columnist, lost his next election. Over the next few years, it was as if a 100-year-old dam had broken. My parents took buses to Washington to march against the war. They quit the country clubs and the big church. And when they moved out of the house in the early 1970s, they went to considerable trouble to find African American buyers, the first in the neighborhood. At some point during this process they became Democrats.
Look at them from a political pollster’s point of view, though, and it was hard to tell they weren’t still Republicans. They lived in privileged neighborhoods and rode the stock market boom of the 1980s. My father continued to climb aboard the same commuter trains that traditionally shuttled suburban Republicans back and forth from Center City. How was a statistician to know that instead of traveling to Bar Harbor or St. Moritz, this particular couple was vacationing in Nicaragua, joining a “human shield”
to defend the Marxist government there from the American-backed Contra rebels? They were camouflaged. Yet even though my parents had abandoned the Republicans, the Democrats didn’t know exactly what to make of them either. Like many Americans (of the minority who are politically engaged), they had broken loose. Politics no longer described their inherited identity. It now fit into a long menu of consumer options. Political mailings arrived every day, along with the credit card come-ons and gardening catalogs. They sifted through them, looking for candidates and issues that appealed to them, and threw the rest into the trash.
If you think about it, this free-range shopping has moved into almost every facet of our lives. People now shop for neighborhoods, religions, and cuisines that suit their lifestyle. Should you keep preparing the chimichangas or goulash your grandmother made or become a vegan? It’s a choice. Millions of us now shop for climates and even countries, settling in Vancouver or Barcelona, or retirement havens on the lakes outside Guadalajara. Even the slope of our nose has become a choice. So why not pick and choose in politics too? That’s what more and more of us do. This means that politicians, who used to locate us by our old groupings, must now find the new tribes and communities we form, often ones based on interests or values. The words Democrat and Republican are now foggy old terms that fail to describe most of us. As far as politicians are concerned, millions of us are lost (or don’t care). Like other marketers, they’re attempting to track us down by following our data. Only then can they engage us.