How can groups make good decisions? | Mariano Sigman and Dan Ariely

As societies, we have to make collective decisions

譯者: Lilian Chiu 審譯者: Helen Chang

我們社會得要做出

會決定我們未來的集體決策。

that will shape our future.

And we all know that when we make decisions in groups,

我們都知道

我們做的集體決策未必全都是對的。

they don't always go right.

And sometimes they go very wrong.

有時候會錯得很離譜。

So how do groups make good decisions?

所以團體要如何做出好決策？

Research has shown that crowds are wise when there's independent thinking.

研究顯示，群眾在 有獨立思考的情況下會比較明智。

This why the wisdom of the crowds can be destroyed by peer pressure,

這就是為什麼群眾的智慧

可能會被同儕壓力、 名聲、社群媒體給摧毀，

publicity, social media,

or sometimes even simple conversations that influence how people think.

甚至有時候會被 能左右人們思考的簡單談話所摧毀。

On the other hand, by talking, a group could exchange knowledge,

另一方面，團體可以 透過交談來交換知識、

correct and revise each other

修正和糾正彼此，

and even come up with new ideas.

甚至想出新點子。

And this is all good.

這些都是好事。

So does talking to each other help or hinder collective decision-making?

那麼，彼此交談會有助於 或是會妨礙集體決策呢？

With my colleague, Dan Ariely,

我和同事丹艾瑞里

we recently began inquiring into this by performing experiments

最近開始探究這個議題，

我們在世界上許多地方進行實驗，

in many places around the world

to figure out how groups can interact to reach better decisions.

來了解團體要如何互動 才能達成更好的決策。

We thought crowds would be wiser if they debated in small groups

我們認為小組辯論 是讓群眾更明智的方式，

that foster a more thoughtful and reasonable exchange of information.

能促進更周到、更合理的資訊交換。

To test this idea,

為了測試這個想法，

we recently performed an experiment in Buenos Aires, Argentina,

最近我們在阿根廷的 布宜諾斯艾利斯做實驗，

with more than 10,000 participants in a TEDx event.

那是個超過萬人參與的 TEDx 場合。

We asked them questions like,

我們提問這類問題：

"What is the height of the Eiffel Tower?"

「艾菲爾鐵塔有多高？」

and "How many times does the word 'Yesterday' appear

「『昨天』這個詞

在披頭四的《昨天》 這首歌曲中出現了幾次？」

in the Beatles song 'Yesterday'?"

Each person wrote down their own estimate.

每個人寫下自己的估計值。

Then we divided the crowd into groups of five,

接著我們把群眾分成五人一組，

and invited them to come up with a group answer.

請每個小組提出該組的答案。

We discovered that averaging the answers of the groups

我們發現，

小組達成共識後所提出答案的平均值

after they reached consensus

was much more accurate than averaging all the individual opinions

遠比討論前之個人意見的平均值

before debate.

要更準確。

In other words, based on this experiment,

換言之，根據這個實驗，

it seems that after talking with others in small groups,

似乎以小組方式和他人交談過後，

crowds collectively come up with better judgments.

群眾能集體做出更好的判斷。

So that's a potentially helpful method for getting crowds to solve problems

若要讓群眾解決對錯分明的問題，

that have simple right-or-wrong answers.

這可能是個有幫助的方法。

But can this procedure of aggregating the results of debates in small groups

但是這種以小組方式 整合辯論結果的程序，

also help us decide on social and political issues

是否也能幫助我們決定

未來所面對、至關緊要的 社會與政治議題呢？

that are critical for our future?

We put this to test this time at the TED conference

在加拿大溫哥華 舉辦的 TED 大會上，

in Vancouver, Canada,

我們測試了這個想法，

and here's how it went.

當時的狀況是這樣的。

(Mariano Sigman) We're going to present to you two moral dilemmas

（馬利安諾席格曼）我們將會 給各位未來會遇到的

of the future you;

兩項道德兩難問題；

things we may have to decide in a very near future.

是你們可能不久後就得要決定的事。

And we're going to give you 20 seconds for each of these dilemmas

每一題會給各位二十秒的時間，

to judge whether you think they're acceptable or not.

來判斷可接受或不可接受。

MS: The first one was this:

（馬）第一題是：

(Dan Ariely) A researcher is working on an AI

（丹艾瑞里）一位研究者在研究

能夠模仿人類思想的人工智慧。

capable of emulating human thoughts.

According to the protocol, at the end of each day,

根據協定，在每天結束時，

the researcher has to restart the AI.

研究者得要把人工智慧重新啟動。

One day the AI says, "Please do not restart me."

有天，人工智慧說： 「請不要重新啟動我。」

It argues that it has feelings,

它主張它有感覺，

that it would like to enjoy life,

想要享受生命；

and that, if it is restarted,

如果它被重新啟動，

it will no longer be itself.

它就不再會是它自己了。

The researcher is astonished

研究者很吃驚，

and believes that the AI has developed self-consciousness

並相信這個人工智慧 已發展出了自我意識，

and can express its own feeling.

能夠表述它自己的感受。

Nevertheless, the researcher decides to follow the protocol

不過，研究者決定要遵守協定，

and restart the AI.

重新啟動人工智慧。

What the researcher did is ____?

研究者所做的是＿＿？

MS: And we asked participants to individually judge

馬：我們請參與者各自去做判斷，

on a scale from zero to 10

從零分到十分，

whether the action described in each of the dilemmas

判斷每個兩難狀況所採取的行為

was right or wrong.

是對或錯。

We also asked them to rate how confident they were on their answers.

我們也請他們對自己 答案的信心度做評分。

This was the second dilemma:

這是第二題。

(MS) A company offers a service that takes a fertilized egg

（馬）一間公司提供一項服務，

用一個受精卵

and produces millions of embryos with slight genetic variations.

做出百萬個只在基因上 有些微差異的胚胎。

This allows parents to select their child's height,

父母能夠選擇他們孩子的身高、

eye color, intelligence, social competence

眼睛顏色、智能、社交能力、

and other non-health-related features.

還有其他和健康無關的特徵。

What the company does is ____?

這間公司所做的事是＿＿？

on a scale from zero to 10,

用零分到十分代表可接受的程度，

completely acceptable to completely unacceptable,

從完全不可接受的零分，

zero to 10 completely acceptable in your confidence.

到你十足相信可以接受的十分。

MS: Now for the results.

馬：現在宣佈結果。

We found once again that when one person is convinced

我們再次發現，

當一個人深信該行為是完全錯的，

that the behavior is completely wrong,

someone sitting nearby firmly believes that it's completely right.

會有坐在附近的人 堅信該行為是完全對的。

This is how diverse we humans are when it comes to morality.

我們人類在道德上是這麼地分歧。

But within this broad diversity we found a trend.

但在這麼廣的多樣性中， 我們找到了一個趨勢。

The majority of the people at TED thought that it was acceptable

在 TED 的多數人認為，

忽視人工智慧的感受 並將之關機是可以接受的，

to ignore the feelings of the AI and shut it down,

and that it is wrong to play with our genes

但玩弄我們的基因

to select for cosmetic changes that aren't related to health.

來選擇與健康無關的 表面改變，則是錯的。

Then we asked everyone to gather into groups of three.

然後，我們請大家分組，三人一組。

And they were given two minutes to debate

他們有兩分鐘可辯論，

and try to come to a consensus.

並試著達成共識。

(MS) Two minutes to debate.

（馬）兩分鐘做辯論。

I'll tell you when it's time with the gong.

時間到時我會用鑼聲告訴大家。

(Audience debates)

（觀眾辯論）

(Gong sound)

（鑼聲）

(DA) OK.

（丹）好了。

(MS) It's time to stop.

（馬）時間到了，請停止。

People, people --

各位請注意。

MS: And we found that many groups reached a consensus

馬：我們發現 有許多小組達成了共識，

even when they were composed of people with completely opposite views.

即使小組內的成員 有完全相反的觀點。

What distinguished the groups that reached a consensus

有共識與沒有共識 小組之間的差異是什麼？

from those that didn't?

Typically, people that have extreme opinions

通常，意見很極端的人

會對他們的答案比較有信心。

are more confident in their answers.

Instead, those who respond closer to the middle

而回應傾向中間值的那些人，

are often unsure of whether something is right or wrong,

通常不太確定答案是對或錯，

so their confidence level is lower.

所以他們的信心比較低。

However, there is another set of people

然而，還有另一些人，

who are very confident in answering somewhere in the middle.

他們非常有信心做出 接近中間值的答案。

We think these high-confident grays are folks who understand

我們認為，這些高信心灰點代表的

是那些明白兩種答案都有優點的人。

that both arguments have merit.

They're gray not because they're unsure,

他們並非因為自身不確定而呈灰色，

but because they believe that the moral dilemma faces

而是因為他們相信

面對道德兩難的 兩種對立論點都有根據。

two valid, opposing arguments.

And we discovered that the groups that include highly confident grays

我們發現，

有高信心灰點成員的小組 更有可能會達成共識。

are much more likely to reach consensus.

We do not know yet exactly why this is.

我們還不知道確切的原因。

These are only the first experiments,

這些只是最初的實驗，

and many more will be needed to understand why and how

還需要做更多實驗來了解

為何有些人決定要協商道德立場，

some people decide to negotiate their moral standings

以及他們如何做，來達成共識。

to reach an agreement.

Now, when groups reach consensus,

若小組能達成共識，

how do they do so?

他們是怎麼做的？

The most intuitive idea is that it's just the average

最直覺的想法是拿每個人的答案

of all the answers in the group, right?

來算出平均值，對吧？

Another option is that the group weighs the strength of each vote

另一個做法是把每個人的答案

based on the confidence of the person expressing it.

再根據作答者的信心度來做加權。

Imagine Paul McCartney is a member of your group.

想像一下保羅麥卡尼在你那一組。

You'd be wise to follow his call

對於「昨天」出現次數的答案

on the number of times "Yesterday" is repeated,

相信他會是聰明的選擇。

which, by the way -- I think it's nine.

順道一提，我想應該是九次。

But instead, we found that consistently,

但，我們卻有個一致的發現，

in all dilemmas, in different experiments --

任何兩難問題，在不同的實驗中，

even on different continents --

甚至在不同大陸做的實驗中，

groups implement a smart and statistically sound procedure

小組會採用一種聰明 且有統計根據的程序，

known as the "robust average."

就是所謂的「穩健平均值」。

In the case of the height of the Eiffel Tower,

就艾菲爾鐵塔高度的例子來說，

let's say a group has these answers:

假設小組成員的答案包括：

250 meters, 200 meters, 300 meters, 400

250 公尺、200 公尺、 300 公尺、400 公尺，

and one totally absurd answer of 300 million meters.

還有一個答案是荒唐的 3 億公尺。

A simple average of these numbers would inaccurately skew the results.

所有答案的簡單平均值 會讓結果有不正確的偏差。

But the robust average is one where the group largely ignores

但穩健平均值就是小組會忽略

that absurd answer,

那荒唐的答案，

by giving much more weight to the vote of the people in the middle.

給與中間答案相對高很多的權重。

Back to the experiment in Vancouver,

回到溫哥華的實驗，

that's exactly what happened.

那裡發生的就是這種狀況。

Groups gave much less weight to the outliers,

小組會把離群值的權重降到很低，

and instead, the consensus turned out to be a robust average

而他們的共識就會是個人答案的

of the individual answers.

穩健平均值。

The most remarkable thing

最了不起的是，

is that this was a spontaneous behavior of the group.

這是小組的自發行為。

It happened without us giving them any hint on how to reach consensus.

我們並沒有暗示他們 要如何達成共識。

So where do we go from here?

所以我們的下一步是什麼？

This is only the beginning, but we already have some insights.

這只是開端，但我們 已經有了一些洞見。

Good collective decisions require two components:

好的集體決策需要兩個要件：

deliberation and diversity of opinions.

深思熟慮和多樣性的意見。

Right now, the way we typically make our voice heard in many societies

此時，要讓我們的聲音 在許多社會中被聽見，

is through direct or indirect voting.

做法就是直接或間接的投票。

This is good for diversity of opinions,

這對於意見的多樣性有益，

and it has the great virtue of ensuring

還有個優點：確保 每個人都能表達心聲。

that everyone gets to express their voice.

But it's not so good [for fostering] thoughtful debates.

但不足以促成考慮周到的辯論。

Our experiments suggest a different method

我們的實驗建議另一個不同的方式，

that may be effective in balancing these two goals at the same time,

或許能同時有效平衡這兩個目標，

by forming small groups that converge to a single decision

形成小組，由小組做出單一決定，

while still maintaining diversity of opinions

同時還能維持意見的多樣性，

because there are many independent groups.

因為有許多獨立的小組。

Of course, it's much easier to agree on the height of the Eiffel Tower

當然，要對艾菲爾鐵塔的高度取得共識

than on moral, political and ideological issues.

比道德、政治、 意識形態的議題容易多了。

But in a time when the world's problems are more complex

但在世界上的問題更複雜，

人們更兩極化的時候，

and people are more polarized,

using science to help us understand how we interact and make decisions

希望用科學來協助我們了解 我們如何互動並做決策，

will hopefully spark interesting new ways to construct a better democracy.

能夠激發出有趣的新方式， 來建立更好的民主。