If you want to be on TV, don't go to Los Angeles or New York. Come to Chicago, where your wish is certain to be fulfilled. In fact, you couldn't avoid it if you wanted to, thanks to the nation's most extensive network of police surveillance cameras. Anytime you walk out your door, you may find an audience.

This is one of Mayor Richard M. Daley's proudest achievements, but the estimated 10,000 devices now in operation are not enough for him. He once expressed his intention to keep adding cameras until there is one "on every street corner in Chicago."

His obvious error is to assume that if some cameras are good, more are better. Daley's policy also rests on a plausible but unproven assumption: that cameras reduce crime by deterring criminals and helping nab those who aren't deterred.

If you are going to spend millions buying, installing and monitoring this technology, you had better be able to show it yields some positive results in practice. Given the experience in this country and abroad, skepticism is in order.

The government of Britain, where cameras are ubiquitous, concluded they have had "no overall effect" on crime. Researchers at the University of Southern California looked at two neighborhoods in Los Angeles and found no visible benefit from this sort of surveillance.

Even in the studies that show cameras help, the question arises: compared to what? Any funds spent on this gadgetry cannot be spent on beat cops, probation officers, laboratory gear or jail cells. The challenge for enthusiasts is to show the technology outperforms other options.

On those issues, the jury is still out. But the latest discoveries, from Chicago, are bound to encourage the spread of surveillance video in law enforcement.

Nancy La Vigne, director of the Justice Policy Center at the Urban Institute in Washington, has directed a study of the impact of cameras in Chicago, Baltimore and Washington, D.C. Her preliminary findings, due to be finalized and published this year, are that they can indeed curb crime - and at a bargain price.

Her team of researchers looked at two high-crime neighborhoods on Chicago's West Side, Humboldt Park and West Garfield Park. In Humboldt Park, she told me, they found "a significant decrease in total monthly crime numbers," including property crime and violent crime. They found no evidence that the cameras merely pushed crime into other areas. In West Garfield Park, on the other hand, they saw "no impact," possibly because there were fewer cameras.

On the cost-effectiveness test, though, La Vigne says the cameras were a solid success. For every $1 of costs, they yielded $4 of societal benefits (reduced crime, savings in courts and corrections, less suffering for victims), despite their failure in West Garfield Park.

In Baltimore, where cameras are concentrated in downtown and monitored actively 24 hours a day (as distinct from the more passive approach in Chicago), La Vigne found the impact on violent crime was even greater - and the benefits exceeded the costs by 50 percent. (In Washington, which deployed only a small number of cameras, they found no effect.)

All this may confirm those who see this technology as an unmixed good. But La Vigne herself worries that too much will be made of these results.

"I'm sure there are diminishing marginal returns," she says, meaning that each extra camera achieves less than the one before. "I'd expect very little impact on low-crime areas." If we have cameras on every corner, many of them will be the functional equivalent of potted plants.

Even if cameras have benefits, they narrow the scope of personal privacy, which should not be sacrificed without a compelling reason. In a crime-infested neighborhood, the loss may clearly be modest compared to the dangers of violent perforation. In more tranquil locales, the burden of proof should be on the supporters.

When cameras are used, common-sense restrictions are in order. The American Civil Liberties Union of Illinois recommends that police show probable cause that someone has committed a crime before they use facial-recognition software or conduct nonstop video tracking of an individual. Another proposal is to delete images after seven days unless there is reason to think they document a crime.

The ultimate question is not whether cameras work. It stands to reason that they can work when used wisely - just as a hammer works for certain tasks. But not everything is a nail.