@sampw

Yup.

The number of people killed by police is not dependent on the crime rate. As in, more crime doesn't result in more people being killed, and less crime doesn't result in fewer people being killed.

There was more crime in the 2010s than 2020s.

More crime in the 2000s than 2010s.

More crime in the 1990s than 2000s.

And the 1980s were much more crime-y and violent than any decade since.

And yet, the number of people killed by cops is higher in the 2020s than 2010s, higher in 2010s than 2000s, etc.

And the reason why is obvious and awful, but hard to unsee, once you see it.

The number of people shot by cops in a year, is effectively a function of:
* Number of cop + citizen interactions per year
* Average probability of a cop + citizen interaction resulting in the cop fatally shooting the suspect

The average probability of a cop fatally shooting a suspect in a given year, can vary with hiring screening, training, policy, rules of engagement, and laws to enforce. But over large populations like "all law enforcement officers," But... this number doesn't change much year to year. 🤷🏿‍♂️

The number of cop + citizen interactions changes a lot per year. That number is largely a function of:
* number of cops
* laws to enforce
* ticket quotas

Suppose there were 400 million law enforcement interactions with innocent Black people in a given year, 400,000 of them used excessive force, and that 400 unarmed Black people were killed in that year.

Police chiefs tell me "👮🏻‍♂️Mekka, you have to admit, the data shows that we mostly do good! Only 1 in a 1000 bad experiences? Only 1 in a thousand of those bad experiences result in death? C'mon!"

To which I reply, "👴🏿Each death is someone's child. And most of those 400 million interactions, were unnecessary."

Based on the above numbers, it should become obvious why data scientists can predict if the number of people killed by cops will go up or down in a given year, based on police payroll budgets and policy changes.

Violent criminals are only ~1% of the population.

Insight: The violent crime rate doesn't affect the above numbers that much, because most police interactions are between a cop and a member of the 99% of non-violent, non-criminal people.