The risk of a false positive is nonzero, but how does it balance against the value of a true-negative? Given that alternatives the police have include "picking up anyone who just looks like they might be a perpetrator" (which all too often translates to racist policing), is there value to communities (especially communities with past racial policing concerns) of tools that tell the cops it's probably not any of the people the system doesn't identify (or identifies as affirmatively not in the area of interest)?