The consensus solution emerging out of the tragic police killings of Michael Brown, Tamir Rice, and Eric Gardner involves equipping police with body-worn cameras. The federal government has offered to provide millions of dollars to support the technology. Without other obvious options, cameras have become the accepted partial solution to an obviously complicated social, racial, and political problem.
Cameras alone, however, cannot identify and track those officers who are most at risk for using excessive force. Cameras provide the necessary transparency after a tragic incident occurs, but cannot provide the warning signs necessary to prevent it. Instead, police administrators must turn predictive analytics already in use to track crime toward the problem of identifying police officers most at risk for using excessive force.
As I wrote before the Darren Wilson grand jury decision "new crime mapping and predictive technologies ordinarily used to prevent crime can be adapted to identify patterns of problematic police-citizen interaction." The key is recognizing that the insights of predictive analytics -- identification, pattern matching, and locational tracking -- can be used to study police-citizen interactions. Predictive policing is a two-way street. The technology offers great new ways to target criminal activity, but also can be quite useful for internal police accountability.
Big data analytics have already begun to revolutionize policing. Predictive policing forecasts potential hotspots of crime. Big data algorithms identify individuals most likely to engage in criminal or gang activity. Patterns of suspicious purchases trigger investigations of the purchasers. Networked databases catalogue suspicious people and activity, and combined with facial recognition and biometric technology portend new investigation possibilities for police. Each of these innovations is based on a simple principle: past data can identify useful patterns correlated with criminal activity.
This type of technology can also be used to identify police officers who show a pattern of problematic decision-making. Daniel Pantaleo, the police officer involved in the choking death of Eric Gardner, has been involved in three prior civil rights investigations. Timothy Loehman, the Cleveland police officer who shot Tamir Rice, had recently left his last police employer after a devastating performance review that called into question his emotional stability when handling guns. These complaint reports are just the formal professional or public court records available. A daily tracking of their actual police performance might well have exposed a pattern about these officers' willingness to use force in effectuating their duties. A systemic evaluation of their policing tactics might have raised a red-flag for police administrators.
The key is to recognize that video can be mined not just to investigate a tragic use of deadly force, but also expose all of the smaller, more telling police citizen interactions before that time. If properly designed, accountability systems can be structured to identify police behavior that correlates with excessive use of force. Smarter body-worn cameras can be equipped with GPS locational mapping, voice recording, and pattern recognition algorithms that can isolate heated interactions, uses of force, weapons, and complaints. With such technology in place, a reviewing supervisor could immediately call up all moments of force in any particular day. Key words could be searched to determine if inappropriate language was being used. Complaints could be checked immediately after any report. A city-wide database of violent contacts could be developed and monitored. Particular police districts or officers with disproportionate patterns of violent contacts could be identified before a tragedy occurs.
Simply put, violent contacts can be mapped in the same way as violent crimes, such that predictions of problematic locations and at-risk officers can be studied.
Whatever one thinks of police tactics in use today, significant damage has been done to the trust that lies at the heart of any police system in a democratic republic. Local police administrators know this reality. The unfortunate part of the recent upheaval is that all police officers are being judged by the actions of a few. Most police officers, even in jurisdictions with a climate of police-citizen tension, do their job professionally and without incident. New technology that provides video and predictive analytics can validate that professional practice, while exposing those who are undertaking more combative or excessive behavior. Police administrators should embrace such internal accountability measures as a way to regain the trust of the community and retain local control.
Police have used fixed surveillance cameras to deter crime for decades now. Those surveillance technologies are evolving to include searchable archives, facial and biometric identification, and networked accessibility. But, these cameras are still reactive. The insight of modern policing is to take that data and make predictive judgments in order to minimize risk. Police have come to understand that fixed recording of past crime is less useful than predictive targeting to prevent crime. That same principle should be considered in designing the body-worn police cameras for police. The goal should not simply be to record the next tragic encounter between a police officer and a citizen, but to prevent it before it happens. Data driven tracking systems can be created to make that goal a reality. And, while there are real privacy costs associated with such cameras and technology (for police officers and the public), the accountability benefits outweigh the concerns.
As companies race to chase the federal money flowing into new police body-worn camera systems, predictive analytics should be a major consideration. Both police and communities should welcome smarter cameras that will ultimately lead to smarter policing.