Santa Cruz, California bans predictive policing in U.S. first

> Are these neighborhoods “over policed” because their residents commit more crimes, or do the residents commit more crimes because their neighborhood is “over policed”?

Your question is flawed because it omits too many variables, including the racial differences in what is enforced as a crime, the endemic poverty and resulting tattered social fabric in high crime neighborhoods that pushes residents - especially younger ones - into situations where committing crime becomes an ever more "rational" choice.

Agreed, I don't think those neighborhoods just happen to have a bunch of violent people, but I keep seeing people make the naive agument that the difference in crime rate for those areas is caused by police presence. I remember after the protests in Baltimore that the police stopped doing as much proactive policing, and people who lived in the bad neighborhoods were complaining that the police weren't doing enough to prevent crime in their neighborhoods. There are underlying social issues that need to be dealt with that won't be fixed by removing police.

Is that global statistics or statistics for "over policed" neighbourhoods the article and discussion is about?

Minneapolis. Seoul.

“In the 1970s and early ’80s, we built 70 percent of our subsidized units in the wealthiest white districts,” Myron Orfield said. “The metro’s affordable-housing plan was one of the best in the country.”

"Only three large metros where at least half the homes are within reach for young middle-class families also finish in the top 10 in the Harvard-Berkeley mobility study: Salt Lake City, Pittsburgh, and Minneapolis–St. Paul. The last is particularly remarkable. The Minneapolis–St. Paul metro area is richer by median household income than Pittsburgh or Salt Lake City (or New York, or Chicago, or Los Angeles). Among residents under 35, the Twin Cities place in the top 10 for highest college-graduation rate, highest median earnings, and lowest poverty rate, according to the most recent census figures. And yet, according to the Center for Housing Policy, low-income families can rent a home and commute to work more affordably in Minneapolis–St. Paul than in all but one other major metro area (Washington, D.C.). Perhaps most impressive, the Twin Cities have the highest employment rate for 18-to-34-year-olds in the country."

"In 2008, Seoul imported a version of Minneapolis’s tax-sharing scheme. Since then, the gap in funding for social services among the city’s districts has narrowed. According to a 2012 analysis by Sun Ki Kwon, then a graduate student at the University of Kentucky, this has helped Seoul’s poorest communities grow their tax bases while only minimally affecting the city’s richest districts."

https://www.theatlantic.com/magazine/archive/2015/03/the-mir...

> There are tons of other stupid things I did when younger where if there had been more cops around I would have gone to jail.

That's what the person you replied to said. Not that more police would have been more arrests based on (just) their pot smoking, but on other behavior as well. I saw a lot of vandalism that led to kids being dragged by their ears back to their parents in middle-class neighborhoods, that would've resulted in arrests (even for misdemeanors) for kids in poor neighborhoods (where I've also lived and seen this difference).

In white neighborhoods small crimes would lead to police dragging the kid home, in minority neighborhoods they'd end up with a misdemeanor record (at least). All of this creates a bias in the algorithms which generally assume that someone with a criminal record of some sort is going to be a higher risk individual (which is probably a safe assumption for many crimes), but it is biased because people aren't charged with crimes at equal rates due to the biases in the existing system that these algorithms are more likely to exacerbate than alleviate. Now that the minority community has a thicker criminal record, the policing in that neighborhood continues to go up which continues to create increased conflict between the police and residents as the police end up arresting people for increasingly minor offenses, along with the major ones.

A lack of policing or difference in attitude of the police in poor and/or minority communities versus middle-class/wealthy and/or white communities is very much present in the US (and probably the world). This leads algorithms like those under discussion to end up inheriting the same or similar biases, whether the developers and marketers intended it or not.