Because as the Zendesk example that started this pointed out, an entire building (probably multiple!) of people were affected by this incident. There was 1 victim. It's going to seem insignificant on a per capita basis. There's thousands of people impacted by it, and possible dozens in the immediate vicinity who could be suffering from ongoing trauma having witnessed it.
Crime is not random lottery. It is concentrated and specific to certain places and circumstances. Thus, pretending that it is a stochastic process that is well characterized by per capita number over the whole city, and that if more people move into a bedroom community 20 miles from you, you automatically become safer because per capita numbers decreased - is innumerate at best. If you are present where the crime is concentrated, you risk is high, regardless of how many people live 20 miles from you but still in the same administrative unit. This should be obvious but some people still insist on focusing exclusively on large-area per-capita numbers.
I don't think you're reading those numbers correctly. The highest crime per capita is in Alberque, New Mexico, the 32nd largest city in the US, and that list is literally the crime rates of the 100 most populated cities, not the 100 cities in the US with the most crime.
Which small cities, in your opinion, are more dangerous than, say, Albuquerque? How many chances, say, a high-tech professional has to regularly find oneself in such a city because his company offices are located there and they have no choice but to go there regularly?