How Game Theory is Reinventing Crime Fighting
Elected officials across the nation from both political parties have begun to examine ways to replace a tough corrections policy with a smart one.
Three years ago, a group of conservative legislators from California slipped off to Texas. Among the purposes of their visit was to learn more about a new approach to controlling crime. The strategy involved investing in community corrections, not new prisons. The somewhat surprising thing was that the plan had been developed in Texas, with strong support from conservatives. Texas, after all, is a state that prides itself on being tough on crime. It executes more inmates than any other state and incarcerates the highest percentage of its population of any big state.
For two decades starting in 1985, Texas had built prisons with gusto, increasing by 300 percent the number of inmate beds. But in 2007, when Gov. Rick Perry produced a budget that asked the Legislature to appropriate $523 million in additional funding for three new prisons -- with more prisons to follow -- legislators balked. Instead, lawmakers decided to invest $240 million in diversion and treatment. By all accounts, this approach has been working. There have been declines in ongoing crime. Parole violations have plummeted. Prison overcrowding has eased.
Texas’ success intrigued the California delegation, but it didn’t inspire them to follow suit. Facing a strong prison workers’ union, opposition from district attorneys and a general unwillingness to relinquish the one tool -- being tough on crime -- that had worked for the GOP in the Golden State, the Californians listened but left with no game plan. “I think they honestly wanted to get something done, but they really felt they couldn’t do anything,” says Texas Rep. Jerry Madden, who was at the meeting as one of the architects of corrections reform in his state. “There were too many other influences they had in their system. It was almost an impossible situation for them.”
Today, California’s corrections system is a trainwreck. The state prison system is so overcrowded that the U.S. Supreme Court recently ruled that conditions violated the Constitution’s 8th Amendment ban on cruel and unusual punishment. Unable to balance its budget, California is currently in the process of shipping 40,000 state inmates to county jails. Texas, meanwhile, has become a model for corrections reform. Last year, at least 11 states, including Arkansas, Kentucky, Ohio and North Carolina, undertook similar sweeping corrections reforms with the intention of limiting the growth of their prison populations. This year, states as diverse as Georgia, Oklahoma, Missouri and Hawaii are expected to take up corrections reform based on ideas that have played out successfully in Texas.
“The Texas story helped spawn a wave of reforms around the country,” says Adam Gelb, who directs Pew’s Public Safety Performance Project. “We hear over and over, ‘If Texas can do this, [the approach] can’t possibly be soft on crime.’”
Cost clearly has been a major impetus for reform. Between 1985 and 2008, state prison populations nearly tripled. According to the Vera Institute of Justice, corrections spending rose even faster, by more than 600 percent. It now makes up 7 percent of state general fund spending. But cutting costs is only part of the story. Ideas matter too. When crime began to spike in the 1960s, criminologists and public policy experts responded with a simple and compelling proposition: Lock away more people for longer. Today, new ways of thinking about public safety -- some of them rooted in game theory, behavioral economics and sociology -- are challenging the perceived wisdom about how to improve public safety and reduce incarceration rates.
Game theory seeks to understand what constitutes a rational course of action in situations where other people’s responses determine outcomes. For decades, academic game theorists have explored how promises, commitments, threats, the elimination of options, and other tactics can affect outcomes and the resulting “equilibrium.” In Texas and in a growing number of states and cities across the country, policymakers have found a smarter approach based on a new generation of research that applies insights from the world of game theory to the criminal justice system. It’s still a very new concept, but the resulting body of work is pointing policymakers toward new and potentially transformative ways of improving public safety while reducing the number of people behind bars. It also grapples with one of the most notable -- and appalling -- features of what some have called the current era of mass incarceration: its destructive effect on many African-American communities.
“Our crime rates have been dropping for nearly 20 years,” says Madden, “but we still have a greater demand for prisons. Why is this?”
A number of cities and states are asking the same thing. In response, elected officials across the nation from both political parties have begun to examine ways to replace a “tough” corrections policy with a “smart” one.
The U.S. incarceration rate -- 700 state and federal prisoners for every 100,000 residents -- is by far the highest rate of incarceration in the developed world. It wasn’t always so. In the first half of the 20th century, U.S. incarceration rates had hovered around 110 per 100,000 residents. Then came the crime explosion of the 1960s. In 1962, the United States experienced 4.2 murders per 100,000 residents. By 1964, the homicide rate had climbed to 6.4, and by 1972, it was 9.4. Robberies were even worse. In 1959, the rate was 51.2 for every 100,000 residents; by 1968, the rate had nearly tripled. Crime, the political scientist James Q. Wilson concluded, “had assumed epidemic proportions.”
Wilson looked primarily to cultural changes to explain this explosion in criminality. However, he posited another cause as well, as did the University of Chicago economist Gary Becker: Crime was rising because the risk of punishment was falling. Crime was up but the number of prison beds was down. By 1974, the “average” punishment per committed burglary was four days of incarceration. The average punishment per committed aggravated assault was eight; for robbery, 28. In short, crime increasingly “paid.” That gave rise to a straightforward solution: Incarcerate more people for longer periods of time. States went on a prison-building spree.
By 1996, the United States had regained the level of punitiveness (as calculated by dividing crimes committed by punishment given out) of 1962, the year before the crime spike of the 1960s began. Crime, meanwhile, had begun to fall. Although academics differ on the details, most agree that increased imprisonment played a significant role in the lowering of crime numbers. But as the crime rate declined, something odd occurred. By 2008, some 2.3 million Americans were in prison or jail, 1 percent of the adult population. Forty percent of the inmates were black. What had happened?
Ohio State University law professor Michelle Alexander offers at least one answer: the misguided war on drugs. In her book, The New Jim Crow: Mass Incarceration in the Age of Colorblindness, she notes that “drug offenses alone account for two-thirds of the rise in the federal inmate population and more than half of the rise in state prisoners between 1985 and 2000.” As a result, 500,000 of the people behind bars today are serving time for drug offenses -- versus fewer than 50,000 in 1980. And that, she argues, has had implications for African-Americans. “Nothing,” she writes, “has contributed more to the systematic mass incarceration of people of color in the United States than the War on Drugs.”
The panic surrounding the emergence of crack cocaine in the 1980s was, she suggests, a media-created phenomenon. Had the war on drugs really been about drugs, then “the drug war could have been waged primarily in overwhelmingly white suburbs or on college campuses. SWAT teams could have rappelled from helicopters in gated suburban communities and raided the homes of high school lacrosse players known for hosting coke and ecstasy parties after their games.” That it did not, she says, reveals the truth about the drug war: “The War on Drugs, cloaked in race-neutral language, offered whites opposed to racial reform a unique opportunity to express their hostility toward blacks and black progress, without being exposed to the charge of racism.”
There is some truth to Alexander’s interpretation. When crack first appeared, some of the reportage was overstated and overwrought. But there also was unprecedented violence associated with the street markets where crack was sold. In the new book, Don’t Shoot: One Man, A Street Fellowship, and the End of Violence in Inner-City America, author David Kennedy, a professor at the John Jay College of Criminal Justice, strongly disputes the claim that the “moral panic” sparked by crack was overblown. Some people, Kennedy writes, believe “it was never really that bad, that the public and political and law enforcement response was just a fevered overreaction.” The reality, he says, is that the crisis was even worse than most people realize.
Kennedy got a first-hand look at the crack market 25 years ago, at Nickerson Gardens, a 1,000-unit housing project in Los Angeles. “I’ve never been so scared before or since,” writes Kennedy of the day spent walking through the project with two police patrolmen. “My lizard hindbrain knew instantly that if they were somehow magicked away all that would ever be found of me would be my bleached bones.” (In fact, he later realized, the dealers had been scared of him; they saw a white guy in the projects in a suit and thought, “Fed.”) To Kennedy, what was unfolding in the Nickerson Gardens of America was like “the end of the world.”
Kennedy shares Alexander’s belief that, with or without the war on drugs, something has gone terribly wrong with the way America polices minority communities. “We are destroying the village in order to save it,” he writes. But where Alexander sees racism, Kennedy sees misunderstanding. The notion that crack was created by the government, a persistant belief among parts of the black community, is wrong, he says. The belief among cops that residents of America’s most dangerous communities are “uncaring, complicit, corrupt, destroyed” is also wrong.
What’s right, he believes, is an approach that he and a group of police, probation officers and others developed in Boston in the mid-1990s. Instead of targeting drugs or gangs, Boston targeted violence.
The shift in focus happened almost by accident. Kennedy and two other researchers wanted to work with the Boston Police Department to understand and disrupt youth access to guns. The department brass sent them to talk to the anti-gang unit. There they heard stories about an amazing approach in Dorchester called Operation Scrap Iron. It had started as a conventional collaboration between the Boston Police Department; the federal Bureau of Alcohol, Tobacco, and Firearms; and the U.S. Attorney’s Office. The agencies targeted “straw” buyers who bought guns and then sold them to underage kids. But by the end of the operation, neighbor gang members were voluntarily dropping guns off at the anti-gang unit. Even more extraordinary, they were going to the police to report when other gangs were hoarding guns. What was happening?
Kennedy realized that the authorities -- gang unit officers, probation officers and Boston’s so-called street workers (in effect, former gang members turned social workers) -- already knew who the small subset of truly violent gang members were. Reducing gang violence wasn’t about ending poverty, shrinking the number of guns or some other huge social issue; it was about getting to this small group of bad actors. What the police had done during Operation Scrap Iron was to talk directly to this group. They’d told the most violent gang, the Wendover Street crew, that everyone would be all over them until the violence stopped. Then they made good on the threat, using every tool possible -- probation checks, public drinking arrests, drug tests, curfew enforcement, area and association restrictions -- until the violence stopped. By the end, the gang was so eager to return to the status quo that they were coming to police when other groups threatened them so that they wouldn’t get violent. Wendover Street had started policing itself.
It didn’t last. Still, Kennedy was intrigued. Gang members did not enjoy a reputation among academics as rational actors. Economists had long since established that crime doesn’t pay very well. (Contrary to the popular imagination, drug dealers selling cocaine earn sums equivalent to working elsewhere at the minimum wage.) People who committed crime, economists theorized, had poor self-control and short time horizons. But Scrap Iron suggested that when authorities delivered a clear warning that certain, specific activities were off limits -- not “don’t be in a gang” or “don’t sell drugs” but rather “don’t use guns to kill people” -- and followed up on the threat, gangs would listen and comply.
As an experiment, the group Kennedy was working with decided to try it with one of the city’s most violent gangs, the Vamp Hill Kings. Representatives from the police, probation agency, the U.S. Attorney’s Office, and the Suffolk County District Attorney’s Office convened a forum with gang members (who were brought into a courthouse by street workers and probation officers) and explained in respectful tones the new rules of the game -- no violence or else. To underscore the threat, the group pointed to the fate of Freddie Cardoza, a member of another gang, the Humboldt Raiders. Cardoza, whom Kennedy describes as “pretty much the city’s worst badass,” had ignored a similar message. Officers stopped him and found a single bullet in his pocket, a no-no for a convicted felon. Cardoza was sent to prison for 15 years. The room went quiet. Then Boston went quiet.
Boston’s crime decline made Kennedy a star. In the years that followed, Kennedy has used this same basic intervention in cities nationwide. But UCLA professor Mark Kleiman argues that the principles Kennedy has applied in the field have a broader application to the criminal justice system as a whole.
Economists who study crime often frame their discussions with a simple equation: The social cost of crime equals the cost of crime plus the cost of crime control. Framing it this way makes an important point. Arrests, prosecutions, incarceration -- the things that many public-sector managers spend their time measuring -- are actually costs. The benefit is crime reduction. Framing matters thusly encourages policymakers to consider questions such as the following: First, how much crime control is too much? Second, what if we could achieve the same level of control at a lower cost?
In his book, When Brute Force Fails: How to Have Less Crime and Less Punishment, Kleiman makes a compelling case that the answer to the second question is that we can. Kennedy’s interventions illustrate the three ideas central to Kleiman’s proposed approach: concentration on the worst offenders, substitution of swift and certain punishment in place of severe punishment, and direct communication of deterrent threats. Central to Kleiman’s work is the insight that “swift and certain punishment, even if not severe, will control the vast bulk of offending behavior.”
Hawaii’s Opportunity Probation with Enforcement (HOPE) experiment offers a striking example of how this works. Started in 2004 by Judge Steven Alm, HOPE identifies probationers at high risk of reoffending -- probation and parole violations account for about a third of prison admissions -- and warns them that they will be subjected to frequent, randomized drug testing. Positive tests will result in an immediate but short return to jail, sometimes simply for the weekend. The results have been dramatic: HOPE probationers are 55 percent less likely to be arrested for a new crime and 53 percent less likely to have their probation revoked than parolees in the control group.
What is particularly striking about HOPE is how few resources are needed to enforce compliance. In game theory terms, Hawaii has moved from a high-violation, high-punishment equilibrium to a low-violation, low-punishment equilibrium. In short, it tipped, just as it had in Boston and other cities where Kennedy-style intervention operations worked. Not surprisingly, the National Institute of Justice recently announced plans to replicate HOPE. Programs are getting under way in Clackamas County, Ore.; Essex County, Mass.; Saline County, Ark.; and Tarrant County, Texas.
Kleiman also emphasizes the importance of concentration. “Concentration,” he finds, “outperforms equal-opportunity sanctioning.” By identifying the worst offenders, warning them of the new zero-tolerance rules for certain activities and then enforcing the policy, authorities can “tip” violations downward. As violations within the targeted group fall, authorities can then extend the zero-tolerance policy to new groups of potential offenders.
Strategies that combine direct communication, concentration of resources, and swift and certain sanctions clearly work (though questions remain about exactly how they work). Kennedy and Kleiman make a convincing argument that not using these tools is itself a choice -- and not a good one. A growing number of policymakers seem to agree. After his disappointing meeting with California Republicans, Madden recalls thinking, “I will know I have made it when I get invited to talk in California.” He got his invitation last October -- to speak at an event with Gov. Jerry Brown.
The message he would deliver to any public official interested in following the Texas model is to figure out what your desired results are, measure how well you are getting there and require that the programs show their results toward that goal. “If you could take a rate of return to prison that was 30 percent for a certain type of criminal,” he says, “and reduce that to 20 percent -- if you can put a program out there that can keep people from committing serious offenses -- then you can have a huge impact.”
Join the Discussion
After you comment, click Post. You can enter an anonymous Display Name or connect to a social profile.