Policing the Poor

Technology has a way of perpetuating biases that have existed since the analog age. Poor neighborhoods — highly policed with bad schools and few opportunities — are the main target of the software being used to “reduce crime.” However, that same software threatens to land poor people in a constant cycle of surveillance, law enforcement interaction and high rates of imprisonment.

In neighborhoods with high poverty rates, many police departments have turned to crime prediction software to help them target those areas in which crimes are the most prevalent. Historical data, such as when crimes occur and where, are fed into software like PredPol. PredPol takes crime data and divides it into two types; part one crimes, which are violent crimes such as homicide and assault; and part two crimes which includes vagrancy, panhandling and consuming small quantities of drugs.

PredPol is designed to be blind to race and ethnicity. This is true because we are not attaching social and racial data to the numbers being fed to the application. Instead proxies are used. Proxies are used in place of actual data. For instance, information such as zip codes can be used to identify race and income, based on statistics on the area. Geographical data is a perfect stand-in for the information to be identified.

Part 1 crimes are typically combined with part 2 crimes. Since part 2 crimes are mostly associated in poor neighborhoods, the number of instances in a particular area are skewed. Part 2 crimes are considered “nuisance crimes” and some police call these ASBs (Anti-Social Behaviors). These victimless actions would normally go unnoticed if the police did not personally see it taking place. No one is going to call the police for panhandling or vagrancy in an impoverished neighborhood. Combining the numbers make it good for police data; the statistics show that a high number of crimes are being solved and the areas are being policed well.

Since adding the two types of crimes increases the amount of reported criminal activity in the area, the data fed into the software comes up with a predictive model and more police are drawn into those areas. Some of the data being fed into PredPol incorporates the location of ATMs and convenience stores. The predictive model shows up as squares the size of two football fields on a map. The number of crimes in a particular area are displayed and policing is assigned to the squares with the highest rate of crime. The result is more arrests and more people end up in cycle of recidivism.

So why do they collect both types of crime data? It is because of the broken window theory. This is the idea that low-level crimes create an atmosphere of chaos. By controlling these crimes, it is believed the police can restore order in these neighborhoods. As the theory goes, fixing broken windows and cleaning graffiti covered walls will go a long way to diminish petty crimes, which lead to bigger crimes. But like many others, this solution created more problems.

Integrated data systems that track the poor — like social services, etc. — are also targeted for use by law enforcement without the consent or the knowledge of the people who supply that data. Operation Talon is an example of social services and law enforcement tag-teaming to criminalizing the poor. Operation Talon is a teaming up of the Office of Inspector General and local welfare offices that give food stamps. The food stamp data is mined to identify people with outstanding warrants. Those that have warrants are lured to appointments in regards to their benefits and are arrested once they show-up. Before welfare reforms of 1996, accessing welfare records required going through legal channels. Now those same records are available upon request without any probable cause, suspicion or process. It can be argued that this was a successful operation and it took criminals off the street. However, this type of effort between law enforcement and social services has no equivalent for people outside of the social service system. In other words, if you are poor, and apply for any type of service, being it for food, shelter, clothing, or are a victim of violence or drug addiction. The information you give while receiving these services is basically up-for-grabs for law enforcement.

According to Virginia Eubanks in her book, “Automating Inequality: How High-Tech Tools Profile, Police and Punish the Poor”, monitoring the poor more tightly is nothing new: “America’s poor and working class people have long been subject to evasive surveillance, midnight raids, and punitive public policy that increase the stigma of hardship and poverty.”

We would do ourselves a favor by imagining how this type of data collection and surveillance would work on white collar crime. White collar crime is not what we visualize when we think of criminality. However, financial crime has a devastating effect on our country. If we used the PredPol software to track financial shenanigans as we do for poor neighborhoods, by arresting people for insider trading, security and corporate fraud or cheating investors on 401k accounts, maybe the system may seem have more equity

The rich and powerful rarely go to jail for these types of crimes, except for a few high-profile cases in which other well-to-do people were cheated out of their money. So even though massive bank fraud lead to the crash of the financial industry, you can’t name a Wall Street Banker that went to jail for it. Law enforcement would need a whole new set of skills in order to track all of the minor transgressions of the financial industry. They would need to have an army of experts to tackle financial corruption. Transactions have to be monitored across networks and algorithms have to be configured to root out possible violations and impropriates in the financial world. There nothing set up for that type of surveillance. The FBI has a long a storied history of tracking check fraud, etc., but not on the scale that is set-up for the poor. Crime fighting was built for the impoverished areas.

Meanwhile the system in place that concentrates on poor neighborhoods are tearing up what little fabric is left that holds the neighborhoods together. It feeds distrust and animosity between the cops and the citizens of those areas, it perpetuates a story of violence, crime and drugs in these areas. And the distrust continues to grow. Low-income people face the lion’s share of the effects of high tech scrutiny. But everyone should be concerned, because what is used on the impoverished now, will be standard operating procedures for everyone in the not too distant future.