Re:Predictive Policing Software Is Pretty Terrible at Predic...
by tech_ken on Thursday October 05, 2023 @09:19AM
(#37778225)
Check out “routine action theory” (which I came across due to the above meta-analysis, can’t explain why it 403d for you, paper is Pratt & Cullen 2005). Posits crime as occurring when three conditions are met:
* Motivated perpetrator
* Vulnerable victim
* Lack of guardianship
Presence of police reduces factor 3, increase in socioeconomic deprivation increases factor 1. It’s unclear to me how you can confidently make the statement that “more police is the most effective” as that doesn’t seem to be attested to anywhere in that study (though I can only read the abstract), however thinking about the balance of these three factors brings our two arguments somewhat into line.
Not sure what kind of meta review was occurring in your link, but there's lots of evidence that adding more police would reduced crime in US cities: https://direct.mit.edu/rest/article-abstract/100/1/167/58429...
Basically every type of crime rate goes down for every cop you add, with murder going down the most. You might be arguing that for a fixed level of policing, most criminal behavior is explained by desperation, but adjusting the level of policing is the most effective way to change the level of crime.
This also intuitively checks out: you could have ~0 crime in a dystopian police state with high levels of desperation, but you would definitely have crime in a police-free state with no desperation (lots of well-off, rich people commit crimes!).