Believing that statistical likelihood should determine social policy is a ludicrous argument.
It is the only logical argument.
There are an estimated 385 million police interactions with the public each year. In 2014 there were 990 civilians killed by police officers. That’s .000002%. I would call that statistically insignificant. By your reasoning, it is so rare, we shouldn’t even look at it when drafting public policy on use of force by the police.
The statistical prevalence of police killings should absolutely be used as a factor when weighing different proposals for the reduction of police violence, because the statistics tells us how bad the problem is.
For instance, I have heard in certain anarcho-capitalist circles the policy proposal issued taht we should do away with all police. If there were 300,000 civilians killed by police each year, then I would absolutely consider that proposal. If there were 30 police killings a year, I absolutely would not. Which means that the reasonableness of that proposal exists on a continuum that is informed by the statistics.
Here’s an exercise for you. Back calculate the per capita number of children murdered by guns in school, and compare it against the per capita number of people murdered by guns in Japan. Compare the two rates. See what you get.
I already know the answer, but I want you to put forth the effort so you have some personal investment in the realization.