New Metropolitan Police database raises concerns of Gangs Matrix repeat

Lessons learnt?

“The Gangs Matrix – which labels young people as gang members based on vague criteria and untested evidence, conflates victims and perpetrators, and entrenches discrimination in public services – should serve as a cautionary tale,” said Rosalind Comyn, policy and campaigns manager at Liberty.

“Support services for young people play a critical role in addressing the root causes of serious violence, but police-led initiatives characterised by widespread data sharing in the absence of robust safeguards risk increasing the surveillance of young people, and may have debilitating effects in terms of their access to housing, education and employment.”

The Met, however, claims that although the database is in its infancy: “Individuals identified as being at risk will be provided support and pathways away from violence through partnership working with local authorities and a range of initiatives.”

Ffrench, on the other hand, has doubts about how early stages the database is. “It can’t be at such an early stage that they’re going to launch it next month,” she said. “There’s only a week and a half left of this month, so they say it’s early stages but they knew from when Amnesty published that report [in May 2018] that the Matrix was dead.”

Following the official launch of the first hub in April, the plan is to have each local authority managing a hub with a central team at New Scotland Yard providing oversight of the whole hub system.

Police use of data, algorithms and machine learning

Closely tied to the use of police data is the idea of predictive policing, whereby the information is taken to predict future behaviours.

In the context of policing, this usually means taking the data and running it through a machine learning tool to do one of two things.

The first is to take the data and map out crime “hotspots” in an attempt to figure out where new crimes will be committed, while the second is largely focused on profiling individuals to try and predict their “risk” of offending.

The Gangs Matrix, for example, was originally set up in the wake of the 2011 Tottenham Riots to identify and monitor individuals considered to be linked to gangs.

So-called “gang nominals” (individuals on the Matrix) were given algorithmically generated “harm scores” to show how likely they are to commit a violent crime, with just two pieces of corroborating evidence needed to justify their inclusion.

This could range from what estate the individual lived on to who they were friends with on Facebook or what music videos they had shared on social media.

Although senior government officials at the time, including former prime minister David Cameron, asserted that gangs had orchestrated the riots, no evidence has ever been produced to confirm this claim.

Matrix seen as discriminatory tool

According to veteran racial equality activist Stafford Scott, writing in the The Guardian in January 2019, “there now appears to be a slowly growing recognition, even among regulators, that the Matrix is, and has been used as, a discriminatory tool in the Met’s so-called war on gangs”.

“The Met claims the Matrix, informed by intelligence, helps identify and assess the most harmful gang members in each of London’s boroughs,” he said. “But those of us who have witnessed the impact of the Matrix know this is far from the truth.”

The Met is not the only police force in the UK to be using these kinds of tools. A recent report by Liberty, Policing by machine, highlights at least 14 forces using similar tools.

Durham Police, for example, have used a program called the Harm Assessment Risk Tool, which bases its predictions on 34 pieces of data.

Of these, 29 relate to the individual’s previous criminal history, while the rest include characteristics such as postcode, which, according to the Liberty report, “act as ‘proxies for race’ by indirectly indicating a person’s ethnicity and fuelling the same biases”.

The report added that research on algorithms in the United States’ criminal justice system shows that even if no race-related data is included in an algorithm, it still learns characteristics in a discriminatory way, “because other pieces of data it did use correlated with race and led to inadvertent profiling by the algorithm”.