The district known as the City of London is two-and-a-half-square kilometers of winding thoroughfares and dark alleyways where ancient churches nestle in the shadow of glass-clad skyscrapers. The City, as it is known to Londoners, is Europe’s financial heart, home to more U.S. banks than New York City and more Japanese banks than Tokyo. For the British government, it is valuable real estate—this small area alone generates 4 percent of the gross national product of the United Kingdom.

That’s why the area is protected by one of the most sophisticated security systems on the planet. The so-called ring of steel, inaugurated in 1998, is a network of cameras that provides comprehensive video coverage of a large part of the City. Every vehicle entering the area is photographed, its license plate checked against a national police database, and an image of its driver stored for posterity. “The ring of steel has had a dramatic effect on crime in this environment,” says Andrew Mellor, a superintendent in the antiterrorism department of the City of London’s police force and the person responsible for managing the system.

Now other cities want to follow London’s lead, with New York City at the head of the queue [see photo, "Crime Central"]. Earlier this year, the New York City Police Department announced that it was installing more than 500 cameras around the city and pushing for its own ring of steel to protect lower Manhattan.

What to expect? If London’s experience is anything to go by, New Yorkers could find that 24/7 surveillance raises difficult questions about the efficacy of widespread monitoring. There’s little doubt that surveillance can bring down some types of crime, but the big question is how well it can deter and prevent terrorist attacks.

The idea for a ring of steel was born in the early 1990s after a bombing campaign by Northern Irish terrorists left four people dead, dozens injured, and parts of the City of London in ruins. The system began as a high-profile police operation to monitor traffic entering the City and to stop suspicious vehicles. But putting police officers on streets coming into the area to monitor traffic was a logistical nightmare and ultimately unsustainable. “We needed a technological solution,” says Mellor.

At the time, the technology for automated license plate recognition was in its infancy. Read rates—the accuracy of the information captured by the monitoring equipment—were highly variable, depending on factors such as the speed of a car, its angle to the camera, and lighting. “No two cameras have the same read rate,” says Minesh Patel, a computer engineer who manages the vast computing resources needed to keep the ring of steel in operation.

Removing these variables turned out to be an important factor in increasing read-rate accuracy. Officials closed all but a few key roads into and out of the area and forced cars to negotiate specially constructed lanes at each entrance. The lanes were designed not just to slow vehicles but also to point them directly at the cameras as they passed. What the antiterrorism team wanted and got was a system with a read accuracy of 94 percent. Not bad for 1998.

But the City got more than it bargained for. Unexpectedly, the introduction of cameras had a big effect on the environment. The traffic-channeling measures not only slowed traffic but also reduced the number of vehicles entering the area, substantially improving air quality. It also allowed city planners to turn many roads that were no longer accessible into pedestrian malls. The result: a more pleasant working environment for many Londoners.

Today, the accuracy of automatic license plate recognition approaches 100 percent for cars traveling at ordinary city speeds in a wide range of lighting conditions. But now that traffic-channeling measures no longer are necessary, the irony is that cities like New York may not benefit in the same way.

One major challenge for surveillance officials is handling the data the London cameras produce. The system consists of over 200 cameras, each sending a 3.8‑megabit-per-second MPEG video feed to the control room of a police station in the heart of the City. Processing this data in real time requires 122 IBM xSeries servers with a total storage capacity of 200 terabytes.

Reading license plates alone is of little use until the details are checked against the country’s 82 million vehicles registered on the remotely accessible Police National Computer. Within seconds, any vehicle that appears suspicious is flagged by the computer in the police control center, and an officer makes the decision to stop it or let it drive on, based on what the national computer indicates.

It’s a hectic business. Last year, the cameras recorded 38 million vehicle entries into the area. Of these, 91 000 were listed for infractions on the national computer; 4161 warranted police action, leading to 539 arrests. Many serious crimes were uncovered as a result of stopping a vehicle for a minor violation. “It gives us a way in,” says Mellor. “With good police work, a traffic offense is just the beginning.”

He gives the example of a black Porsche Cayenne that was flagged by the computer last 13 February because the driver had not paid the car’s leasing bills. The police stopped the vehicle, searched it, and found US $20 000 in the glove compartment, triggering a major money-laundering investigation.

The system has become a major deterrent for certain types of crime, but sometimes the presence of cameras may merely push crime to areas where there is no monitoring. Then, too, use of cameras has bred some new misconduct, such as people offering to make license plates using unusual typographical fonts that are hard for computers to read.

One thing that hasn’t been much of a public concern is privacy. “People seem to accept that we need these cameras to deter terrorists,” says Mellor. But how effective has the system actually been in warding off terrorist threats? It’s not an easy question to answer.

There’s no way of knowing, really, how many attacks against the City might have occurred—or even how attractive a target the City is. Mellor believes that that part of London may not be at as high a risk as some other areas. “A suicide bomber can kill more people in the Bluewater Shopping Centre [in Greenlithe, just outside London] on a Saturday afternoon,” he says.

On 7 July last year, when four suicide bombers detonated explosives in London’s crowded transport system, killing 50 people, all the bombers were photographed entering the system. While the images have been invaluable in investigating the bombing and may help deter further bombings, they obviously did not stop the attack. Cameras alone will not be able to prevent all attacks, and not all attacks will be stoppable. “You just have to be realistic about the risk,” says Mellor, pragmatically.