Thanks for such a thoughtful response. On #5, it appears to be based on police reports, interviews, and corresponding docs (arrests, hospital admissions). The a

Thanks for such a thoughtful response. On #5, it appears to be based on police reports, interviews, and corresponding docs (arrests, hospital admissions). The arrests and hospital visit counts seem objective and the authors don't appear to have an agenda. I don't have any particular expertise or strong opinions about this topic. My interest is that the threshold/evidence requirement for telling people to avoid a certain resource should be set pretty high. Separately, the data confirms your #1: https://www.bjs.gov/content/pub/pdf/cpp15.pdf https://www.bjs.gov/content/pub/pdf/cpp15.pdf (https://news.ycombinator.com/item?id=23467379 https://news.ycombinator.com/item?id=23467379). The numerator is absolutely massive -- 53 million people per year. Roughly half of those are traffic stops, but even excluding those, it's still big. Citizens were asked the level of escalation of their most recent police-initiated contact or traffic accident and 0.3% said that a gun was pointed (page 17). That 0.3% is from citizens, not police, so it's probably an upper bound. For comparison, about 3% of people's most recent contact included any use of force, mostly handcuffs (1.8%) and push/grab/hit/kick (0.7%).