Slashdot Mirror
'Moneyball' Approach Reduces Crime In New York City
HughPickens.com writes The NYT reports that NY County District Attorney Cyrus Vance Jr.'s most significant initiative has been to transform, through the use of data, the way district attorneys fight crime. "The question I had when I came in was, Do we sit on our hands waiting for crime to tick up, or can we do something to drive crime lower?" says Vance. "I wanted to develop what I call intelligence-driven prosecution." When Vance became DA in 2009, it was glaringly evident that assistant D.A.s fielding the 105,000-plus cases a year in Manhattan seldom had enough information to make nuanced decisions about bail, charges, pleas or sentences. They were narrowly focused on the facts of cases in front of them, not on the people committing the crimes. They couldn't quickly sort minor delinquents from irredeemably bad apples. They didn't know what havoc defendants might be wreaking in other boroughs.
Vance divided Manhattan's 22 police precincts into five areas and assigned a senior assistant D.A. and an analyst to map the crime in each area. CSU staff members met with patrol officers, detectives and Police Department field intelligence officers and asked police commanders to submit a list of each precinct's 25 worst offenders — so-called crime drivers, whose "incapacitation by the criminal-justice system would have a positive impact on the community's safety." Seeded with these initial cases, the CSU built a searchable database that now includes more than 9,000 chronic offenders (PDF), virtually all of whom have criminal records. A large percentage are recidivists who have been repeatedly convicted of grand larceny, one of the top index crimes in Manhattan, but the list also includes active gang members, people whom the D.A. considers "uncooperative witnesses," and a fluctuating number of violent "priority targets," which currently stands at 81. "These are people we want to know about if they are arrested," says Kerry Chicon. "We are constantly adding, deleting, editing and updating the intelligence in the Arrest Alert System. If someone gets out of a gang, or goes to prison for a long time, or moves out of the city or the state, or ages out of being a focus for us, or dies, we edit the system accordingly — we do that all the time."
"It's the 'Moneyball' approach to crime," says Chauncey Parker. "The tool is data; the benefit, public safety and justice — whom are we going to put in jail? If you have 10 guys dealing drugs, which one do you focus on? The assistant district attorneys know the rap sheets, they have the police statements like before, but now they know if you lift the left sleeve you'll find a gang tattoo and if you look you'll see a scar where the defendant was once shot in the ankle. Some of the defendants are often surprised we know so much about them."
"It's the 'Moneyball' approach to crime," says Chauncey Parker. "The tool is data; the benefit, public safety and justice — whom are we going to put in jail? If you have 10 guys dealing drugs, which one do you focus on? The assistant district attorneys know the rap sheets, they have the police statements like before, but now they know if you lift the left sleeve you'll find a gang tattoo and if you look you'll see a scar where the defendant was once shot in the ankle. Some of the defendants are often surprised we know so much about them."
I would very much like to know the racial makeup of that list. Given it came from the police themselves, it certainly leads to questions about how such individuals end up on those lists.
Put the fucking race card away.
I think you're missing the point, because if data is really being used how it should be, in the most efficient way, this goes way past "profiling", which is essentially the opposite approach in terms of detail, and heads into "accuracy".
Well the second paragraph of the summary makes it pretty clear it isn't just a database of "people who look like they could be criminals". They are repeat offenders of serious crimes. I don't really even get what you mean by "biased slice of the population". Yeah it's biased, because they have to include bad guys in the list. Otherwise what do you mean? Data isn't racist, which was my original point. I'm assuming unless they are the most bigotted people on the planet and somehow programmed that into their algorithm, their lists include a pretty fair percentage of each race, according to their relative rates of committing the crimes they are singling out as important.
Well the second paragraph of the summary makes it pretty clear it isn't just a database of "people who look like they could be criminals". They are repeat offenders of serious crimes. I don't really even get what you mean by "biased slice of the population". Yeah it's biased, because they have to include bad guys in the list. Otherwise what do you mean? Data isn't racist, which was my original point. I'm assuming unless they are the most bigotted people on the planet and somehow programmed that into their algorithm, their lists include a pretty fair percentage of each race, according to their relative rates of committing the crimes they are singling out as important.
His point is that the police may be racially profiling to begin with. If they are more suspicious of black people, more likely to arrest a black person to begin with, then the data base is going to be artificially skewed towards information about black people. There may be plenty of white people that are doing the exact same thing without ever being caught because they aren't getting stop and frisked and found to be in possession of drugs, for instance.