(TN) With little fanfare and even less public transparency, at least 60 cities in the US and Europe have implemented crime forecasting systems. It’s time for a national discussion on police policies and citizen protections. ⁃ TN Editor
(WEF) What if you could predict where a crime will take place before it occurred, even determining the time of the incident and the identity of the culprit in advance? Sounds a bit like science fiction, right? Social scientists have long believed that historical crime trends often influence future patterns.
The revolution in advanced machine learning is putting their theories to the test. A new generation of crime forecasting tools is about to dramatically change the nature of law enforcement – and our privacy – forever. And it is more important than ever that we shine a light on the algorithms driving these innovations.
Lessons from one of the world’s most crime-affected cities
While still early days, at least 60 police departments in US and European cities have rolled rolling-out crime forecasting systems – with mixed results. Researchers are struggling to make sense of the outcomes. One of the problems is that police don’t like sharing what they’re doing with the public. Another is that it is exceedingly difficult to unpack the algorithms they use since they are proprietary. We simply don’t know what’s inside the black box.
In order to better understand the potential of crime prediction, we recently designed an open-source crime prediction platform for a particularly crime-affected city, Rio de Janeiro. On average roughly 1,000 people are murdered annually in the sprawling metropolis of 6 million. That’s twice the number of people murdered in all of Canada, a country with six times the population. And the crime situation has worsened over the past few years, though is not as terrible as many believe.
In Rio de Janeiro, like most cities around the world, it is hard to get a clear reading of security and safety. Public statistics are in not easily accessible and there are often long delays before they are released. Making matters worse, when local news outlets also run crime stories, they typically lead with sensationalist headlines that do more to spread fear than offer insight. It’s hardly a surprise, then, that 81% of the locals believe they are at risk of being murdered.
As in many cities around the world, Rio’s residents suffer from a dangerous knowledge gap when it comes to assessing the risk of crime. For example, few residents know that murder rates are still 50% of what they were a decade back. There is a worrisome disconnect between how people perceive criminal violence and its actual prevalence. This fear of crime dramatically affects the day-to-day decisions made by many of the city’s dwellers, including whether or not to turn to private security, gated communities or firearms for personal defense.
What seismology tells us about predicting crime
The good news is that violent crime can be reversed, or at least partly avoided. In many cities, the spread of smartphones, social media and data literacy is provoking changes in situational awareness and behaviour. Services like Amazon’s rating systems, Facebook, Instagram, Snapchat, Telegram and Yelp are creating a culture where people now expect to have access to data on every conceivable product and service – and shun offerings where data is unavailable.
What’s more, improvements in advanced machine learning mean that it’s possible to create much more accurate and targeted analytics to analyse crime dynamics, even in complex cities such as like Rio de Janeiro. Crime forecasting is just one example of this. It is based on the expectation that crime is hyper-concentrated in specific places and contagious among certain kinds of people.
The underlying mathematical models for crime prediction can be traced to an unlikely source – seismology. Very generally, crime is analogous to earthquakes: built-in features of the environment strongly influence associated aftershocks. For example, crimes associated with a particular nightclub, apartment block, or street corner can influence the intensity and spread of future criminal activity.