Data-mining tactics surface in the government's national security efforts


This city, famous for being America's playground, has also become its security lab. Like nowhere else in the United States, Las Vegas has embraced the twin trends of data mining and high-tech surveillance, with arguably more cameras per square foot than any airport or sports arena in the country. Even the city's cabs and monorail have cameras. As the U.S. government ramps up its efforts to forestall terrorist attacks, some privacy advocates view the city as a harbinger of things to come.

In secret rooms in casinos across Las Vegas, surveillance specialists are busy analyzing information about players and employees. Relying on thousands of cameras in nearly every cranny of the casino, they evaluate suspicious behavior. They ping names against databases that share information with other casinos, sometimes using facial-recognition software to validate a match. And in the marketing suites, casino staffers track players' every wager, every win or loss, the better to target high-rollers for special treatment and low- and middle-rollers for promotions.

"You could almost look at Vegas as the incubator of a whole host of surveillance technologies," said James X. Dempsey, policy director for the Center for Democracy and Technology. Those technologies, he said, have already spread to other commercial venues: malls, stadiums, amusement parks. And although that is "problematic," he said, "the spread of the techniques to counterterrorism is doubly worrisome. Finding a terrorist is much harder than finding a card counter, and the consequences of being wrongly labeled a terrorist are much more severe than being excluded from a casino."

The casino industry, like the national security industry, is seeking information to answer a fundamental question: Who are you?

"It's, are you a good guy or a bad guy? A threat or a non-threat?" explained Derk Boss, the Stratosphere's vice president for surveillance, whose crew operates under what he calls the IOU system: Identify, Observe and Understand.

"There are going to be people that just want to come and gamble and enjoy your services," he said. "And there are going to be people that are going to come to take your money. Our job is to distinguish between those two groups."

In the surveillance room, 50 monitors are linked to 2,000 cameras, from the casino entrance to the tower observation deck. Two employees keep an eye on the monitors. Guests are on camera from the moment they enter &

except in their rooms and in bathrooms. An investigator tracking a suspect could go back and review old tape, assembling a mosaic of a visitor's moves for the past two weeks.

What happens in Vegas does indeed stay in Vegas &

for a lot longer than most patrons realize.

On a recent Friday night, the surveillance team at the Stratosphere hotel and casino is watching a casino host they suspect of handing out unwarranted "comps," or vouchers for free rooms and meals to guests. Might he be taking kickbacks?

Down on the floor, the pit boss is observing players, looking for "tells" &

behavioral signs of cheaters or other undesirables. The night before, investigators identified a blackjack player as a card counter. Casinos dislike card counters because they can determine when the cards are to their advantage and raise their bets accordingly. When the pit boss told the card counter he could bet only the minimum amount, he cashed in his chips and left.

Casinos have tried to use facial recognition software to identify known cheats in real time, but with little success. Casino lighting is often dim, and a player who wants to conceal his identity can hide behind a hat, sunglasses or a false beard.

But in a few years, some say, iris scan technology will be mature enough to use in gaming. Casinos might ask people to sit for a scan of the iris, which, like a fingerprint, has a unique pattern. That pattern would be transformed into a template to be matched against a database.

After Sept. 11, 2001, several airports tested facial recognition software, with little success. But the government is continuing to invest in biometric technologies, and the U.S. military already uses iris scans on suspects captured in Iraq and Afghanistan.

On occasion, national security and casino security interests directly intersect. Jeff Jonas discovered that after he developed a computer program for the casino industry that helps detect cheats using aliases.

A 43-year-old technology visionary and high-school dropout, Jonas soon realized that his system could also identify employees colluding with gamblers, say, by discovering that they shared a home address. He calls his program NORA &

for Non-Obvious Relationship Awareness.

Every time a player registers for a loyalty card or a hotel room, Jonas explained from his lab near the Strip, the player's name, address and other data are sent to NORA. Also in the casinos' NORA database is information about employees and vendors.

NORA can spot links that a casino employee probably would never discover, such as a phone number shared by two different names, Jonas said. It once identified a casino promotions director who picked a winning ticket that belonged to her sister, he said.

The idea was so powerful that the CIA's private investment arm, In-Q-Tel, poured more than $1 million into NORA to help root out corruption in federal agencies. Then, after the Sept. 11 attacks, it became clear that link analysis could be useful in tracking terrorist networks.

In 2002, Jonas shared his technology with Pentagon officials, who were researching a more controversial technique called pattern-based data mining. Their aim was to identify terror networks from patterns of behavior, by plowing through vast beds of data such as hotel, flight and rental-car reservations. Jonas, now an IBM chief scientist, said that narrowly focused link analysis is less invasive because it starts with a known suspect rather than casting about in the general population.

At the U.S. Treasury's Financial Crimes Enforcement Network, for example, investigators have used link analysis to track money laundering. From one Suspicious Activity Report &

which financial institutions are required to send to the government &

they have identified a money launderer's partners in crime. FinCEN has a decade's worth of data on 170 million report forms. "We find a tremendous amount of connectivity," said Steve Hudak, FinCEN spokesman. "We find suspects linked by addresses, suspects linked by phone numbers. So we definitely know that these people are operating together."

But privacy advocates warn that the farther it moves from the suspect, the more likely link analysis is to snare innocent people.

"We often hear of the surveillance technology du jour, but what we're seeing now in America is a collection of surveillance technologies that work together," said Barry Steinhardt, American Civil Liberties Union technology and liberty project director. "It isn't just video surveillance or face recognition or license plate readers or RFID chips. It's that all these technologies are converging to create a surveillance society ."

Share This Story