Cyrus Vance Jr.’s ‘Moneyball’ Approach to Crime | Hacker News Reader

Cyrus Vance Jr.’s ‘Moneyball’ Approach to Crime

nytimes.com

41 pointskanamekun24 comments

One of the biggest dangers in Manhattan these days is running over and killing people with cars. In 2013 333 people were murdered in New York, while 286 people were killed by car crashes. Vance steadfastly refuses to prosecute motor vehicle violent crime. As long as this blind spot persists he cannot claim to be taking a scientific or quantitative approach to reducing crime.

http://www.streetsblog.org/2014/11/26/vance-serves-up-excuse...

The link you sent says Vance does prosecute 'motor crime' (DUIs, fleeing the scene,etc..) Also you forgot to add that it is inherently difficult to prosecute because a jury doesn't see negligent drivers as criminals. You already know about the clement case that Vance lost.

I intend to agree that car crashes is one of society's biggest and underrated problems, but your really laying out a big strawman here

> a jury doesn't see negligent drivers as criminals

That's a little harsh. They probably do see them as criminals.

However, they can also see themselves being in the defendant's chair for this "crime". Nobody is perfect, and everybody has a story of how they screwed up while driving.

Let he who is without sin cast the first stone.

If you define killing people with cars to be out of scope then you're just drawing the system boundaries incorrectly. Let me ask you this: if it's not a crime to kill someone with a car, as long as you're sober and you don't flee the scene, why does anyone murder someone else in New York with a knife or a gun? Wouldn't running them down with a car be the rational choice? After all, they aren't prosecuting the guy who ran over and killed a 3-year-old in a crosswalk, even though there's a video recording of the act. You can be captured on video running down and killing someone and not be prosecuted. Only a fool would choose another weapon.

Because crime in the US is a matter of intent. If you intend to kill someone, and then do, the exact means of killing are largely irrelevant; you have committed some sort of homicide crime. (The exact means of killing may determine what particular homicide crime has been committed, such as manslaughter vs murder.) That is why people above have been saying that most accidents are not criminal--the driver does not intend to hit the cyclist/pedestrian/whatever. Negligence is generally not a crime in the US except in exceptional circumstances (i.e., negligent homicide, in which the driver's gross negligence while driving--such as texting while the car is moving--results in accident in which someone else dies).

The vast majority of those 286 crash deaths aren't criminal though..

You concluded that how?

Because the vast majority of vehicle accident aren't criminal? Why would this be any different. If you've ever driven in Manhattan you'd know how easy it is for a mistake to happen, and it would be a mess to prove who's at fault since traffic laws and J walking are hardly enforced.

What is that even supposed to mean? Just because people don't intend to run over others doesn't mean it's not criminal when they do so through negligence. The phrase "if you've ever driven in Manhattan" seems to suggest you're well aware that plenty of drivers don't exactly reach to the level of competence we would expect from someone manuevering tons of steel in a dense city where the number of pedestrians frequently exceeds that of cars.

This line of thinking is exactly why we now have 300 deaths from traffic, or as many as murder cases. People think negligence in driving, be it speeding, fiddling with your phone, accelerating for that yellow and dodging instead of yielding to pedestrians in crosswalks are somehow excusable. They're not when you are driving a car.

Although I agree that this is certainly true, I cannot vouch for its historic truth. It is possible that the vast reduction in the murder rate is primarily due to the fact that vehicle accidents are (no longer) considered murder, rather than any actual reduction in violent crime.

Not stating this as fact -- I haven't even done my homework to look at vehicle-related deaths over time -- but as motivation for the possible blind spot pointed out by thrownaway2424. Historical crime data is notorious for this because statistics are gathered by law enforcement and have a subjective aspect based on the time that they are gathered.

Motor vehicle deaths, especially when corrected for miles driven or total population, have dropped precipitously:

http://en.wikipedia.org/wiki/List_of_motor_vehicle_deaths_in...

Murder, aggravated assault, rape, robbery, etc. are all down across the board, in some cases by more than 50% in 20 years:

http://www.fbi.gov/about-us/cjis/ucr/crime-in-the-u.s/2012/c...

There are always some subjective decisions on data reporting, but that's much more prevalent in data comparisons between countries rather than over time within a single country. The FBI has been very consistent in their reporting.

Civilized countries have long since settled this question with a simple formula: if you are in a car and you hurt or kill someone, it's your fault.

Really?

So the child who ran out chasing after a ball is your fault? The bicyclist wearing black, at night, in the rain, with no reflectors or lights is your fault?

We are human. We have moments of inattention ALL THE TIME. This is proven by many, many, many psychological studies. You don't convict someone of a crime because they committed an accidental error.

If you want to cut down on this, fund self-driving cars.

Yes, really. This is the law in Belgium for example.

I would love to see a source for the claim that accidents are criminally prosecuted in Belgium.. The data I've found looks like ~5% of trips in Belgium are by drunk drivers, ~35% of Belgians drive over the speed limit, drivers only wear seat belts ~80% of the time and they aren't liable for their passengers' seat belt use..

It sounds like there should be thousands of Belgians in jail for traffic crimes, but extrapolating from their population / imprisonment rate (11.2M & 108/100k), it looks like they only have 12k total people in jail. Their annual violent crime rate is ~50/100k and if you assume even an average of a 2-year sentence for violent crimes, there is no room left in jail for these traffic criminals...

[http://archive.etsc.eu/documents/Final_Traffic_Law_Enforceme...]

My armchair opinion is that most of the people that are driving in the city are towards the wealthy end of the spectrum (or tourists that don't know any better), which makes going after them political suicide.

[Note: I'm ignoring taxis / buses here. I imagine that killing someone with your taxi affects you significantly enough for you to try and avoid it, even without jail time or criminal charges.]

The wealthy don't personally drive in Manhattan. People pay good money not to have to drive in Manhattan.

There's a quote in the article from an assistant DA claiming it is a 'moneyball' approach but seems more like a hash of somewhat data-oriented approaches.

It's been a long time since I've read the Moneyball book, and never saw the movie, but it seems to me there are two levels of 'moneyball' approach:

1. Use data to make allocate resources differently than gut instinct/received wisdom (of baseball scouts). This level can be applied very widely. Maybe they (Manhattan DA office) are doing this level to decide who to put resources into prosecuting. But clearly they're not doing it comprehensively, given massive prosecutions for marijuana.

2. Do (1) to make a bet in a competitive environment that goes a different way than most competitors. I don't see this level applying to what the DA is doing at all, though I imagine it could be done in the hiring of law enforcement/prosecution, which seems to be competitive, at least at the executive level. But there's probably a lack of data to make such bets. (I haven't looked, but an idle question I have is whether there's data to support shifting police work to a female-dominated profession; if there is maybe that's a bet that could be made.)

I say this in part because I think there are non-sports fields in which both levels could be very directly translated, eg http://gondwanaland.com/mlog/2012/03/04/altmetrics-moneyball...

I am starting to detest articles that do not provide any comparisons to the numbers they throw out.

In 1974 and 2009 there were 648 and 59 murders in Manhattan respectively.

While in US total in 1974 and 2009 there were 20710[9366] and 15399[15350] murders respectively. Number in square bracket is alternative number from data in second table on cited page below.

Thus it looks that there is significant reduction of murders in Manhattan specifically. Perhaps most striking statistics is that in 1974 murders in Manhattan constituted over 3% of total murders in the country while having only 0.7% of population. In contrast in 2009 murders in Manhattan constituted 0.4% of nation total while having 0.5% of population. I am using data only form first table here.

I am sure that there are a host of other factors that are affecting this, notably rise of property prices which have pushed out substantial portion of lower income population out of Manhattan.

I used data from: http://www.disastercenter.com/crime/uscrime.htm and https://www.google.com/search?q=manhattan+population

The government has designated ~9,000 people as the most dangerous criminals, "virtually all of whom have criminal records", and uses that to guide how these people are treated. Based on what I read, you are put on this list without trial, without your knowledge, and without appeal.

What could go wrong? The Times devotes a little space to this question, far down in the article, but doesn't seem to see the obvious, major flaws. How many times will we repeat these mistakes before there is enough of a widespread, general understanding of them?

Part of the problem is that it's not affecting the people with power in our society. If there's a mistake, we probably won't know about it -- from our perspective, nothing is perfect, but the victim could end up in jail and have their life ruined. If this was a list of government-designated criminal hackers, and it affected your ability to use IT and your job ...

EDIT: Many edits

I suspect that this list is really just a database sort over a weighted set of criteria. That sort gets you the "most problematic" criminals by some criteria.

After than we have the following:

* the person is in jail, in which case being on the list is largely irrelevant

* the person is out of jail, but not committing crimes, in which case being on the list is largely irrelevant. They will presumably fall off of the list as they fail to commit crimes

* the person is out of jail, but has just been apprehended for committing a crime.

For the last situation, the list is hugely important, because it helps you deal with the crime in context.

Do you have other sources of information besides the article? I don't remember much in it that supports those claims.

"Inspired by “broken windows” policing strategies advocated by William J. Bratton, who at the time was head of the city’s transit cops and is now the New York City police commissioner, D.A.s began to prosecute offenses once considered small potatoes. The iconic example in New York was the campaign against the “squeegee men” cleaning car windshields when drivers were trapped in traffic."

What a timely article, on the same day that a grand jury returns a "no bill" for a cop who choked a man to death for the crime of selling loose cigarettes.

Killing a few black and brown people (and ruining the lives of thousands more by sending them to prison) for petty offenses is a small price to pay, as long as Times Square is made safe for "Elmo impersonators and wide-eyed naïfs in fanny packs".

The title made me think of Analytics Made Skeezy[1]. I am a little sad that there's so little data, as others have already pointed out.

[1] http://analyticsmadeskeezy.com/

Reply on news.ycombinator.com