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

譯者: Marssi Draw 審譯者: William Choi

In 2007, I became the attorney general

2007年時，我成為紐澤西州

of the state of New Jersey.

總檢察長。

Before that, I'd been a criminal prosecutor,

在那之前，我擔任刑事檢察官，

first in the Manhattan district attorney's office,

一開始在曼哈頓區檢察官辦公室，

and then at the United States Department of Justice.

後來在美國司法部。

But when I became the attorney general,

但是當我成為總檢察長後，

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

發生了兩件事， 改變我對刑事司法的看法。

The first is that I asked what I thought

第一件事是我提出了一些我認為

were really basic questions.

很基本的問題。

I wanted to understand who we were arresting,

我想了解我們逮捕誰、

who we were charging,

控告誰、

and who we were putting in our nation's jails

以及把誰關進我國的拘留所

and prisons.

和監獄。

I also wanted to understand

我也想了解

if we were making decisions

我們做決定的方式

in a way that made us safer.

是否能讓我們更安全。

And I couldn't get this information out.

我找不出答案。

It turned out that most big criminal justice agencies

結果是多數的大型刑事司法機關，

like my own

就像我工作的地方，

didn't track the things that matter.

沒有追蹤重要的事情。

So after about a month of being incredibly frustrated,

因此歷經大約一個月的強烈挫折感後，

I walked down into a conference room

我走進一間會議室，

that was filled with detectives

裡面擠滿警探

and stacks and stacks of case files,

和堆積如山的案件資料，

and the detectives were sitting there

警探坐在那，

with yellow legal pads taking notes.

手拿黃色標準便條紙在做筆記。

They were trying to get the information

他們試圖得到我在找的資訊，

I was looking for

by going through case by case

逐一檢視

for the past five years.

過去五年的案件。

And as you can imagine,

如你所料，

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

最後得到的結果不太好。

It turned out that we were doing

結論是我們辦了

a lot of low-level drug cases

很多低階的街頭毒品案件，

on the streets just around the corner

就在附近的街上，

from our office in Trenton.

離我們在翠登的辦公室不遠。

The second thing that happened

第二件發生的事情是

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

我在紐澤西肯頓警局待的那一天。

Now, at that time, Camden, New Jersey,

那時候紐澤西州肯頓

was the most dangerous city in America.

是全美最危險的城市，

I ran the Camden Police Department because of that.

這是我去肯頓警局的原因。

I spent the day in the police department,

我在警局待一整天，

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

被帶到一間房間， 和資深警官在一起，

all of whom were working hard

他們都很努力

and trying very hard to reduce crime in Camden.

試著降低肯頓的犯罪率。

And what I saw in that room,

當我們討論如何降低犯罪率時，

as we talked about how to reduce crime,

我在那房裡看到

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

一大堆警官拿著 很多黃色的小型便利貼。

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

他們拿著一張黃色便利貼， 在上面寫點東西，

and they would put it up on a board.

把它貼在公布欄上。

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

其中一位會說：「兩個星期前有搶案。

We have no suspects."

沒有嫌犯。」

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

另一位說：「上星期這附近 有槍擊案。沒有嫌犯。」

We weren't using data-driven policing.

我們沒有運用任何數據處理治安。

We were essentially trying to fight crime

基本上，我們打算用黃色便利貼

with yellow Post-it notes.

來打擊犯罪。

Now, both of these things made me realize

這兩件事讓我了解

fundamentally that we were failing.

我們徹底失敗了。

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

我們甚至不知道誰在 我們的刑事司法體系裡，

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

我們沒有重要資料的數據，

and we didn't share data or use analytics

也沒有共享數據、運用分析

or tools to help us make better decisions

或工具來幫助我們做更好的決定，

and to reduce crime.

並減少犯罪。

And for the first time, I started to think

我第一次思考

about how we made decisions.

我們是如何做決定。

When I was an assistant D.A.,

我擔任地區助理檢察官

and when I was a federal prosecutor,

和聯邦檢察官，

I looked at the cases in front of me,

研究眼前的案件時，

and I generally made decisions based on my instinct

幾乎都是靠直覺

and my experience.

和經驗做決定。

When I became attorney general,

當我成為總檢察長，

I could look at the system as a whole,

能夠全面檢視體制，

and what surprised me is that I found

讓我驚訝的是發現了

that that was exactly how we were doing it

我們就是那樣做，

across the entire system --

整個體制都是如此──

in police departments, in prosecutors's offices,

在警察局、檢察署、

in courts and in jails.

法院和監獄。

And what I learned very quickly

很快我就了解

is that we weren't doing a good job.

我們做得不好，

So I wanted to do things differently.

於是就想用不同的方式做事。

I wanted to introduce data and analytics

我想把數據、邏輯分析

and rigorous statistical analysis

和精密統計分析

into our work.

運用到工作上。

In short, I wanted to moneyball criminal justice.

簡而言之，我想用 魔球的方式處理刑事司法。

Now, moneyball, as many of you know,

如在座許多人所知，

is what the Oakland A's did,

魔球是奧克蘭運動家隊 所運用的策略，

where they used smart data and statistics

他們用數據和統計

to figure out how to pick players

找出選擇球員的方法

that would help them win games,

去幫助球隊贏球，

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

他們從前根據棒球球探意見，

who used to go out and watch players

球探會出門去看球員，

and use their instinct and experience,

然後以直覺和經驗，

the scouts' instincts and experience,

球探的直覺和經驗，

to pick players, from one to use

去挑選球員，從運用

smart data and rigorous statistical analysis

數據和精密統計分析

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

找出要怎麼選出 能讓他們贏得比賽的球員。

It worked for the Oakland A's,

對奧克蘭運動家隊奏效了，

and it worked in the state of New Jersey.

對紐澤西州也奏效了。

We took Camden off the top of the list

我們讓肯頓不再名列

as the most dangerous city in America.

美國最危險城市之一。

We reduced murders there by 41 percent,

我們把當地兇殺案減少了 41%，

which actually means 37 lives were saved.

意謂著救了 37 條人命。

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

我們將城裡各種犯罪活動減少了 26% 。

We also changed the way we did criminal prosecutions.

我們也改變刑事訴訟的方式，

So we went from doing low-level drug crimes

從處理低階的毒品犯罪，

that were outside our building

那些發生在我們的大樓外，

to doing cases of statewide importance,

轉變為遍及全州的重要案件，

on things like reducing violence with the most violent offenders,

像是減少高度危險暴力犯的再犯率、

prosecuting street gangs,

起訴街頭幫派、

gun and drug trafficking, and political corruption.

槍枝和毒品運送，以及政治貪汙。

And all of this matters greatly,

這一切帶來的影響甚大，

because public safety to me

因為公共安全對我來說

is the most important function of government.

是政府最重要的功能。

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

如果我們不安全， 我們就無法接受教育，

we can't be healthy,

就無法擁有健康，

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

我們就無法做所有生活中想做的事。

And we live in a country today

今天我們居住的國家

where we face serious criminal justice problems.

正面對嚴重的刑事司法問題。

We have 12 million arrests every single year.

我們每年有 1,200 萬件逮捕案。

The vast majority of those arrests

這些逮捕案最大部分的是

are for low-level crimes, like misdemeanors,

低階犯罪，像是輕罪，

70 to 80 percent.

佔 70% 到 80%。

Less than five percent of all arrests

只有不到 5% 的逮捕

are for violent crime.

是暴力犯罪。

Yet we spend 75 billion,

然而我們花費 750 億美元，

that's b for billion,

是「百億」元，

dollars a year on state and local corrections costs.

在州和地方一年的懲治支出上。

Right now, today, we have 2.3 million people

現在，我們有 230 萬人

in our jails and prisons.

在監獄和拘留所。

And we face unbelievable public safety challenges

我們面對難以致信的公安問題

because we have a situation

因為面對的處境是

in which two thirds of the people in our jails

拘留所內三分之二的嫌疑犯

are there waiting for trial.

正在等著審判，

They haven't yet been convicted of a crime.

他們還沒被判有罪，

They're just waiting for their day in court.

等著上法庭的那一天。

And 67 percent of people come back.

67% 的嫌疑犯會回來。

Our recidivism rate is amongst the highest in the world.

我們是全球累犯率最高的國家之一。

Almost seven in 10 people who are released

在監被釋放的人之中， 幾乎 10 個就有 7 個

from prison will be rearrested

會再次被逮捕，

in a constant cycle of crime and incarceration.

呈現不斷犯罪和監禁的循環。

So when I started my job at the Arnold Foundation,

因此當我開始在阿諾德基金會工作，

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

回頭來看這一大堆問題，

and I came back to thinking about how

回頭來思考

we had used data and analytics to transform

該如何使用數據和邏輯分析來轉變

the way we did criminal justice in New Jersey.

我們在紐澤西刑事司法採取的方式。

And when I look at the criminal justice system

當我檢視現今

in the United States today,

美國的刑事司法體制時，

I feel the exact same way that I did

我發現和當年在紐澤西起頭時

about the state of New Jersey when I started there,

相同的情況，

which is that we absolutely have to do better,

毫無疑問我們在那做得更好了，

and I know that we can do better.

我也知道我們可以做得更好。

So I decided to focus

因此我決定著眼在

on using data and analytics

使用數據和邏輯分析，

to help make the most critical decision

協助我們在公共安全中

in public safety,

做最重要的決定，

and that decision is the determination

而那個決定即是

of whether, when someone has been arrested,

在某疑犯被逮捕時的判定，

whether they pose a risk to public safety

不管是他們危及公共安全

and should be detained,

該被拘留，

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

又或是他們沒有危及公共安全

and should be released.

而該被釋放。

Everything that happens in criminal cases

每件在刑事案件中發生的事

comes out of this one decision.

都來自於這個決定。

It impacts everything.

這個決定影響每一件事，

It impacts sentencing.

影響每一個判決，

It impacts whether someone gets drug treatment.

影響某疑犯是否接受藥物治療，

It impacts crime and violence.

影響暴力和犯罪。

And when I talk to judges around the United States,

當我和全美法官談話時，

which I do all the time now,

── 我現在常這麼做 ──

they all say the same thing,

他們都說一樣的話，

which is that we put dangerous people in jail,

那就是我們把危險人物關進牢裡，

and we let non-dangerous, nonviolent people out.

讓不危險、非暴力的人出來。

They mean it and they believe it.

他們很認真，也深信不疑。

But when you start to look at the data,

但當你開始檢視數據，

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

附帶一提的是， 法官沒有看過數據，

when we start to look at the data,

當我們開始檢視數據，

what we find time and time again,

就會一次又一次地發現

is that this isn't the case.

根本不是如此。

We find low-risk offenders,

我們見到低風險犯人

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

佔了所有刑事司法總人數的一半，

we find that they're in jail.

我們發現他們在坐牢。

Take Leslie Chew, who was a Texas man

看看一個名為萊斯理的德州人，

who stole four blankets on a cold winter night.

他在一個寒冷冬夜偷了四件毛毯。

He was arrested, and he was kept in jail

他被逮捕，關在牢裡，

on 3,500 dollars bail,

保釋金為 3,500 美元，

an amount that he could not afford to pay.

他繳不出保釋金，

And he stayed in jail for eight months

因此留在牢裡八個月，

until his case came up for trial,

直到案子開審，

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

花費納稅人超過 9,000 美元。

And at the other end of the spectrum,

而在相反的那一端，

we're doing an equally terrible job.

我們做得一樣很糟。

The people who we find

我們見到的是

are the highest-risk offenders,

高風險的犯人，

the people who we think have the highest likelihood

我們認為這些人若被釋放，

of committing a new crime if they're released,

將會極有可能再次犯罪，

we see nationally that 50 percent of those people

這些人全國大概有一半

are being released.

都被釋放了，

The reason for this is the way we make decisions.

原因出自於我們做決定的方式。

Judges have the best intentions

當法官要做出關於風險的這些決定時，

when they make these decisions about risk,

他們是出於好意，

but they're making them subjectively.

但是卻主觀地做出決定，

They're like the baseball scouts 20 years ago

就像 20 年前的棒球球探，

who were using their instinct and their experience

用直覺和經驗

to try to decide what risk someone poses.

嘗試裁定某人會造成什麼危險。

They're being subjective,

他們主觀意識強，

and we know what happens with subjective decision making,

而我們知道主觀的決定 會帶來什麼結果，

which is that we are often wrong.

那就是我們經常會做錯。

What we need in this space

我們在這裡需要的是

are strong data and analytics.

強而有力的數據和邏輯分析。

What I decided to look for

我決定找出

was a strong data and analytic risk assessment tool,

強而有力的數據 和邏輯分析風險評估工具，

something that would let judges actually understand

透過科學和客觀的方式

with a scientific and objective way

讓法官確實了解

what the risk was that was posed

他們面前的人

by someone in front of them.

可能造成什麼風險。

I looked all over the country,

我檢視整個國家，

and I found that between five and 10 percent

發現美國的所有司法轄區之中

of all U.S. jurisdictions

約有 5% 到 10%

actually use any type of risk assessment tool,

確實使用某種型式的風險評估工具，

and when I looked at these tools,

當我研究這些工具，

I quickly realized why.

很快就理解事出何因。

They were unbelievably expensive to administer,

這些工具管理起來貴得嚇人，

they were time-consuming,

耗費時間，

they were limited to the local jurisdiction

而且只能限制在

in which they'd been created.

他們所在的司法轄區。

So basically, they couldn't be scaled

基本上，他們無法擴大規模

or transferred to other places.

或是移轉到其它地方。

So I went out and built a phenomenal team

因此我建立了一個出色的團隊，

of data scientists and researchers

由數據科學家、研究人員

and statisticians

和統計學家組成，

to build a universal risk assessment tool,

來建立全面的風險評估工具，

so that every single judge in the United States of America

如此一來每一位美國法官

can have an objective, scientific measure of risk.

都可以擁有客觀、科學的風險測量。

In the tool that we've built,

在我們設計的工具中，

what we did was we collected 1.5 million cases

我們搜集了

from all around the United States,

全美 150 萬個案件，

from cities, from counties,

它們來自城市、郡市、

from every single state in the country,

來自全國每一個州、

the federal districts.

來自聯邦地區。

And with those 1.5 million cases,

而這 150 萬個案件

which is the largest data set on pretrial

是美國現今審判前

in the United States today,

最大的資料組，

we were able to basically find that there were

基本上我們能找出

900-plus risk factors that we could look at

900 個以上的危險因子， 我們可以從其中檢視，

to try to figure out what mattered most.

嘗試找出什麼是最重要的。

And we found that there were nine specific things

我們發現有特定的九件事

that mattered all across the country

在全國各地都很重要，

and that were the most highly predictive of risk.

而那些是最容易看得出來的風險。

And so we built a universal risk assessment tool.

我們建置出一套全面的風險評估工具，

And it looks like this.

看起來就像這樣，

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

就像你看到的，我們會放入一些資訊，

but most of it is incredibly simple,

但大部分都是很簡單的東西，

it's easy to use,

操作也簡單，

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

像是著眼在被告之前的犯罪記錄，

whether they've been sentenced to incarceration,

不管是否被判監禁，

whether they've engaged in violence before,

不管是否曾涉入暴力案件，

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

或只是未曾出庭。

And with this tool, we can predict three things.

有了這個工具，我們可以預測三件事。

First, whether or not someone will commit

首先，如果某疑犯被釋放的話，

a new crime if they're released.

他會否再犯罪。

Second, for the first time,

第二，這是第一次

and I think this is incredibly important,

── 我想這十分重要 ──

we can predict whether someone will commit

我們可以預測某疑犯如果被釋放，

an act of violence if they're released.

會不會從事暴力犯罪。

And that's the single most important thing

那是法官跟你說話時，

that judges say when you talk to them.

對他而言最重要的事。

And third, we can predict whether someone

第三，我們可以預測

will come back to court.

某疑犯會否回到法庭上。

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

每一位美國法官都能使用這個工具，

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

因為它是以全面性的資料組建立。

What judges see if they run the risk assessment tool

當法官們操作這個風險評估工具時，

is this -- it's a dashboard.

看到的就是這個介面。

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

在最上面的是新犯罪活動評分，

six of course being the highest,

六分當然是最高分，

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

接著在中間可以看到 「增加的暴力風險」。

What that says is that this person

意謂著這個疑犯

is someone who has an elevated risk of violence

有較高機率的暴力風險，

that the judge should look twice at.

法官應該再多看一眼。

And then, towards the bottom,

接著，往底部看，

you see the Failure to Appear Score,

你會看到「未能出庭指數」，

which again is the likelihood

這再次意謂著

that someone will come back to court.

某疑犯會回到法院的可能性。

Now I want to say something really important.

接下來我要說的十分重要。

It's not that I think we should be eliminating

我認為我們並不是應該排除

the judge's instinct and experience

法官在這個過程中的

from this process.

直覺和經驗。

I don't.

不是這個意思。

I actually believe the problem that we see

我確實相信我們看到的問題

and the reason that we have these incredible system errors,

以及造成這些體制裡 重大錯誤的原因，

where we're incarcerating low-level, nonviolent people

我們監禁低階、非暴力的人，

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

卻把高風險的危險人物放出來，

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

是因為我們沒有客觀的風險評估。

But what I believe should happen

但我相信我們應該

is that we should take that data-driven risk assessment

拿這份依照數據產生的風險評估，

and combine that with the judge's instinct and experience

結合法官的直覺和經驗，

to lead us to better decision making.

讓我們做出更好的決定。

The tool went statewide in Kentucky on July 1,

這項工具七月一日開始 在肯塔基州全州使用，

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

我們還要擴展到全美許多轄區。

Our goal, quite simply, is that every single judge

我們的目標很簡單， 就是讓每一個美國法官

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

在五年內都可運用這套

within the next five years.

以數據為導向的風險工具。

We're now working on risk tools

我們現在設計

for prosecutors and for police officers as well,

檢察官和警官也能使用這個風險工具，

to try to take a system that runs today

試著讓這套系統在現今美國運作，

in America the same way it did 50 years ago,

就像 50 年前的方式一樣，

based on instinct and experience,

根據直覺和經驗，

and make it into one that runs

讓它改變為根據

on data and analytics.

數據和邏輯分析。

Now, the great news about all this,

現在這一切最棒的是

and we have a ton of work left to do,

我們有一大堆工作等著我們去做，

and we have a lot of culture to change,

有很多文化要改變，

but the great news about all of it

但這一切最棒的是

is that we know it works.

我們知道那有用。

It's why Google is Google,

這是 Google 之所以 是 Google 的原因，

and it's why all these baseball teams use moneyball

這就是為什麼所有這些棒球隊運用

to win games.

魔球策略來贏球。

The great news for us as well

同樣對我們來說很棒的是

is that it's the way that we can transform

這是我們能夠改變

the American criminal justice system.

美國刑事司法體系的方式，

It's how we can make our streets safer,

這是我們可以讓街道更安全的方式，

we can reduce our prison costs,

我們可以減少監獄支出，

and we can make our system much fairer

我們可以讓體制更公平

and more just.

且更正義。

Some people call it data science.

有些人稱它為數據科學。

I call it moneyballing criminal justice.

我稱它為魔球的刑事司法。

Thank you.

謝謝大家。

(Applause)

（掌聲）