Anne Milgram: Why smart statistics are the key to fighting crime

翻译人员: Yuexin Gao 校对人员: liu angela

In 2007, I became the attorney general

2007年，我担任了新泽西州的

of the state of New Jersey.

司法部长。

Before that, I'd been a criminal prosecutor,

在那之前，我曾是一名刑事检察官，

first in the Manhattan district attorney's office,

先是在曼哈顿地区检查官办公室，

and then at the United States Department of Justice.

后来是在国家司法部。

But when I became the attorney general,

但是在担任司法部长之后，

two things happened that changed the way I see criminal justice.

发生了两件事让我改变了对刑事司法的看法

The first is that I asked what I thought

第一个是我提出我所认为的

were really basic questions.

很基本的问题。

I wanted to understand who we were arresting,

我想要了解我们逮捕的是什么人，

who we were charging,

我们指控的是什么人，

and who we were putting in our nation's jails

还有我们是将什么样的人关进看守所

and prisons.

和监狱。

I also wanted to understand

我也想要了解

if we were making decisions

我们所做的决定是否

in a way that made us safer.

会让民众更加安全。

And I couldn't get this information out.

但我无法获取这类信息

It turned out that most big criminal justice agencies

原来多数大型刑事司法机构

like my own

就像我工作的地方

didn't track the things that matter.

他们并没有对真正重要的事情进行持续的跟踪调查和记录。

So after about a month of being incredibly frustrated,

所以经历了约一个月的异常沮丧之后，

I walked down into a conference room

我走进一个会议室

that was filled with detectives

满屋都是探员

and stacks and stacks of case files,

和成堆成堆的案件档案，

and the detectives were sitting there

探员们坐在那里

with yellow legal pads taking notes.

用黄色便笺簿作着笔记。

They were trying to get the information

他们试图获取的信息

I was looking for

就是我一直在寻找的

by going through case by case

通过逐个分析

for the past five years.

过去的五年间的所有案件。

And as you can imagine,

你可以想象

when we finally got the results, they weren't good.

我们终于得出的结果并不是很理想。

It turned out that we were doing

原来我们一直在做

a lot of low-level drug cases

很多低级的毒品案件

on the streets just around the corner

就在拐角处的街道上

from our office in Trenton.

离我们在特伦顿的办公室不远。

The second thing that happened

第二件事是

is that I spent the day in the Camden, New Jersey police department.

我在卡姆登的新泽西州警察局耗了一天。

Now, at that time, Camden, New Jersey,

当时，新泽西州的卡姆登

was the most dangerous city in America.

是美国最危险的城市。

I ran the Camden Police Department because of that.

我跑了一趟卡姆登警察局就是因为这个原因。

I spent the day in the police department,

我在警察局待了一整天，

and I was taken into a room with senior police officials,

被带到了一个高级警官待的房间，

all of whom were working hard

那里所有人都在努力工作

and trying very hard to reduce crime in Camden.

并很努力的试图减少卡姆登的犯罪活动

And what I saw in that room,

在那个房间里，

as we talked about how to reduce crime,

当我们谈到如何减少犯罪，

were a series of officers with a lot of little yellow sticky notes.

有很多拿着小小的黄色便笺的警官。

And they would take a yellow sticky and they would write something on it

他们会揭下一张黄色便笺，在上面写点东西

and they would put it up on a board.

然后把它贴在板儿上。

And one of them said, "We had a robbery two weeks ago.

其中一个警官说，“我们有一宗劫案发生在两个星期前

We have no suspects."

但没发现疑犯。”

And another said, "We had a shooting in this neighborhood last week. We have no suspects."

另一个说：“上周在这附近发生了一场枪击事件,没发现疑犯”

We weren't using data-driven policing.

我们未曾使用过数据分析来维持治安。

We were essentially trying to fight crime

我们基本上在试图

with yellow Post-it notes.

用黄色便利签打击犯罪。

Now, both of these things made me realize

这两件事让我意识到

fundamentally that we were failing.

从根本上说，我们过去一直表现欠佳。

We didn't even know who was in our criminal justice system,

我们甚至不知道有谁涉及到刑事犯罪，

we didn't have any data about the things that mattered,

我们没有相关重要事件的任何数据，

and we didn't share data or use analytics

我们未曾共享数据，使用分析技术

or tools to help us make better decisions

或分析工具，以帮助我们做出更好地判断

and to reduce crime.

并减少犯罪。

And for the first time, I started to think

我第一次开始思考

about how we made decisions.

我们是如何作出决定的。

When I was an assistant D.A.,

当我还是助理地方检察官，

and when I was a federal prosecutor,

和联邦检察官的时候，

I looked at the cases in front of me,

我看着面前的那些案件，

and I generally made decisions based on my instinct

我所做出的决定通常是依据我的直觉

and my experience.

和我的经验。

When I became attorney general,

后来我成为司法部长，

I could look at the system as a whole,

我可以全面的观察整个司法系统

and what surprised me is that I found

令人吃惊的是我发现

that that was exactly how we were doing it

我们的这种做法恰恰适用于

across the entire system --

整个司法系统 — —

in police departments, in prosecutors's offices,

警察部门，检察官办公室，

in courts and in jails.

法庭和监狱。

And what I learned very quickly

很快，我了解到

is that we weren't doing a good job.

我们过去的工作成果并不令人满意。

So I wanted to do things differently.

所以我想做些改变。

I wanted to introduce data and analytics

我想将数据、逻辑分析

and rigorous statistical analysis

和严格的统计分析

into our work.

纳入到我们的工作。

In short, I wanted to moneyball criminal justice.

总之，我想在刑事司法上做到点球成金。

Now, moneyball, as many of you know,

正如大家所知，点球成金，

is what the Oakland A's did,

是奥克兰运动家棒球队的方法，

where they used smart data and statistics

他们使用了智能数据和统计学

to figure out how to pick players

找出如何挑选球员的办法

that would help them win games,

这将有助于他们赢得比赛，

and they went from a system that was based on baseball scouts

依据一个由棒球侦查员组成的系统

who used to go out and watch players

他们频繁的观察球员的表现

and use their instinct and experience,

利用他们的直觉和经验，

the scouts' instincts and experience,

侦查员的直觉和经验，

to pick players, from one to use

挑选球员，通过使用

smart data and rigorous statistical analysis

智能数据和严格的统计分析

to figure out how to pick players that would help them win games.

找出挑选球员的办法，这会帮助他们赢得比赛。

It worked for the Oakland A's,

奥克兰运动家棒球队就是这样运作的，

and it worked in the state of New Jersey.

这对新泽西州也是适用的。

We took Camden off the top of the list

我们已经将卡姆登从

as the most dangerous city in America.

美国最危险的城市的名单中剔除。

We reduced murders there by 41 percent,

我们减少了41%的谋杀率，

which actually means 37 lives were saved.

实际上相当于拯救了37条生命。

And we reduced all crime in the city by 26 percent.

整个城市的犯罪行为减少了26%。

We also changed the way we did criminal prosecutions.

我们也改变了刑事诉讼的工作方式。

So we went from doing low-level drug crimes

所以我们从调查低级毒品犯罪案件

that were outside our building

它们就发生在办公大楼外面

to doing cases of statewide importance,

转移到调查全州范围内的重要案件，

on things like reducing violence with the most violent offenders,

比如减少暴力罪犯的暴力行为，

prosecuting street gangs,

起诉街头帮派，

gun and drug trafficking, and political corruption.

枪支和毒品的不法交易，还有政治腐败。

And all of this matters greatly,

所有这些事项都很重要，

because public safety to me

因为我认为，保证公共安全

is the most important function of government.

是政府最重要的职能。

If we're not safe, we can't be educated,

如果人身安全无法保证，我们就不能接受教育，

we can't be healthy,

就不能保持身体健康，

we can't do any of the other things we want to do in our lives.

就不能做生活中想要做的任何事。

And we live in a country today

今天我们生活的国家

where we face serious criminal justice problems.

面临着严重的刑事司法问题。

We have 12 million arrests every single year.

我们每年有1200 万起拘捕行动。

The vast majority of those arrests

绝大多数的拘捕行动

are for low-level crimes, like misdemeanors,

是针对低级的犯罪行为，像轻罪，

70 to 80 percent.

这些占据了70%至80%。

Less than five percent of all arrests

不到5%的拘捕行动

are for violent crime.

是针对暴力犯罪。

Yet we spend 75 billion,

然而，我们每年花费750亿美元，

that's b for billion,

以十亿为单位，

dollars a year on state and local corrections costs.

作为国家和地方的修正成本。

Right now, today, we have 2.3 million people

此时此刻，有230万人

in our jails and prisons.

被监禁在看守所和监狱里。

And we face unbelievable public safety challenges

我们在公共安全方面面临着惊人的挑战

because we have a situation

因为现在的形势是

in which two thirds of the people in our jails

监狱中有三分之二的人

are there waiting for trial.

在那里等待审判。

They haven't yet been convicted of a crime.

他们至今还没有被判定有罪。

They're just waiting for their day in court.

他们一直等着出庭受审，

And 67 percent of people come back.

其中有67%的人会重返社会。

Our recidivism rate is amongst the highest in the world.

我们是世界上累犯率最高的国家之一。

Almost seven in 10 people who are released

几乎每释放10个人就有7个

from prison will be rearrested

将会再次被逮捕

in a constant cycle of crime and incarceration.

这是一个恒定的犯罪和监禁的周期。

So when I started my job at the Arnold Foundation,

所以，当我开始在阿诺德基金会工作时，

I came back to looking at a lot of these questions,

回头看了很多这类问题，

and I came back to thinking about how

重新思考了我们怎样

we had used data and analytics to transform

利用数据和分析转变了

the way we did criminal justice in New Jersey.

新泽西州的刑事司法。

And when I look at the criminal justice system

后来我注意到

in the United States today,

当今国家的刑事司法系统，

I feel the exact same way that I did

我觉得应使用同样的方法

about the state of New Jersey when I started there,

即首先在新泽西州使用的那种方法，

which is that we absolutely have to do better,

毫无疑问我们要做得更好，

and I know that we can do better.

而且我知道我们可以做得更好。

So I decided to focus

所以我决定将重点集中在

on using data and analytics

使用数据和分析

to help make the most critical decision

以帮助作出最关键的判断

in public safety,

在公共安全方面，

and that decision is the determination

这一决定是判断

of whether, when someone has been arrested,

已经被逮捕的某个人，

whether they pose a risk to public safety

是否会对公共安全构成风险

and should be detained,

而被拘留

or whether they don't pose a risk to public safety

还是不会对公共安全造成风险

and should be released.

应被释放。

Everything that happens in criminal cases

刑事案件中发生的一切

comes out of this one decision.

都出自这一决定。

It impacts everything.

它影响了全局。

It impacts sentencing.

它影响了判刑。

It impacts whether someone gets drug treatment.

它影响到是否有人需要药物治疗。

It impacts crime and violence.

它影响了犯罪和暴力行为。

And when I talk to judges around the United States,

当我同全美众多法官交谈时，

which I do all the time now,

我现在无时无刻不在做这件事，

they all say the same thing,

他们都说着同样的话，

which is that we put dangerous people in jail,

我们把危险的人关进监狱，

and we let non-dangerous, nonviolent people out.

把没有危险的人、非暴力的人放出去。

They mean it and they believe it.

他们是认真的，他们相信自己所做的。

But when you start to look at the data,

但当你开始查看那些数据，

which, by the way, the judges don't have,

顺便提一句，那些法官没看过，

when we start to look at the data,

当我们开始查看数据，

what we find time and time again,

我们一次又一次的发现，

is that this isn't the case.

这不是个案。

We find low-risk offenders,

我们发现，

which makes up 50 percent of our entire criminal justice population,

占刑事司法总人数的50%的低风险罪犯

we find that they're in jail.

被关在监狱里。

Take Leslie Chew, who was a Texas man

举个例子，莱斯利丘是德克萨斯州人

who stole four blankets on a cold winter night.

在一个寒冷的冬夜偷了四个毯子。

He was arrested, and he was kept in jail

他被拘捕，然后被关进监狱

on 3,500 dollars bail,

需要三千五百美元保释金，

an amount that he could not afford to pay.

这是一笔他支付不起的金额。

And he stayed in jail for eight months

他在监狱里呆了八个月

until his case came up for trial,

直到他的案子开庭，

at a cost to taxpayers of more than 9,000 dollars.

共花了纳税人9,000 多美元税款。

And at the other end of the spectrum,

在另一个极端，

we're doing an equally terrible job.

我们所做的工作也同样糟糕，

The people who we find

那些我们抓获的

are the highest-risk offenders,

高危险罪犯，

the people who we think have the highest likelihood

那些被认为一旦释放会有极高的可能性

of committing a new crime if they're released,

再次犯罪的人，

we see nationally that 50 percent of those people

在全国范围内，其中的50%

are being released.

正在回归社会。

The reason for this is the way we make decisions.

是我们做决定的方式导致的这种结果。

Judges have the best intentions

法官怀着善意

when they make these decisions about risk,

做出这些有风险的决定，

but they're making them subjectively.

但，是主观的决定。

They're like the baseball scouts 20 years ago

他们就像20 年前的棒球侦查员

who were using their instinct and their experience

他们凭本能和经验

to try to decide what risk someone poses.

试着去判断某个人制造的危险。

They're being subjective,

他们是主观的，

and we know what happens with subjective decision making,

我们知道主观决策会导致什么，

which is that we are often wrong.

那就是我们常常犯错。

What we need in this space

我们需要的是

are strong data and analytics.

有力的数据和分析。

What I decided to look for

我决定寻找

was a strong data and analytic risk assessment tool,

一个有力的数据分析的风险评估工具，

something that would let judges actually understand

这个工具会让法官

with a scientific and objective way

以科学的客观的方式去了解

what the risk was that was posed

什么样的风险

by someone in front of them.

摆在他们面前。

I looked all over the country,

我找遍了全国，

and I found that between five and 10 percent

发现5%至10%的

of all U.S. jurisdictions

美国管辖区域

actually use any type of risk assessment tool,

实际使用了某些类型的风险评估工具，

and when I looked at these tools,

我查看了这些评估工具之后，

I quickly realized why.

很快意识到其中缘由。

They were unbelievably expensive to administer,

它们应用起来非常昂贵，

they were time-consuming,

非常耗时，

they were limited to the local jurisdiction

它们被限制在地方管辖区域

in which they'd been created.

因为它们就出自那里。

So basically, they couldn't be scaled

因此，基本上，它们不能扩展

or transferred to other places.

或转移到其他地方。

So I went out and built a phenomenal team

所以我组建了一个出色的团队

of data scientists and researchers

由数据科学家和研究人员

and statisticians

还有统计人员组成

to build a universal risk assessment tool,

建立一个通用的风险评估工具，

so that every single judge in the United States of America

这样一来，全美的每一个法官

can have an objective, scientific measure of risk.

都可以做一个客观、科学的风险评估。

In the tool that we've built,

在这个已经建立的这个工具里，

what we did was we collected 1.5 million cases

我们收集了150 万个案件

from all around the United States,

来自美国各地，

from cities, from counties,

包括城市，县，

from every single state in the country,

国内的每个州，

the federal districts.

联邦区。

And with those 1.5 million cases,

有了那150 万个案例，

which is the largest data set on pretrial

这是美国审判前最大的数据库

in the United States today,

截止到今天，

we were able to basically find that there were

我们基本上能够找到

900-plus risk factors that we could look at

我们可以查看的九百多个危险因素

to try to figure out what mattered most.

试着找出最重要的问题。

And we found that there were nine specific things

我们发现具体有九件事

that mattered all across the country

在全国范围内都很重要

and that were the most highly predictive of risk.

是可预测的最高风险。

And so we built a universal risk assessment tool.

于是我们建立了一种通用的风险评估工具。

And it looks like this.

它看起来就像这个。

As you'll see, we put some information in,

正如你所看到的，我们把一些信息列在上面，

but most of it is incredibly simple,

但大多数都格外简单，

it's easy to use,

它使用起来很容易，

it focuses on things like the defendant's prior convictions,

它侧重的方面是被告的前科，

whether they've been sentenced to incarceration,

他们是否曾被判处监禁，

whether they've engaged in violence before,

他们是否曾被卷入过暴力事件，

whether they've even failed to come back to court.

他们是否甚至没能回到法庭。

And with this tool, we can predict three things.

使用此工具，我们可以预测三件事。

First, whether or not someone will commit

第一，他们是否会再次犯罪

a new crime if they're released.

如果被释放的话。

Second, for the first time,

第二，我第一次觉得，

and I think this is incredibly important,

这一点非常重要，

we can predict whether someone will commit

我们可以预测他们是否会

an act of violence if they're released.

进行暴力活动，如果被释放的话。

And that's the single most important thing

法官说这是最重要的一件事

that judges say when you talk to them.

当你向他们问话的时候。

And third, we can predict whether someone

第三，我们可以预测是否他们

will come back to court.

会回到法庭。

And every single judge in the United States of America can use it,

美国的任何一名法官都可以使用它，

because it's been created on a universal data set.

因为它是由通用的数据库制成的。

What judges see if they run the risk assessment tool

如果他们使用了这个风险评估工具你就会看到

is this -- it's a dashboard.

这个--一个评分板。

At the top, you see the New Criminal Activity Score,

在顶部的，是新的刑事犯罪活动评分，

six of course being the highest,

六当然是最高分，

and then in the middle you see, "Elevated risk of violence."

然后在中间你可以看到“增长的暴力风险”

What that says is that this person

说的是，这个人

is someone who has an elevated risk of violence

进行暴力行为的概率有所升高

that the judge should look twice at.

这是法官应该注意的重点。

And then, towards the bottom,

然后在底部，

you see the Failure to Appear Score,

你看到的是未能出庭的分数，

which again is the likelihood

这也是判断

that someone will come back to court.

此人会回到法庭的可能性。

Now I want to say something really important.

现在我想说些非常重要的事。

It's not that I think we should be eliminating

我不认为应该排除

the judge's instinct and experience

法官的直觉和经验

from this process.

在整个过程中。

I don't.

我不这样想。

I actually believe the problem that we see

事实上，我相信我们看到的问题

and the reason that we have these incredible system errors,

以及系统内出现令人难以置信的错误的原因，

where we're incarcerating low-level, nonviolent people

也就是我们关押低级、非暴力的人

and we're releasing high-risk, dangerous people,

我们释放高风险的、危险的人的原因，

is that we don't have an objective measure of risk.

是因为我们没有客观的衡量风险。

But what I believe should happen

但我相信

is that we should take that data-driven risk assessment

我们应该将这种数据驱动的风险评估

and combine that with the judge's instinct and experience

与法官的直觉和经验相结合

to lead us to better decision making.

会使我们做出更好的决策。

The tool went statewide in Kentucky on July 1,

该评估工具于7月1日在肯塔基州全面推行，

and we're about to go up in a number of other U.S. jurisdictions.

我们还要在许多其他美国司法管辖区内推行。

Our goal, quite simply, is that every single judge

我们的目标很简单，就是让全美的每一个法官

in the United States will use a data-driven risk tool

都能使用这种数据驱动的风险评估工具

within the next five years.

在未来五年内实现。

We're now working on risk tools

我们现在正在研究风险工具

for prosecutors and for police officers as well,

以便检察官和警官使用，

to try to take a system that runs today

想要把过去50年不变的系统

in America the same way it did 50 years ago,

继续运行下去，

based on instinct and experience,

基于直觉和经验，

and make it into one that runs

并把它变成一种运用

on data and analytics.

数据分析的系统。

Now, the great news about all this,

现在，

and we have a ton of work left to do,

我们仍有大量的工作要做，

and we have a lot of culture to change,

仍要改变相关文化，

but the great news about all of it

但关于这所有一切有个好消息

is that we know it works.

那就是我们知道这很管用。

It's why Google is Google,

这就是为什么数据成就了谷歌，

and it's why all these baseball teams use moneyball

为什么棒球队会因使用点球成金的方法

to win games.

而打赢比赛。

The great news for us as well

还有一个好消息是

is that it's the way that we can transform

我们可以用这种方法转变

the American criminal justice system.

美国的刑事司法系统。

It's how we can make our streets safer,

它可以使我们的周边环境更安全，

we can reduce our prison costs,

可以减少我们的监狱花销，

and we can make our system much fairer

可以让我们的司法系统更公平

and more just.

更公正。

Some people call it data science.

有人说这是数据科学。

I call it moneyballing criminal justice.

我称之为点球成金式刑事司法。

Thank you.

谢谢。

(Applause)

（掌声）