Who polices the police? The perennial question usually refers to the use of force, or controversial stop-and-frisk policies. Now, policing’s “data revolution” is under scrutiny.

A new report, published last week by Amnesty International UK, excoriated London’s Metropolitan Police for compiling a “racially discriminatory” database on some 3,806 Londoners considered gang members or associates. According to Amnesty, the police are sharing their names with housing, education and employment services without their consent, harming their access to support in the process. The report called for an immediate investigation and an end to the database.

But data collection is now a valuable tool for police forces across the world. Violent crime clusters around prolific offenders and specific locations, and data analysis lets law enforcement put resources where they are most needed. How can police forces balance the need for data with the rights of citizens?

Dirty data

Amnesty charged London’s police with three key failings. First is racial profiling: 78% of those listed on the database, known as the Gangs Matrix, are black, but only 27% of serious youth violence is committed by black offenders.

Second is the legality of how that data was gathered. Amnesty found officers befriending young people on social media without appropriate warrants. It also criticised the criteria for being listed: listening to UK rap was one particularly egregious criterion among many.

And third is the question of data storage and protection. “Police are failing to adhere to data protection laws,” said Olly Sprague, Amnesty’s policing expert. The report uncovered inconsistent data protection practices between London boroughs, and widespread information sharing between police forces, housing services and job centres. While the police claimed that sharing information was intended for “safeguarding” purposes, Sprague said Amnesty found “zero evidence” of where the Matrix had been used for that end. Instead, information sharing “impacted people’s lives in an immensely damaging way,” said Sprague, making it harder for those listed to find housing and deterring schools from accepting students listed on the Matrix.

“Police are failing to adhere to data protection laws”

The report focused exclusively on the failings of British policing, but the critiques could apply much more widely. Data was first used to map crimes and develop policing strategies in New York in 1995 through a program called CompStat, which is still in use. Now, data is being used to predict offending in at least 20 of the US’ largest police forces, and in many more countries across South America and Europe.

A statement by a coalition of civil rights bodies in 2016, including the American Civil Liberties Union, decried the unchecked rise of data-driven policing in the US as “limited and biased”, emphasising the unreliability of police data, and the biases that creep in at the point of collection.

To some, the problems of racial profiling and unregulated data collection might pale in comparison to the urgency of getting a handle on violent crime. But according to Ben Bradford, UCL’s professor of global policing, any practice which undermines people’s trust in the police can have dire consequences.

Communities who don’t trust the police are “less likely to report crime, less likely to provide information, less likely to work with police in their communities to solve problems, and they may be more likely to commit crime as a direct result of that lack of legitimacy,” he said. The damage that can be caused by irresponsible data sharing is just the latest cause for concern.

Restoring the faith

The tensions between law enforcement’s need for surveillance and the privacy rights of citizens are unlikely to be resolved completely, but some measures can mitigate the harms to communities who feel over-surveilled but under-protected. Action is needed on two fronts: first, to ensure the lawful and ethical use of data once it is collected; and second, to make the process of information gathering itself more consensual.

On the first axis, one approach is to use independent watchdogs to bring policing into line with existing data protection regulations. The UK’s new Centre for Data Ethics and Innovation might prove a valuable example of this, so long as it has teeth to pursue malpractice.

A second is to demand greater transparency in how data and technology are being used by police forces. Algorithmic accountability is one new tactic, whereby the workings of automated decision-making systems are made available for public scrutiny.

To improve the way the police gathers data is a more challenging task by all accounts. Police data is not crime data: the former is based on the number of arrests or crime reports, not the actual number of crimes committed. Police data rests on where police are deployed, who they arrest and who they process — all of which introduce human bias into the equation, cementing the racial disparities found in databases, stop-and-frisks, arrests and convictions.

“The only way the police get intelligence which is useful is by working with the community”

Insofar as these tactics increase suspicion of police forces, they force officers to gather intelligence by scrutinising communities from the outside. Rebuilding trust would allow engagement and collaboration from within.

According to Matteo Tiratelli, a policing researcher at the University of Manchester, the solution to this breed of data problem is resolutely analogue. “The only way the police get intelligence which is useful is by working with the community,” he said. The police should act as a trusted outlet for residents to speak to them and design solutions to local problems, which could reduce the need for external surveillance.

Involving the community in policing has worked in countless settings across the world. In the Brazilian city of Belo Horizonte, creating police task forces that worked alongside community groups helped cut crime by 69%. In the Los Angeles neighbourhood of Watts, gaining the community's trust is at the centre of a new project which includes a five-year residency requirement for sworn officers and co-screening of new officers by communities.

Trust, Bradford and Tiratelli argue, is the bedrock of effective policing. If communities trust their officers, they are more likely to provide tip-offs, lessening the need for the police to engage in questionably lawful snooping. Data-driven and tech-savvy policing offers new possibilities for cutting crime worldwide, but offer no replacement for the importance of human interactions between officers and the communities they serve.

(Picture credit: Flickr/ep_jhu)

Edward Siddons
edward.siddons@apolitical.co