TEDxCambridge
Tricia Wang: The human insights missing from big data
王聖捷: 因大數據而失準的視野
Filmed:
Readability: 4.5
1,688,539 views
為什麼即使擁有空前巨大的數據量，還有這麼多公司選錯了？藉由諾基亞、網飛、和古希臘先知的故事，王聖捷簡潔易懂地解釋了大數據是什麼，還揭示了大數據隱藏的陷阱。她建議我們應重視「厚數據」──也就是與現實中的人們溝通而獲得的寶貴、不可計量的洞察觀點──以期做出正確的商業選擇，並戰勝未知。
Tricia Wang - Technology ethnographer
With astronaut eyes and ethnographer curiosity, Tricia Wang helps corporations grow by discovering the unknown about their customers. Full bio
With astronaut eyes and ethnographer curiosity, Tricia Wang helps corporations grow by discovering the unknown about their customers. Full bio
Double-click the English transcript below to play the video.
00:12
In ancient Greece,
0
885
1545
古希臘時期，
00:15
when anyone from slaves to soldiers,
poets and politicians,
poets and politicians,
1
3436
3943
不論是奴隸或士兵，詩人或政治家，
00:19
needed to make a big decision
on life's most important questions,
on life's most important questions,
2
7403
4004
當他們人生遇到重大問題時，
需要做出重要的決定，
需要做出重要的決定，
00:23
like, "Should I get married?"
3
11431
1391
像是「我該結婚嗎？」
00:24
or "Should we embark on this voyage?"
4
12846
1857
或是「我該開始這次的航行嗎？」
00:26
or "Should our army
advance into this territory?"
advance into this territory?"
5
14727
2928
或是「我的士兵該進攻這個領地嗎？」
00:29
they all consulted the oracle.
6
17679
2579
他們都會請示先知。
00:33
So this is how it worked:
7
21020
1440
運行模式是這樣的：
00:34
you would bring her a question
and you would get on your knees,
and you would get on your knees,
8
22484
3112
你把問題告訴她，接著屈膝跪下，
00:37
and then she would go into this trance.
9
25620
1871
然後她就會進入出神狀態。
00:39
It would take a couple of days,
10
27515
1549
這會花上幾天的時間，
00:41
and then eventually
she would come out of it,
she would come out of it,
11
29088
2163
最終她會回神，
00:43
giving you her predictions as your answer.
12
31275
2536
答復你她的預知。
00:46
From the oracle bones of ancient China
13
34910
2566
從古中國的甲骨文，
到古希臘，再到馬雅曆，
到古希臘，再到馬雅曆，
00:49
to ancient Greece to Mayan calendars,
14
37500
2345
00:51
people have craved for prophecy
15
39869
2296
人們都渴求著預言，
00:54
in order to find out
what's going to happen next.
what's going to happen next.
16
42189
3137
為了知道接下來會發生什麼事。
00:58
And that's because we all want
to make the right decision.
to make the right decision.
17
46516
3239
而這是因為我們都想做正確的決定，
01:01
We don't want to miss something.
18
49779
1545
我們不希望漏掉了什麼。
01:03
The future is scary,
19
51892
1743
未來令人害怕。
01:05
so it's much nicer
knowing that we can make a decision
knowing that we can make a decision
20
53659
2717
所以能在某種程度上
保障決定的結果，是很棒的事。
保障決定的結果，是很棒的事。
01:08
with some assurance of the outcome.
21
56400
1982
01:11
Well, we have a new oracle,
22
59079
1611
我們有了新的先知，
01:12
and it's name is big data,
23
60714
2145
名字叫大數據。
01:14
or we call it "Watson"
or "deep learning" or "neural net."
or "deep learning" or "neural net."
24
62883
3939
也可以稱它為「華生」、
「深度學習」或「人工神經網路」。
「深度學習」或「人工神經網路」。
01:19
And these are the kinds of questions
we ask of our oracle now,
we ask of our oracle now,
25
67340
4012
如今我們會問先知這樣的問題：
01:23
like, "What's the most efficient way
to ship these phones
to ship these phones
26
71376
3922
「要將這批手機從中國
運到瑞典，怎樣最有效率？」
運到瑞典，怎樣最有效率？」
01:27
from China to Sweden?"
27
75322
1823
01:29
Or, "What are the odds
28
77169
1800
或是「我的小孩出生就有
遺傳疾病的機率是多少？」
遺傳疾病的機率是多少？」
01:30
of my child being born
with a genetic disorder?"
with a genetic disorder?"
29
78993
3363
01:34
Or, "What are the sales volume
we can predict for this product?"
we can predict for this product?"
30
82952
3244
或是「預期這產品的銷售量多少？」
01:40
I have a dog. Her name is Elle,
and she hates the rain.
and she hates the rain.
31
88108
4047
我養了隻狗，名叫埃萊，最討厭下雨。
01:44
And I have tried everything
to untrain her.
to untrain her.
32
92179
3306
我用盡方法來訓練她，
讓她適應下雨。
讓她適應下雨。
01:47
But because I have failed at this,
33
95509
2771
但因為我失敗了，
01:50
I also have to consult
an oracle, called Dark Sky,
an oracle, called Dark Sky,
34
98304
3286
我還是得諮詢一位叫
Dark Sky（天氣預報公司）的先知，
Dark Sky（天氣預報公司）的先知，
01:53
every time before we go on a walk,
35
101614
1635
每次散步之前都會諮詢，
01:55
for very accurate weather predictions
in the next 10 minutes.
in the next 10 minutes.
36
103273
3577
以獲得接下來十分鐘的準確天氣預報。
02:01
She's so sweet.
37
109535
1303
她真的很貼心。
02:03
So because of all of this,
our oracle is a $122 billion industry.
our oracle is a $122 billion industry.
38
111827
5707
基於這些理由，我們的「先知」
是個 1220 億美元的產業。
是個 1220 億美元的產業。
02:10
Now, despite the size of this industry,
39
118006
3376
先不論這個產業的規模，
02:13
the returns are surprisingly low.
40
121406
2456
令人驚訝的是它極低的報酬率。
02:16
Investing in big data is easy,
41
124342
2494
投資大數據很簡單，
02:18
but using it is hard.
42
126860
1933
運用大數據卻很難。
02:21
Over 73 percent of big data projects
aren't even profitable,
aren't even profitable,
43
129981
4040
73% 以上的大數據計畫根本不賺錢，
02:26
and I have executives
coming up to me saying,
coming up to me saying,
44
134045
2431
有些業務主管跑來跟我說，
02:28
"We're experiencing the same thing.
45
136500
1789
「我們都面臨了同樣的問題。
02:30
We invested in some big data system,
46
138313
1753
我們投資了幾個大數據系統，
02:32
and our employees aren't making
better decisions.
better decisions.
47
140090
2968
但我們的員工卻還是不能
做出更優的決定。
做出更優的決定。
02:35
And they're certainly not coming up
with more breakthrough ideas."
with more breakthrough ideas."
48
143082
3162
他們當然也沒有想出
更多突破性的點子。」
更多突破性的點子。」
02:38
So this is all really interesting to me,
49
146914
3184
這些對我來說都很有趣，
02:42
because I'm a technology ethnographer.
50
150122
2010
因為我是個科技人類學家。
02:44
I study and I advise companies
51
152630
2564
我研究並給予公司建議，
02:47
on the patterns
of how people use technology,
of how people use technology,
52
155218
2483
告訴他們人們使用科技的形態，
02:49
and one of my interest areas is data.
53
157725
2678
我有興趣的領域之一就是數據。
02:52
So why is having more data
not helping us make better decisions,
not helping us make better decisions,
54
160427
5193
為什麼獲得更多數據
卻沒有幫我們做更好的決定，
卻沒有幫我們做更好的決定，
02:57
especially for companies
who have all these resources
who have all these resources
55
165644
2783
特別是那些有資源，
可以投資大數據系統的公司？
可以投資大數據系統的公司？
03:00
to invest in these big data systems?
56
168451
1736
03:02
Why isn't it getting any easier for them?
57
170211
2398
為什麼他們沒有更好地做決定？
03:05
So, I've witnessed the struggle firsthand.
58
173990
2634
我第一時間就目睹了這項困境。
03:09
In 2009, I started
a research position with Nokia.
a research position with Nokia.
59
177374
3484
2009 年，我開始了
在諾基亞的研究工作。
在諾基亞的研究工作。
03:13
And at the time,
60
181232
1158
當時，諾基亞是世界上
最大的手機公司之一，
最大的手機公司之一，
03:14
Nokia was one of the largest
cell phone companies in the world,
cell phone companies in the world,
61
182414
3158
03:17
dominating emerging markets
like China, Mexico and India --
like China, Mexico and India --
62
185596
3202
在中國、墨西哥、印度等
新興市場中佔有主要地位──
新興市場中佔有主要地位──
03:20
all places where I had done
a lot of research
a lot of research
63
188822
2502
我在這些地方都做了很多研究，
03:23
on how low-income people use technology.
64
191348
2676
研究低收入的人怎麼使用科技產品。
03:26
And I spent a lot of extra time in China
65
194048
2330
我在中國花了特別多時間
03:28
getting to know the informal economy.
66
196402
2592
來了解地下經濟。
03:31
So I did things like working
as a street vendor
as a street vendor
67
199018
2401
所以我當過街頭攤販，
03:33
selling dumplings to construction workers.
68
201443
2574
賣水餃給建築工人。
03:36
Or I did fieldwork,
69
204041
1358
我也做過實地調查，
03:37
spending nights and days
in internet cafés,
in internet cafés,
70
205423
2958
在網咖中日日夜夜地待著，
03:40
hanging out with Chinese youth,
so I could understand
so I could understand
71
208405
2546
和中國年輕人來往，這樣我才知道
03:42
how they were using
games and mobile phones
games and mobile phones
72
210975
2284
他們怎麼玩遊戲、使用手機，
03:45
and using it between moving
from the rural areas to the cities.
from the rural areas to the cities.
73
213283
3370
以及他們從農村地區
移居到城市時的使用情形。
移居到城市時的使用情形。
03:50
Through all of this qualitative evidence
that I was gathering,
that I was gathering,
74
218335
3927
透過我收集的定性資料，
03:54
I was starting to see so clearly
75
222286
2824
我開始清楚看見
03:57
that a big change was about to happen
among low-income Chinese people.
among low-income Chinese people.
76
225134
4472
即將發生在低收入中國人身上的巨變。
04:03
Even though they were surrounded
by advertisements for luxury products
by advertisements for luxury products
77
231020
4367
雖然他們身邊圍繞著奢侈品的廣告，
04:07
like fancy toilets --
who wouldn't want one? --
who wouldn't want one? --
78
235411
3495
像是花俏的馬桶──誰不想要呢──
04:10
and apartments and cars,
79
238930
2890
還有公寓和車，
04:13
through my conversations with them,
80
241844
1820
從和他們的對話中，
04:15
I found out that the ads
the actually enticed them the most
the actually enticed them the most
81
243688
3841
我發現最吸引他們的廣告，
04:19
were the ones for iPhones,
82
247553
1996
是 iPhone 的廣告，
04:21
promising them this entry
into this high-tech life.
into this high-tech life.
83
249573
3052
那些廣告向他們保證了
進入高科技生活的途徑。
進入高科技生活的途徑。
04:25
And even when I was living with them
in urban slums like this one,
in urban slums like this one,
84
253469
3163
即使我和他們一起
住在這樣的城市貧民窟，
住在這樣的城市貧民窟，
04:28
I saw people investing
over half of their monthly income
over half of their monthly income
85
256656
2996
我也看到人們將半個月以上的收入
04:31
into buying a phone,
86
259676
1623
拿去買手機，
04:33
and increasingly, they were "shanzhai,"
87
261323
2302
而且越來越多都是「山寨品」，
04:35
which are affordable knock-offs
of iPhones and other brands.
of iPhones and other brands.
88
263649
3388
也就是他們買得起的
iPhone 或其他品牌的仿冒品。
iPhone 或其他品牌的仿冒品。
04:40
They're very usable.
89
268303
1625
這些仿冒品很堪使用。
04:42
Does the job.
90
270890
1322
原廠有的功能都能用。
04:44
And after years of living
with migrants and working with them
with migrants and working with them
91
272750
5789
我和移民一起住、一起工作了數年，
04:50
and just really doing everything
that they were doing,
that they were doing,
92
278563
3434
真的是他們做什麼，我就做什麼，
04:54
I started piecing
all these data points together --
all these data points together --
93
282021
3597
我開始將所有數據拼湊在一起──
04:57
from the things that seem random,
like me selling dumplings,
like me selling dumplings,
94
285642
3123
不論是看似不相關的事，
像是我賣水餃的事，
像是我賣水餃的事，
05:00
to the things that were more obvious,
95
288789
1804
或是較明顯相關的事，
像是追蹤他們花多少錢付手機費。
05:02
like tracking how much they were spending
on their cell phone bills.
on their cell phone bills.
96
290617
3232
所以我才有辦法描繪出
這麼多整體畫面
這麼多整體畫面
05:05
And I was able to create
this much more holistic picture
this much more holistic picture
97
293873
2639
05:08
of what was happening.
98
296536
1156
來說明當時正發生什麼事。
05:09
And that's when I started to realize
99
297716
1722
這時我才開始理解到
05:11
that even the poorest in China
would want a smartphone,
would want a smartphone,
100
299462
3509
連中國最窮的人也想要智慧型手機，
05:14
and that they would do almost anything
to get their hands on one.
to get their hands on one.
101
302995
4985
且他們幾乎會不擇手段拿到手。
05:21
You have to keep in mind,
102
309073
2404
你們要記得，
05:23
iPhones had just come out, it was 2009,
103
311501
3084
當時是 2009 年，iPhone 才剛出現，
05:26
so this was, like, eight years ago,
104
314609
1885
這是八年前的事，
05:28
and Androids had just started
looking like iPhones.
looking like iPhones.
105
316518
2437
安卓手機才剛開始像 iPhone。
05:30
And a lot of very smart
and realistic people said,
and realistic people said,
106
318979
2507
很多聰明又現實的人說，
05:33
"Those smartphones -- that's just a fad.
107
321510
2207
「智慧型手機只是一時的流行。
05:36
Who wants to carry around
these heavy things
these heavy things
108
324243
2996
誰會想帶著這麼重的東西到處走，
05:39
where batteries drain quickly
and they break every time you drop them?"
and they break every time you drop them?"
109
327263
3487
又很快就沒電，
還會一掉地就壞？」
05:44
But I had a lot of data,
110
332793
1201
但我有很多數據，
05:46
and I was very confident
about my insights,
about my insights,
111
334018
2260
我對自己的洞察觀點非常有自信，
05:48
so I was very excited
to share them with Nokia.
to share them with Nokia.
112
336302
2829
我興奮地把數據告訴諾基亞。
05:53
But Nokia was not convinced,
113
341332
2517
但我沒能說服諾基亞，
05:55
because it wasn't big data.
114
343873
2335
因為那不是大數據。
05:59
They said, "We have
millions of data points,
millions of data points,
115
347022
2404
他們說：「我們有幾百萬則數據，
06:01
and we don't see any indicators
of anyone wanting to buy a smartphone,
of anyone wanting to buy a smartphone,
116
349450
4247
而我們沒見到任何數據
指出有人想買智慧型手機，
指出有人想買智慧型手機，
06:05
and your data set of 100,
as diverse as it is, is too weak
as diverse as it is, is too weak
117
353721
4388
你的 100 組數據太缺乏多樣性，
06:10
for us to even take seriously."
118
358133
1714
我們完全無法重視這項數據。」
06:12
And I said, "Nokia, you're right.
119
360908
1605
我說：「諾基亞，你說的沒錯。
06:14
Of course you wouldn't see this,
120
362537
1560
你當然不會看到有人要買，
06:16
because you're sending out surveys
assuming that people don't know
assuming that people don't know
121
364121
3371
因為你所發送問卷的假設前提
是人們不知道智慧型手機是什麼，
06:19
what a smartphone is,
122
367516
1159
所以你的數據當然不會反映
06:20
so of course you're not going
to get any data back
to get any data back
123
368699
2366
06:23
about people wanting to buy
a smartphone in two years.
a smartphone in two years.
124
371089
2572
兩年內想買智慧型手機的人的想法。
06:25
Your surveys, your methods
have been designed
have been designed
125
373685
2118
你問卷、研究方法的設計理念
06:27
to optimize an existing business model,
126
375827
2022
都是想讓現有的業務型態更好，
06:29
and I'm looking
at these emergent human dynamics
at these emergent human dynamics
127
377873
2608
而我關注的是這些正浮現的人類動態，
06:32
that haven't happened yet.
128
380505
1354
那些是過去沒有發生的，
06:33
We're looking outside of market dynamics
129
381883
2438
我們看的是市場動態之外，
06:36
so that we can get ahead of it."
130
384345
1631
這樣我們才能先走一步。」
06:39
Well, you know what happened to Nokia?
131
387373
2244
你們知道諾基亞怎麼樣了嗎？
06:41
Their business fell off a cliff.
132
389641
2365
他們的產業跌落谷底。
06:44
This -- this is the cost
of missing something.
of missing something.
133
392791
3727
這就是錯失的代價。
06:49
It was unfathomable.
134
397163
1999
那代價是深不可測的。
06:52
But Nokia's not alone.
135
400003
1651
但不是只有諾基亞這樣。
06:54
I see organizations
throwing out data all the time
throwing out data all the time
136
402258
2581
我看到各機構一天到晚丟棄數據，
06:56
because it didn't come from a quant model
137
404863
2561
因為數據並非來自數量大的模型，
06:59
or it doesn't fit in one.
138
407448
1768
或對不上數量大的模型數據。
07:02
But it's not big data's fault.
139
410219
2048
但這不是大數據的錯。
07:04
It's the way we use big data;
it's our responsibility.
it's our responsibility.
140
412942
3907
是我們用錯方法，
是我們的責任。
07:09
Big data's reputation for success
141
417730
1911
但一般認為大數據的成功之處
07:11
comes from quantifying
very specific environments,
very specific environments,
142
419665
3759
在於量化的對象非常的特定，
07:15
like electricity power grids
or delivery logistics or genetic code,
or delivery logistics or genetic code,
143
423448
4913
像是電網、物流運送或遺傳密碼，
07:20
when we're quantifying in systems
that are more or less contained.
that are more or less contained.
144
428385
4318
也就是些基本上可操縱的系統。
07:24
But not all systems
are as neatly contained.
are as neatly contained.
145
432727
2969
但並非所有的系統
都能被操縱得好好的。
都能被操縱得好好的。
07:27
When you're quantifying
and systems are more dynamic,
and systems are more dynamic,
146
435720
3258
若你在量化的系統是動態的，
07:31
especially systems
that involve human beings,
that involve human beings,
147
439002
3799
特別是那些有人參與其中的系統，
07:34
forces are complex and unpredictable,
148
442825
2426
會產生影響的事物複雜又難以預測，
07:37
and these are things
that we don't know how to model so well.
that we don't know how to model so well.
149
445275
3486
我們不太知道怎樣建立這些模型。
07:41
Once you predict something
about human behavior,
about human behavior,
150
449204
2813
即使你一時預測了人的行動，
07:44
new factors emerge,
151
452041
1855
又會出現新的要素，
07:45
because conditions
are constantly changing.
are constantly changing.
152
453920
2365
因為情況持續在改變。
07:48
That's why it's a never-ending cycle.
153
456309
1803
正因如此，這是個永無止境的迴圈。
07:50
You think you know something,
154
458136
1464
你以為你瞭解了一件事，
07:51
and then something unknown
enters the picture.
enters the picture.
155
459624
2242
另一件未知的事物便進入了你的視野。
07:53
And that's why just relying
on big data alone
on big data alone
156
461890
3322
所以純粹依靠大數據
07:57
increases the chance
that we'll miss something,
that we'll miss something,
157
465236
2849
便增加了我們錯失的機率，
08:00
while giving us this illusion
that we already know everything.
that we already know everything.
158
468109
3777
但同時讓我們以為我們無所不知。
08:04
And what makes it really hard
to see this paradox
to see this paradox
159
472406
3856
為什麼我們很難發現這個矛盾，
08:08
and even wrap our brains around it
160
476286
2659
甚至也很難去理解它，
08:10
is that we have this thing
that I call the quantification bias,
that I call the quantification bias,
161
478969
3691
是因為我們有我所謂的「量化成見」，
08:14
which is the unconscious belief
of valuing the measurable
of valuing the measurable
162
482684
3922
也就是無意識地認為可量化的
比不可量化的更有價值。
08:18
over the immeasurable.
163
486630
1594
08:21
And we often experience this at our work.
164
489222
3284
我們工作時常有這樣的經驗。
08:24
Maybe we work alongside
colleagues who are like this,
colleagues who are like this,
165
492530
2650
或許我們和這樣想的同事一起工作，
08:27
or even our whole entire
company may be like this,
company may be like this,
166
495204
2428
或者整個公司都這樣想，
08:29
where people become
so fixated on that number,
so fixated on that number,
167
497656
2546
人們過於迷戀數字，
08:32
that they can't see anything
outside of it,
outside of it,
168
500226
2067
以至於看不見除此之外的任何東西，
08:34
even when you present them evidence
right in front of their face.
right in front of their face.
169
502317
3948
即使你將證據貼到他們臉上，給他們看。
08:39
And this is a very appealing message,
170
507123
3371
這是個十分吸引人的訊息，
08:42
because there's nothing
wrong with quantifying;
wrong with quantifying;
171
510518
2343
因為量化並沒有錯；
08:44
it's actually very satisfying.
172
512885
1430
量化事實上很讓人滿意。
08:46
I get a great sense of comfort
from looking at an Excel spreadsheet,
from looking at an Excel spreadsheet,
173
514339
4362
我看著 Excel 電子表格就覺得安心，
08:50
even very simple ones.
174
518725
1401
即使是很簡單的也一樣。
08:52
(Laughter)
175
520150
1014
（笑聲）
08:53
It's just kind of like,
176
521188
1152
那種感覺就是，
08:54
"Yes! The formula worked. It's all OK.
Everything is under control."
Everything is under control."
177
522364
3504
「好的！方程式沒問題。
一切都很好。都在掌控之中。」
一切都很好。都在掌控之中。」
08:58
But the problem is
178
526792
2390
問題是，
09:01
that quantifying is addictive.
179
529206
2661
量化會使人上癮。
09:03
And when we forget that
180
531891
1382
我們一旦忘記這件事，
09:05
and when we don't have something
to kind of keep that in check,
to kind of keep that in check,
181
533297
3038
若我們沒能做到時時確認是否上癮，
09:08
it's very easy to just throw out data
182
536359
2118
我們很容易直接扔掉這樣的資料：
09:10
because it can't be expressed
as a numerical value.
as a numerical value.
183
538501
2718
僅僅因為它無法用數值量化。
09:13
It's very easy just to slip
into silver-bullet thinking,
into silver-bullet thinking,
184
541243
2921
很容易認為會有完美解決一切的絶招，
09:16
as if some simple solution existed.
185
544188
2579
就好像有某種簡單的解決方法一樣。
09:19
Because this is a great moment of danger
for any organization,
for any organization,
186
547600
4062
因為這對任何一間機構來說，
都是危機的重要時刻，
都是危機的重要時刻，
09:23
because oftentimes,
the future we need to predict --
the future we need to predict --
187
551686
2634
時常，我們要預測的未來，
09:26
it isn't in that haystack,
188
554344
2166
並不是在這安穩的草堆裡，
09:28
but it's that tornado
that's bearing down on us
that's bearing down on us
189
556534
2538
而是在它之外，
是即將襲擊我們的暴風中心。
是即將襲擊我們的暴風中心。
09:31
outside of the barn.
190
559096
1488
09:34
There is no greater risk
191
562960
2326
沒有什麼比對未知
一無所知來得有風險，
一無所知來得有風險，
09:37
than being blind to the unknown.
192
565310
1666
09:39
It can cause you to make
the wrong decisions.
the wrong decisions.
193
567000
2149
那會使你做出錯誤的決定。
09:41
It can cause you to miss something big.
194
569173
1974
那可能使你錯失重要的事物。
09:43
But we don't have to go down this path.
195
571734
3101
但我們不用這樣做。
09:47
It turns out that the oracle
of ancient Greece
of ancient Greece
196
575453
3195
到頭來，是古希臘的先知
握有顯示道路的神秘鑰匙。
握有顯示道路的神秘鑰匙。
09:50
holds the secret key
that shows us the path forward.
that shows us the path forward.
197
578672
3966
09:55
Now, recent geological research has shown
198
583654
2595
近年的地質研究顯示，
09:58
that the Temple of Apollo,
where the most famous oracle sat,
where the most famous oracle sat,
199
586273
3564
最有名的先知所在的阿波羅神廟，
10:01
was actually built
over two earthquake faults.
over two earthquake faults.
200
589861
3084
事實上座落在兩個地震斷層上。
10:04
And these faults would release
these petrochemical fumes
these petrochemical fumes
201
592969
2886
這些斷層會從地殼下釋出石油煙氣，
10:07
from underneath the Earth's crust,
202
595879
1685
而那位先知就直接坐在那些斷層上方，
10:09
and the oracle literally sat
right above these faults,
right above these faults,
203
597588
3866
10:13
inhaling enormous amounts
of ethylene gas, these fissures.
of ethylene gas, these fissures.
204
601478
3588
從縫隙中吸入數不盡的乙烯氣體。
10:17
(Laughter)
205
605090
1008
（笑聲）
10:18
It's true.
206
606122
1173
那是真的。
10:19
(Laughter)
207
607319
1017
（笑聲）
10:20
It's all true, and that's what made her
babble and hallucinate
babble and hallucinate
208
608360
3509
那都是真的，那就是為什麼
她講話含糊不清還看到幻覺，
她講話含糊不清還看到幻覺，
10:23
and go into this trance-like state.
209
611893
1724
並進入類似出神的狀態。
10:25
She was high as a kite!
210
613641
1770
她感覺自己都飛上天了！
10:27
(Laughter)
211
615435
4461
（笑聲）
10:31
So how did anyone --
212
619920
2779
所以大家要怎麼──
10:34
How did anyone get
any useful advice out of her
any useful advice out of her
213
622723
3030
大家要怎麼在這個狀態下
得到有用的建議？
得到有用的建議？
10:37
in this state?
214
625777
1190
10:39
Well, you see those people
surrounding the oracle?
surrounding the oracle?
215
627497
2381
看到那些圍繞先知的人們了嗎？
10:41
You see those people holding her up,
216
629902
1879
你可以看到那些人支撐著她，
10:43
because she's, like, a little woozy?
217
631805
1717
因為她好像有點頭昏眼花？
10:45
And you see that guy
on your left-hand side
on your left-hand side
218
633546
2308
有沒有發現她左邊的男子
10:47
holding the orange notebook?
219
635878
1598
正拿著橘色小冊子？
10:50
Well, those were the temple guides,
220
638105
1730
那些是神廟的引導人員，
10:51
and they worked hand in hand
with the oracle.
with the oracle.
221
639859
3016
他們與先知密切合作。
10:56
When inquisitors would come
and get on their knees,
and get on their knees,
222
644084
2516
當有人來下跪詢問時，
10:58
that's when the temple guides
would get to work,
would get to work,
223
646624
2340
神廟的引導人員就開始工作了，
11:00
because after they asked her questions,
224
648988
1864
在來者向先知詢問一些問題後，
11:02
they would observe their emotional state,
225
650876
2001
他們會觀察來者的精神狀態，
11:04
and then they would ask them
follow-up questions,
follow-up questions,
226
652901
2324
然後他們會問來者一些後續問題，
11:07
like, "Why do you want to know
this prophecy? Who are you?
this prophecy? Who are you?
227
655249
2834
像是：「為什麼你想知道
這個預言？你是誰？
這個預言？你是誰？
11:10
What are you going to do
with this information?"
with this information?"
228
658107
2264
你會怎麼運用這個資訊？」
11:12
And then the temple guides would take
this more ethnographic,
this more ethnographic,
229
660395
3182
接著神廟的引導人員會
用人類學的角度來看，
用人類學的角度來看，
11:15
this more qualitative information,
230
663601
2156
用質性資訊的角度來看，
11:17
and interpret the oracle's babblings.
231
665781
2075
然後翻譯先知含糊不清的話。
11:21
So the oracle didn't stand alone,
232
669428
2292
所以先知並非自己承攬一切任務，
11:23
and neither should our big data systems.
233
671744
2148
我們的大數據系統同樣也不該如此。
11:26
Now to be clear,
234
674630
1161
我要澄清一下，
11:27
I'm not saying that big data systems
are huffing ethylene gas,
are huffing ethylene gas,
235
675815
3459
我並非在說大數據系統
在呼吸着乙烯氣體，
在呼吸着乙烯氣體，
11:31
or that they're even giving
invalid predictions.
invalid predictions.
236
679298
2353
甚至給予沒用的預測。
11:33
The total opposite.
237
681675
1161
完全相反。
11:34
But what I am saying
238
682860
2068
我想說的是，
11:36
is that in the same way
that the oracle needed her temple guides,
that the oracle needed her temple guides,
239
684952
3832
就像先知需要神廟的引導人員那樣，
11:40
our big data systems need them, too.
240
688808
2288
大數據系統同樣也需要。
11:43
They need people like ethnographers
and user researchers
and user researchers
241
691120
4109
大數據需要人類學家以及用戶研究人員
11:47
who can gather what I call thick data.
242
695253
2506
來收集我所謂的「厚數據」──
11:50
This is precious data from humans,
243
698502
2991
來自於人們的寶貴數據，
11:53
like stories, emotions and interactions
that cannot be quantified.
that cannot be quantified.
244
701517
4102
像是故事、情緒和互動，
這些無法計量的事物。
這些無法計量的事物。
11:57
It's the kind of data
that I collected for Nokia
that I collected for Nokia
245
705643
2322
就像我收集給諾基亞的那種數據，
數據樣本規模非常小，
11:59
that comes in in the form
of a very small sample size,
of a very small sample size,
246
707989
2669
12:02
but delivers incredible depth of meaning.
247
710682
2955
但傳達的涵義卻極其的深。
12:05
And what makes it so thick and meaty
248
713661
3680
它如此厚重、內容豐富的原因是
12:10
is the experience of understanding
the human narrative.
the human narrative.
249
718445
4029
那些從人們的話語中
明白更多信息的經驗。
明白更多信息的經驗。
12:14
And that's what helps to see
what's missing in our models.
what's missing in our models.
250
722498
3639
這才能幫助我們看到
模型裡缺少了什麼東西。
模型裡缺少了什麼東西。
12:18
Thick data grounds our business questions
in human questions,
in human questions,
251
726851
4045
厚數據以人類問題為根基
來說明經濟問題，
來說明經濟問題，
12:22
and that's why integrating
big and thick data
big and thick data
252
730920
3562
這就是為什麼結合大數據和厚數據
12:26
forms a more complete picture.
253
734506
1689
能讓我們得到的訊息更加完整。
12:28
Big data is able to offer
insights at scale
insights at scale
254
736772
2881
大數據能在一定程度上洞悉問題，
12:31
and leverage the best
of machine intelligence,
of machine intelligence,
255
739677
2647
並最大程度發揮機器智能，
12:34
whereas thick data can help us
rescue the context loss
rescue the context loss
256
742348
3572
而厚數據能幫我們找到
那缺失的背景資訊，
那缺失的背景資訊，
能讓大數據便於使用，
12:37
that comes from making big data usable,
257
745944
2098
12:40
and leverage the best
of human intelligence.
of human intelligence.
258
748066
2181
並最大程度發揮人類智能。
12:42
And when you actually integrate the two,
that's when things get really fun,
that's when things get really fun,
259
750271
3552
若你真的把這兩個結合在一起
事情就會變得非常有趣，
事情就會變得非常有趣，
12:45
because then you're no longer
just working with data
just working with data
260
753847
2436
如此一來，運用的就不只是
你早就收集的數據。
你早就收集的數據。
12:48
you've already collected.
261
756307
1196
你還可以運用尚未收集的數據。
12:49
You get to also work with data
that hasn't been collected.
that hasn't been collected.
262
757527
2737
你就可以知道「為什麼」：
12:52
You get to ask questions about why:
263
760288
1719
12:54
Why is this happening?
264
762031
1317
為什麼會變成這樣？
12:55
Now, when Netflix did this,
265
763778
1379
所以說，網飛這樣做
12:57
they unlocked a whole new way
to transform their business.
to transform their business.
266
765181
3035
就開啟了轉換商業模式的全新方式。
13:01
Netflix is known for their really great
recommendation algorithm,
recommendation algorithm,
267
769406
3956
網飛以擁有優秀的推薦演算法而聞名，
13:05
and they had this $1 million prize
for anyone who could improve it.
for anyone who could improve it.
268
773386
4797
且發給任何能改善系統的人
一百萬美元獎金。
一百萬美元獎金。
13:10
And there were winners.
269
778207
1314
有人贏了獎金。
13:12
But Netflix discovered
the improvements were only incremental.
the improvements were only incremental.
270
780255
4323
但網飛發現效能提升還是不夠明顯。
13:17
So to really find out what was going on,
271
785404
1964
為了知道發生了什麼事，
13:19
they hired an ethnographer,
Grant McCracken,
Grant McCracken,
272
787392
3741
他們僱用了人類學家，
格蘭特．麥克拉肯，
格蘭特．麥克拉肯，
13:23
to gather thick data insights.
273
791157
1546
來收集厚數據以準確洞察理解。
13:24
And what he discovered was something
that they hadn't seen initially
that they hadn't seen initially
274
792727
3924
他發現了網飛最初未能
從量化數據中看出來的，
從量化數據中看出來的，
13:28
in the quantitative data.
275
796675
1355
13:31
He discovered that people loved
to binge-watch.
to binge-watch.
276
799072
2728
他發現人們喜歡刷劇。
（註：短時間內狂看電視劇）
（註：短時間內狂看電視劇）
13:33
In fact, people didn't even
feel guilty about it.
feel guilty about it.
277
801824
2353
事實上，人們甚至不覺得有什麼不對。
13:36
They enjoyed it.
278
804201
1255
他們非常享受這個過程。
13:37
(Laughter)
279
805480
1026
（笑聲）
13:38
So Netflix was like,
"Oh. This is a new insight."
"Oh. This is a new insight."
280
806530
2356
網飛覺得：「噢，這是個新洞見。」
13:40
So they went to their data science team,
281
808910
1938
於是叫他們的數據科學組
13:42
and they were able to scale
this big data insight
this big data insight
282
810872
2318
把這洞察放大到
量化數據的規模來衡量。
量化數據的規模來衡量。
13:45
in with their quantitative data.
283
813214
2587
13:47
And once they verified it
and validated it,
and validated it,
284
815825
3170
一旦他們再次確認了它的準確性，
13:51
Netflix decided to do something
very simple but impactful.
very simple but impactful.
285
819019
4761
網飛便決定做一件簡單
卻影響很大的事情。
卻影響很大的事情。
13:56
They said, instead of offering
the same show from different genres
the same show from different genres
286
824834
6492
他們說：
「與其提供不同類型但相似的影集，
14:03
or more of the different shows
from similar users,
from similar users,
287
831350
3888
或是給類似的觀眾
欣賞更多不同的影集，
欣賞更多不同的影集，
14:07
we'll just offer more of the same show.
288
835262
2554
只要同一影集提供更多集就好了。
14:09
We'll make it easier
for you to binge-watch.
for you to binge-watch.
289
837840
2105
我們讓你更容易刷劇。」
14:11
And they didn't stop there.
290
839969
1486
而他們並沒有止步於此。
14:13
They did all these things
291
841479
1474
他們用一樣的方式，
14:14
to redesign their entire
viewer experience,
viewer experience,
292
842977
2959
重新設計了整個觀眾體驗，
14:17
to really encourage binge-watching.
293
845960
1758
來真正地鼓勵大家刷劇。
14:20
It's why people and friends disappear
for whole weekends at a time,
for whole weekends at a time,
294
848230
3241
這就是為什麼朋友會消失整個星期，
14:23
catching up on shows
like "Master of None."
like "Master of None."
295
851495
2343
追上「無為大師」等戲劇的進度。
14:25
By integrating big data and thick data,
they not only improved their business,
they not only improved their business,
296
853862
4173
結合大數據與厚數據，
不只讓產業進步，
14:30
but they transformed how we consume media.
297
858059
2812
也轉變了我們使用媒體的型態。
14:32
And now their stocks are projected
to double in the next few years.
to double in the next few years.
298
860895
4552
預期他們的股票
會在接下來幾年內翻倍。
會在接下來幾年內翻倍。
14:38
But this isn't just about
watching more videos
watching more videos
299
866280
3830
這不只是關於看了更多影片，
14:42
or selling more smartphones.
300
870134
1620
或賣了更多智慧型手機，等等。
14:44
For some, integrating thick data
insights into the algorithm
insights into the algorithm
301
872143
4050
對於一些公司來說，
結合厚數據洞察和演算法，
14:48
could mean life or death,
302
876217
2263
可能讓他們起死回生，
14:50
especially for the marginalized.
303
878504
2146
特別是那些已被邊緣化的公司。
14:53
All around the country,
police departments are using big data
police departments are using big data
304
881738
3434
全國的警察局都用大數據來防止犯罪，
14:57
for predictive policing,
305
885196
1963
來設定保證金金額，
14:59
to set bond amounts
and sentencing recommendations
and sentencing recommendations
306
887183
3084
並用加劇偏見的方式來建議判刑。
15:02
in ways that reinforce existing biases.
307
890291
3147
15:06
NSA's Skynet machine learning algorithm
308
894296
2423
美國國家安全局的天網學習演算法
15:08
has possibly aided in the deaths
of thousands of civilians in Pakistan
of thousands of civilians in Pakistan
309
896743
5444
可能致使幾千名巴基斯坦平民死亡，
15:14
from misreading cellular device metadata.
310
902211
2721
肇因於錯誤判讀了行動電話的數據。
15:19
As all of our lives become more automated,
311
907131
3403
當我們的生活變得更加自動化，
15:22
from automobiles to health insurance
or to employment,
or to employment,
312
910558
3080
從汽車、健康保險或者就業，
15:25
it is likely that all of us
313
913662
2350
很可能我們所有人
15:28
will be impacted
by the quantification bias.
by the quantification bias.
314
916036
2989
都會受量化偏見的影響。
15:32
Now, the good news
is that we've come a long way
is that we've come a long way
315
920972
2621
好消息是
我們從吸入乙烯氣體到做出預測
15:35
from huffing ethylene gas
to make predictions.
to make predictions.
316
923617
2450
已有長足的進步。
15:38
We have better tools,
so let's just use them better.
so let's just use them better.
317
926091
3070
我們有了更好的工具，
那麽讓我們更好地利用它。
那麽讓我們更好地利用它。
15:41
Let's integrate the big data
with the thick data.
with the thick data.
318
929185
2323
讓我們將大數據與厚數據結合。
15:43
Let's bring our temple guides
with the oracles,
with the oracles,
319
931532
2261
讓我們使神廟的引導人員
與先知一起合作，
與先知一起合作，
15:45
and whether this work happens
in companies or nonprofits
in companies or nonprofits
320
933817
3376
不論做這項工作的是
公司、非營利組織、
政府，甚至軟體，
15:49
or government or even in the software,
321
937217
2469
全部都有其意義，
15:51
all of it matters,
322
939710
1792
15:53
because that means
we're collectively committed
we're collectively committed
323
941526
3023
因為這代表我們全體一起努力
15:56
to making better data,
324
944573
2191
來得到更好的數據，
15:58
better algorithms, better outputs
325
946788
1836
更好的演算法、更好的產品，
16:00
and better decisions.
326
948648
1643
以及更好的決定。
16:02
This is how we'll avoid
missing that something.
missing that something.
327
950315
3558
這就是避免錯失的方法。
16:07
(Applause)
328
955222
3948
（掌聲）
ABOUT THE SPEAKER
Tricia Wang - Technology ethnographerWith astronaut eyes and ethnographer curiosity, Tricia Wang helps corporations grow by discovering the unknown about their customers.
Why you should listen
For Tricia Wang, human behavior generates some of the most perplexing questions of our times. She has taught global organizations how to identify new customers and markets hidden behind their data, amplified IDEO's design thinking practice as an expert-in-residence, researched the social evolution of the Chinese internet, and written about the "elastic self," an emergent form of interaction in a virtual world. Wang is the co-founder of Sudden Compass, a consulting firm that helps companies unlock new growth opportunities by putting customer obsession into practice.
Wang's work has been featured in The Atlantic, Al Jazeera, and The Guardian. Fast Company spotlighted her work in China: "What Twitter Can Learn From Weibo: Field Notes From Global Tech Ethnographer Tricia Wang." In her latest op-ed on Slate, she discusses how attempts to stop terrorists on social media can harm our privacy and anonymity. Her Medium post, "Why Big Data Needs Thick Data," is a frequently cited industry piece on the importance of an integrated data approach. One of her favorite essays documents her day in the life of working as a street vendor in China.
Known for her lively presentations that are grounded in her research and observations about human behavior and data, Wang has spoken at organizations such as Proctor & Gamble, Nike, Wrigley, 21st Century Fox and Tumblr. Her most recent talk at Enterprise UX delved into why corporate innovation usually doesn’t work and what to do about it. She delivered the opening keynote at The Conference to a crowd of marketers and creatives, delving into the wild history of linear perspective and its influence on how we think and form organizations.
Wang holds affiliate positions at Data & Society, Harvard University's Berkman Klein Center for Internet Studies and New York University's Interactive Telecommunication Program. She oversees Ethnography Matters, a site that publishes articles about applied ethnography and technology. She co-started a Slack community for anyone who uses ethnographic methods in industry.
Wang began her career as a documentary filmmaker at NASA, an HIV/AIDS activist, and an educator specializing in culturally responsive pedagogy. She is also proud to have co-founded the first national hip-hop education initiative, which turned into the Hip Hop Education Center at New York University, and to have built after-school technology and arts programs for low-income youth at New York City public schools and the Queens Museum of Arts. Her life philosophy is that you have to go to the edge to discover what’s really happening. She's the proud companion of her internet famous dog, #ellethedog.
Tricia Wang | Speaker | TED.com