‘Red Flags’ as New Documents Point to Blind Spots of NYPD ‘Predictive Policing’
The New York Police Department, working mostly in secret with private companies and in-house data scientists, has developed its own crime-forecasting technology. A new tranche of previously unseen documents, provided exclusively to the Daily Beast, offer the deepest look to date at the NYPD's efforts at so-called predictive policing—a technique that independent researchers and academics have excoriated for “tech-washing” racially biased policing methods by using algorithmic analysis of flawed data to offer a veneer of objectivity to justify objectionable law enforcement tactics.
The efforts began in 2014, under former Police Commissioner Bill Bratton, who has labeled crime forecasting “the wave of the future,” and who had initiated one of the earliest and most advanced predictive policing programs during his term leading the Los Angeles Police Department.
Two private firms participating in a 2016 predictive policing pilot with the NYPD sought to use non-law enforcement socioeconomic data such as welfare records, renter and homeowner data, and Census records that have questionable utility in predicting crime. The department did not continue its relationship with the private-sector prediction companies over concerns about accuracy and forecasting methodology used by the firms and instead built out its own in-house prediction algorithms .