I'm not sure why this argument is persuasive. It would be like saying Palestine, OH is still safe because the risk of cancer per square mile is very good. If anything, we might naively expect that density itself causes homicide since proximity means opportunity. But on homicide incidence, SF (6.35)¹ tracks closely to the national average (7.8)².
Another argument posed in the article is that decriminalization is screwing with record keeping. But for homicides? Homicide is the most difficult to obscure.
---
[1]: https://en.wikipedia.org/wiki/List_of_United_States_cities_b...