when anyone from slaves to soldiers, poets and politicians,

古希臘時期，

needed to make a big decision on life's most important questions,

不論是奴隸或士兵，詩人或政治家，

like, "Should I get married?"

當他們人生遇到重大問題時， 需要做出重要的決定，

or "Should we embark on this voyage?"

像是「我該結婚嗎？」

or "Should our army advance into this territory?"

或是「我該開始這次的航行嗎？」

they all consulted the oracle.

或是「我的士兵該進攻這個領地嗎？」

So this is how it worked:

他們都會請示先知。

you would bring her a question and you would get on your knees,

運行模式是這樣的：

and then she would go into this trance.

你把問題告訴她，接著屈膝跪下，

It would take a couple of days,

然後她就會進入出神狀態。

and then eventually she would come out of it,

這會花上幾天的時間，

giving you her predictions as your answer.

最終她會回神，

From the oracle bones of ancient China

答復你她的預知。

to ancient Greece to Mayan calendars,

從古中國的甲骨文， 到古希臘，再到馬雅曆，

people have craved for prophecy

人們都渴求著預言，

in order to find out what's going to happen next.

為了知道接下來會發生什麼事。

And that's because we all want to make the right decision.

而這是因為我們都想做正確的決定，

We don't want to miss something.

我們不希望漏掉了什麼。

The future is scary,

未來令人害怕。

so it's much nicer knowing that we can make a decision

所以能在某種程度上 保障決定的結果，是很棒的事。

with some assurance of the outcome.

我們有了新的先知，

Well, we have a new oracle,

名字叫大數據。

and it's name is big data,

也可以稱它為「華生」、 「深度學習」或「人工神經網路」。

or we call it "Watson" or "deep learning" or "neural net."

如今我們會問先知這樣的問題：

And these are the kinds of questions we ask of our oracle now,

「要將這批手機從中國 運到瑞典，怎樣最有效率？」

like, "What's the most efficient way to ship these phones

或是「我的小孩出生就有 遺傳疾病的機率是多少？」

from China to Sweden?"

或是「預期這產品的銷售量多少？」

Or, "What are the odds

我養了隻狗，名叫埃萊，最討厭下雨。

of my child being born with a genetic disorder?"

我用盡方法來訓練她， 讓她適應下雨。

Or, "What are the sales volume we can predict for this product?"

但因為我失敗了，

I have a dog. Her name is Elle, and she hates the rain.

我還是得諮詢一位叫 Dark Sky（天氣預報公司）的先知，

And I have tried everything to untrain her.

每次散步之前都會諮詢，

But because I have failed at this,

以獲得接下來十分鐘的準確天氣預報。

I also have to consult an oracle, called Dark Sky,

她真的很貼心。

every time before we go on a walk,

基於這些理由，我們的「先知」 是個 1220 億美元的產業。

for very accurate weather predictions in the next 10 minutes.

先不論這個產業的規模，

She's so sweet.

令人驚訝的是它極低的報酬率。

So because of all of this, our oracle is a $122 billion industry.

投資大數據很簡單，

Now, despite the size of this industry,

運用大數據卻很難。

the returns are surprisingly low.

73% 以上的大數據計畫根本不賺錢，

Investing in big data is easy,

有些業務主管跑來跟我說，

but using it is hard.

「我們都面臨了同樣的問題。

Over 73 percent of big data projects aren't even profitable,

我們投資了幾個大數據系統，

and I have executives coming up to me saying,

但我們的員工卻還是不能 做出更優的決定。

"We're experiencing the same thing.

他們當然也沒有想出 更多突破性的點子。」

We invested in some big data system,

這些對我來說都很有趣，

and our employees aren't making better decisions.

因為我是個科技人類學家。

And they're certainly not coming up with more breakthrough ideas."

我研究並給予公司建議，

So this is all really interesting to me,

告訴他們人們使用科技的形態，

because I'm a technology ethnographer.

我有興趣的領域之一就是數據。

I study and I advise companies

為什麼獲得更多數據 卻沒有幫我們做更好的決定，

on the patterns of how people use technology,

特別是那些有資源， 可以投資大數據系統的公司？

and one of my interest areas is data.

為什麼他們沒有更好地做決定？

So why is having more data not helping us make better decisions,

我第一時間就目睹了這項困境。

especially for companies who have all these resources

2009 年，我開始了 在諾基亞的研究工作。

to invest in these big data systems?

當時，諾基亞是世界上 最大的手機公司之一，

Why isn't it getting any easier for them?

在中國、墨西哥、印度等 新興市場中佔有主要地位──

So, I've witnessed the struggle firsthand.

我在這些地方都做了很多研究，

In 2009, I started a research position with Nokia.

研究低收入的人怎麼使用科技產品。

And at the time,

我在中國花了特別多時間

Nokia was one of the largest cell phone companies in the world,

來了解地下經濟。

dominating emerging markets like China, Mexico and India --

所以我當過街頭攤販，

all places where I had done a lot of research

賣水餃給建築工人。

on how low-income people use technology.

我也做過實地調查，

And I spent a lot of extra time in China

在網咖中日日夜夜地待著，

getting to know the informal economy.

和中國年輕人來往，這樣我才知道

So I did things like working as a street vendor

他們怎麼玩遊戲、使用手機，

selling dumplings to construction workers.

以及他們從農村地區 移居到城市時的使用情形。

Or I did fieldwork,

透過我收集的定性資料，

spending nights and days in internet cafés,

我開始清楚看見

hanging out with Chinese youth, so I could understand

即將發生在低收入中國人身上的巨變。

how they were using games and mobile phones

雖然他們身邊圍繞著奢侈品的廣告，

and using it between moving from the rural areas to the cities.

像是花俏的馬桶──誰不想要呢──

Through all of this qualitative evidence that I was gathering,

還有公寓和車，

I was starting to see so clearly

從和他們的對話中，

that a big change was about to happen among low-income Chinese people.

我發現最吸引他們的廣告，

Even though they were surrounded by advertisements for luxury products

是 iPhone 的廣告，

like fancy toilets -- who wouldn't want one? --

那些廣告向他們保證了 進入高科技生活的途徑。

and apartments and cars,

即使我和他們一起 住在這樣的城市貧民窟，

through my conversations with them,

我也看到人們將半個月以上的收入

I found out that the ads the actually enticed them the most

拿去買手機，

were the ones for iPhones,

而且越來越多都是「山寨品」，

promising them this entry into this high-tech life.

也就是他們買得起的 iPhone 或其他品牌的仿冒品。

And even when I was living with them in urban slums like this one,

這些仿冒品很堪使用。

I saw people investing over half of their monthly income

原廠有的功能都能用。

into buying a phone,

我和移民一起住、一起工作了數年，

and increasingly, they were "shanzhai,"

真的是他們做什麼，我就做什麼，

which are affordable knock-offs of iPhones and other brands.

我開始將所有數據拼湊在一起──

They're very usable.

不論是看似不相關的事， 像是我賣水餃的事，

Does the job.

或是較明顯相關的事，

And after years of living with migrants and working with them

像是追蹤他們花多少錢付手機費。

and just really doing everything that they were doing,

所以我才有辦法描繪出 這麼多整體畫面

I started piecing all these data points together --

來說明當時正發生什麼事。

from the things that seem random, like me selling dumplings,

這時我才開始理解到

to the things that were more obvious,

連中國最窮的人也想要智慧型手機，

like tracking how much they were spending on their cell phone bills.

且他們幾乎會不擇手段拿到手。

And I was able to create this much more holistic picture

你們要記得，

of what was happening.

當時是 2009 年，iPhone 才剛出現，

And that's when I started to realize

這是八年前的事，

that even the poorest in China would want a smartphone,

安卓手機才剛開始像 iPhone。

and that they would do almost anything to get their hands on one.

很多聰明又現實的人說，

You have to keep in mind,

「智慧型手機只是一時的流行。

iPhones had just come out, it was 2009,

誰會想帶著這麼重的東西到處走，

so this was, like, eight years ago,

又很快就沒電，

and Androids had just started looking like iPhones.

還會一掉地就壞？」

And a lot of very smart and realistic people said,

但我有很多數據，

"Those smartphones -- that's just a fad.

我對自己的洞察觀點非常有自信，

Who wants to carry around these heavy things

我興奮地把數據告訴諾基亞。

where batteries drain quickly and they break every time you drop them?"

但我沒能說服諾基亞，

But I had a lot of data,

因為那不是大數據。

and I was very confident about my insights,

他們說：「我們有幾百萬則數據，

so I was very excited to share them with Nokia.

而我們沒見到任何數據 指出有人想買智慧型手機，

But Nokia was not convinced,

你的 100 組數據太缺乏多樣性，

because it wasn't big data.

我們完全無法重視這項數據。」

They said, "We have millions of data points,

我說：「諾基亞，你說的沒錯。

and we don't see any indicators of anyone wanting to buy a smartphone,

你當然不會看到有人要買，

and your data set of 100, as diverse as it is, is too weak

因為你所發送問卷的假設前提

for us to even take seriously."

是人們不知道智慧型手機是什麼，

And I said, "Nokia, you're right.

所以你的數據當然不會反映

Of course you wouldn't see this,

兩年內想買智慧型手機的人的想法。

because you're sending out surveys assuming that people don't know

你問卷、研究方法的設計理念

what a smartphone is,

都是想讓現有的業務型態更好，

so of course you're not going to get any data back

而我關注的是這些正浮現的人類動態，

about people wanting to buy a smartphone in two years.

那些是過去沒有發生的，

Your surveys, your methods have been designed

我們看的是市場動態之外，

to optimize an existing business model,

這樣我們才能先走一步。」

and I'm looking at these emergent human dynamics

你們知道諾基亞怎麼樣了嗎？

that haven't happened yet.

他們的產業跌落谷底。

We're looking outside of market dynamics

這就是錯失的代價。

so that we can get ahead of it."

那代價是深不可測的。

Well, you know what happened to Nokia?

但不是只有諾基亞這樣。

Their business fell off a cliff.

我看到各機構一天到晚丟棄數據，

This -- this is the cost of missing something.

因為數據並非來自數量大的模型，

It was unfathomable.

或對不上數量大的模型數據。

But Nokia's not alone.

但這不是大數據的錯。

I see organizations throwing out data all the time

是我們用錯方法，

because it didn't come from a quant model

是我們的責任。

or it doesn't fit in one.

但一般認為大數據的成功之處

But it's not big data's fault.

在於量化的對象非常的特定，

It's the way we use big data; it's our responsibility.

像是電網、物流運送或遺傳密碼，

Big data's reputation for success

也就是些基本上可操縱的系統。

comes from quantifying very specific environments,

但並非所有的系統 都能被操縱得好好的。

like electricity power grids or delivery logistics or genetic code,

若你在量化的系統是動態的，

when we're quantifying in systems that are more or less contained.

特別是那些有人參與其中的系統，

But not all systems are as neatly contained.

會產生影響的事物複雜又難以預測，

When you're quantifying and systems are more dynamic,

我們不太知道怎樣建立這些模型。

especially systems that involve human beings,

即使你一時預測了人的行動，

forces are complex and unpredictable,

又會出現新的要素，

and these are things that we don't know how to model so well.

因為情況持續在改變。

Once you predict something about human behavior,

正因如此，這是個永無止境的迴圈。

new factors emerge,

你以為你瞭解了一件事，

because conditions are constantly changing.

另一件未知的事物便進入了你的視野。

That's why it's a never-ending cycle.

所以純粹依靠大數據

You think you know something,

便增加了我們錯失的機率，

and then something unknown enters the picture.

但同時讓我們以為我們無所不知。

And that's why just relying on big data alone

為什麼我們很難發現這個矛盾，

increases the chance that we'll miss something,

甚至也很難去理解它，

while giving us this illusion that we already know everything.

是因為我們有我所謂的「量化成見」，

And what makes it really hard to see this paradox

也就是無意識地認為可量化的

and even wrap our brains around it

比不可量化的更有價值。

is that we have this thing that I call the quantification bias,

我們工作時常有這樣的經驗。

which is the unconscious belief of valuing the measurable

或許我們和這樣想的同事一起工作，

over the immeasurable.

或者整個公司都這樣想，

And we often experience this at our work.

人們過於迷戀數字，

Maybe we work alongside colleagues who are like this,

以至於看不見除此之外的任何東西，

or even our whole entire company may be like this,

即使你將證據貼到他們臉上，給他們看。

where people become so fixated on that number,

這是個十分吸引人的訊息，

that they can't see anything outside of it,

因為量化並沒有錯；

even when you present them evidence right in front of their face.

量化事實上很讓人滿意。

And this is a very appealing message,

我看著 Excel 電子表格就覺得安心，

because there's nothing wrong with quantifying;

即使是很簡單的也一樣。

it's actually very satisfying.

（笑聲）

I get a great sense of comfort from looking at an Excel spreadsheet,

那種感覺就是，

even very simple ones.

「好的！方程式沒問題。 一切都很好。都在掌控之中。」

(Laughter)

問題是，

It's just kind of like,

量化會使人上癮。

"Yes! The formula worked. It's all OK. Everything is under control."

我們一旦忘記這件事，

But the problem is

若我們沒能做到時時確認是否上癮，

that quantifying is addictive.

我們很容易直接扔掉這樣的資料：

And when we forget that

僅僅因為它無法用數值量化。

and when we don't have something to kind of keep that in check,

很容易認為會有完美解決一切的絶招，

it's very easy to just throw out data

就好像有某種簡單的解決方法一樣。

because it can't be expressed as a numerical value.

因為這對任何一間機構來說， 都是危機的重要時刻，

It's very easy just to slip into silver-bullet thinking,

時常，我們要預測的未來，

as if some simple solution existed.

並不是在這安穩的草堆裡，

Because this is a great moment of danger for any organization,

而是在它之外， 是即將襲擊我們的暴風中心。

because oftentimes, the future we need to predict --

沒有什麼比對未知 一無所知來得有風險，

it isn't in that haystack,

那會使你做出錯誤的決定。

but it's that tornado that's bearing down on us

那可能使你錯失重要的事物。

outside of the barn.

但我們不用這樣做。

There is no greater risk

到頭來，是古希臘的先知 握有顯示道路的神秘鑰匙。

than being blind to the unknown.

近年的地質研究顯示，

It can cause you to make the wrong decisions.

最有名的先知所在的阿波羅神廟，

It can cause you to miss something big.

事實上座落在兩個地震斷層上。

But we don't have to go down this path.

這些斷層會從地殼下釋出石油煙氣，

It turns out that the oracle of ancient Greece

而那位先知就直接坐在那些斷層上方，

holds the secret key that shows us the path forward.

從縫隙中吸入數不盡的乙烯氣體。

Now, recent geological research has shown

（笑聲）

that the Temple of Apollo, where the most famous oracle sat,

那是真的。

was actually built over two earthquake faults.

（笑聲）

And these faults would release these petrochemical fumes

那都是真的，那就是為什麼 她講話含糊不清還看到幻覺，

from underneath the Earth's crust,

並進入類似出神的狀態。

and the oracle literally sat right above these faults,

她感覺自己都飛上天了！

inhaling enormous amounts of ethylene gas, these fissures.

（笑聲）

(Laughter)

所以大家要怎麼──

It's true.

大家要怎麼在這個狀態下 得到有用的建議？

(Laughter)

看到那些圍繞先知的人們了嗎？

It's all true, and that's what made her babble and hallucinate

你可以看到那些人支撐著她，

and go into this trance-like state.

因為她好像有點頭昏眼花？

She was high as a kite!

有沒有發現她左邊的男子

(Laughter)

正拿著橘色小冊子？

So how did anyone --

那些是神廟的引導人員，

How did anyone get any useful advice out of her

他們與先知密切合作。

in this state?

當有人來下跪詢問時，

Well, you see those people surrounding the oracle?

神廟的引導人員就開始工作了，

You see those people holding her up,

在來者向先知詢問一些問題後，

because she's, like, a little woozy?

他們會觀察來者的精神狀態，

And you see that guy on your left-hand side

然後他們會問來者一些後續問題，

holding the orange notebook?

像是：「為什麼你想知道 這個預言？你是誰？

Well, those were the temple guides,

你會怎麼運用這個資訊？」

and they worked hand in hand with the oracle.

接著神廟的引導人員會 用人類學的角度來看，

When inquisitors would come and get on their knees,

用質性資訊的角度來看，

that's when the temple guides would get to work,

然後翻譯先知含糊不清的話。

because after they asked her questions,

所以先知並非自己承攬一切任務，

they would observe their emotional state,

我們的大數據系統同樣也不該如此。

and then they would ask them follow-up questions,

我要澄清一下，

like, "Why do you want to know this prophecy? Who are you?

我並非在說大數據系統 在呼吸着乙烯氣體，

What are you going to do with this information?"

甚至給予沒用的預測。

And then the temple guides would take this more ethnographic,

完全相反。

this more qualitative information,

我想說的是，

and interpret the oracle's babblings.

就像先知需要神廟的引導人員那樣，

So the oracle didn't stand alone,

大數據系統同樣也需要。

and neither should our big data systems.

大數據需要人類學家以及用戶研究人員

Now to be clear,

來收集我所謂的「厚數據」──

I'm not saying that big data systems are huffing ethylene gas,

來自於人們的寶貴數據，

or that they're even giving invalid predictions.