Pop culture has conditioned us to expect miracles from technology: from the cyberpunk-driven PC that brings down the Defense Department's mainframe to the mild-mannered Mac that is one upgrade short of sentience.

The reality is that technology is always a work in progress, and when you throw humans into the mix, perfection is an elusive goal. That gap between expectation and performance is evident in Rochester's five-year-old ShotSpotter system.

The Rochester Police Department has 116 ShotSpotter sensors located around the city, positioned where data and analysis have shown the highest volumes of gunfire. The sensors are designed to detect gunfire and immediately relay that information to the city's 911 center. An operator verifies that the sound is gunfire and not fireworks or a backfiring car, for example, and then dispatches police.

"It's about as instantaneous as you're going to get," says Mike Wood, RPD's deputy chief of operations. "That's one of the advantages of it: for us to be able to detect gunfire immediately."

The RPD also monitors the ShotSpotter system from its camera room at the Public Safety Building. The system immediately creates an audio file of an incident which can be reviewed by police. From the file, police can learn how many shots were fired, whether multiple guns were used, and other information.

"If you hear "Pop! Pop!" and then, "Boom! Boom!" you don't have to be an expert sometimes to hear the difference between a .22 and a shotgun," Wood says.

ShotSpotter can help investigators reconstruct events surrounding a shooting, Wood says.

And ShotSpotter evidence can be used to verify or disprove elements of a suspect's story, such as how many shots were fired, if another person fired back, and other details.

The RPD has integrated ShotSpotter with the citywide surveillance camera system. The cameras automatically re-orient themselves in the direction of a ShotSpotter alert - meaning police could have sound and video of an incident or its aftermath.

It all sounds good so far. But the numbers paint a less rosy picture. Sometimes the alerts turn out to be for sounds other than gunfire, and many times officers respond to a scene only to come away empty handed. According to data provided by the RPD, between 30 percent to 40 percent of all ShotSpotter activations over the last four years were ultimately judged by a 911 operator not to be gunfire. And of the 3,306 ShotSpotter alerts from January to November 2011,crime reports were written in only 50 cases.

"In many cases, officers will respond to find nothing to follow-up on: no one to speak with - victim, witnesses - and no physical evidence," Wood says. "When there is no additional information, no report is written."

The 50 crime reports resulted in six arrests, three on firearm-related charges, police say.

The sheriff in Broward County, Florida, decided last year to dismantle the county's ShotSpotter system because, he said, it wasn't cost effective. Officers, he said, were responding to too many false alarms.

"It was great technology," he said in the Sun Sentinel. "But it just needed some more fine-tuning."

But Wood says that just because a body doesn't turn up immediately doesn't mean a shooting didn't happen or that the ShotSpotter recording won't prove useful later. While it could be that the system is not performing as it should due to weather, construction noise, or other factors, he says, it could also mean that witnesses to the shooting are reluctant to talk to police. And the recording could help police piece together a case later on if, for example, a shooting victim turns up at a local hospital, Wood says.

Wood says the system is effective, although the efficacy can be tough to measure.

"I guess it's how you define it," he says. "It is a tool for us? Does it help us? Yes. If you ask certain patrol officers, sometimes it might be a bit of a burden because there are calls that - it's like the repeat burglar alarm that is nothing."

Technology can be seductive for law enforcement. And even in this era of cash-strapped municipalities, officials are usually reluctant to cut police budgets too steeply.

"This thing will give you information about shots fired," says one criminal justice expert who asked not to be identified. "How can you say no to that? It's tough to resist."

Police and other criminal justice experts say the way ShotSpotter can be useful is to identify trends and "hot spots" for shootings. And that data should trigger other efforts, they say, like increased police patrols and the involvement of groups like Pathways to Peace, a youth violence prevention and intervention group.

ShotSpotter alerts may not immediately lead to arrests, but they may contribute to a body of information that's helpful: that ultimately saves somebody or produces an arrest.

The mistake, experts say, would be to look at the alerts in isolation, "thinking that a single shot is going to tell you everything you want to know."

Wood agrees. RPD crime analysts periodically review ShotSpotter data, he says, for possible hot spots of shooting activity. Officers in those areas are then sent out to gather intelligence on what may be happening there, Wood says.