When can a theory kill people? When a theory leads to policies that increase the causes of death rather than reducing them. That is what London officials did with the cholera epidemic in the 19th Century. And it may be just what America is doing with the problem of homicide in the 21st.

click on image to enlarge

The cholera epidemic of 1848-49 was instigated by a plan any bio-terrorist would love: systematic poisoning of the water supply. The theory was that disease was caused by “miasma:” bad smells from human waste. The rapidly expanding population of London disposed of human waste by simply dumping chamber pots into the cellar. Smells grew so offensive that public health leaders resolved to achieve reform in waste removal. But not knowing the germ theory of disease, their reform was terribly misguided. Where London once had banned the deposit of waste into drainage sewers, such deposits were now required. Where the Thames had been clear enough in London to support a fishing industry, the river became clouded with filth. Where drinking water from the Thames that was piped to homes had been clean, it suddenly teemed with bacteria.

On the eve of a cholera epidemic, the Metropolitan Commission on sewers proudly announced that 30,000 cesspools had been abolished and the waste flushed directly into the river. Their inhabitants were now – in theory – safer from the allegedly infectious “miasma” of smell around those cesspools. What resulted was the rapid onslaught of 15,000 deaths, as Steven Johnson recently pointed out in The Ghost Map.

As homicide rates now rise across many U.S. cities, we may well ask if it is being caused by a theory wrongly accepted as “obviously” true. Imprisonment is the miasma theory of crime, and the “broken windows” policy is the equivalent of dumping waste in the drinking water. Since the 1970s, many academics and public officials have embraced the theory that lack of punishment was a major cause of crime. Consequently, our incarceration rates have climbed to the highest levels in the world. Our homicide rates, however, have risen, fallen, and risen again, with no clear connection to rates of imprisonment.

Police have tried to fight crime, and especially drug abuse and sales, by making many more arrests. This policy was based on an interpretation of the “broken windows” theory that little crimes lead to big ones, so that “zero tolerance” for little crime is needed to reduce big crimes. But like shifting human waste from basement cesspools into the drinking water, this policy is only half right.

What zero tolerance did was detect and deter people who carry guns illegally in public places. Evidence in seven tests reviewed by the National Academy of Sciences suggests that focused gun detection reduces homicide and gunshot wounds. This “gun-borne” theory of homicide applies not only to neighborhoods, but also to specific individuals. As University of Pennsylvania criminologist Richard Berk has found, a tiny handful (2%) of all Philadelphia adults on probation or parole will kill someone, accounting for as many as half of all homicides. These potential killers can be predicted with great accuracy based solely on their prior history of prosecution. Making sure these few people do not carry guns, and are provided with maximum rehabilitation and mental health services, could be our most cost-effective solution to rising homicide rates.

There is no evidence, however, that the reduction of little crimes, in general, leads to less crime. A $40 million MacArthur Foundation/NIJ study of videotaped samples of street behavior from thousands of Chicago street blocks found no correlation between little crime and big crime, as Harvard sociologist Robert Sampson has reported. But Sampson’s work has been widely ignored by policymakers.

Street disorder in poverty areas, like London’s smelly “miasma,” is offensive to local residents, who cry out for action. But simply answering those calls and making more arrests only reduces the purity of justice. Zero tolerance policies swamp the courts, leaving prosecutors and judges unable to convict many of the guilty. Trivial and serious offenses alike go unpunished, with many of the most dangerous people avoiding prison. At the same time, many other people do get sent on to prison, with fewer parole officers to support them on re-entry to the community. Spreading punishment so widely leaves community corrections too thin to make a difference, although even those resources can be concentrated on parolees most likely to kill.

The facts show that most murders are committed by people, and at places, that can be predicted by careful analysis of criminal justice records. Policies of treatment and surveillance that focus on those high-risk people and places might, in theory, reduce the homicide rate. The “broken windows” theory that increases incarceration rates indiscriminately across poor areas might, in fact, only increase the number of murders. Police chiefs have responded to recent homicide increases by asking for more police officers. Unless police patrols are directed with precise mapping of gun crimes, more police and arrests may only cause more crime.

These theories urgently need further testing. The first thing a new Congress should do is not to fund more police, but more tests of homicide reduction strategies. When policies based on a prominent theory are followed by an epidemic, it is time to gather more facts. Nothing is more dangerous than a well-intentioned theory that defies the facts.

Lawrence W. Sherman is Wolfson Professor of Criminology at Cambridge University and Director of the Jerry Lee Center of Criminology at the University of Pennsylvania.