TED2009
Nate Silver: Does racism affect how you vote?
ネイト・シルバー：人種は投票に影響を及ぼすか？
Filmed:
Readability: 4
498,847 views
ネイト・シルバーが政治における人種についての物議をかもす疑問に答えます。果たしてオバマの人種は一部の地域で彼の得票数を削ぐことになったのか？ 統計と俗説のぶつかり合うこの魅惑的な講演は、街づくりがいかに寛容さを促進させうるかという驚くべき洞察で締めくくられます。
Nate Silver - Statistician
Math whiz and baseball fan Nate Silver was mainly known for predicting outcomes in fantasy ballgames -- until his technique hit a home run calling the outcome of the 2008 election primaries. Full bio
Math whiz and baseball fan Nate Silver was mainly known for predicting outcomes in fantasy ballgames -- until his technique hit a home run calling the outcome of the 2008 election primaries. Full bio
Double-click the English transcript below to play the video.
00:15
I want to talk about the election.
0
0
3000
選挙について話そうと思います
00:18
For the first time in the United States, a predominantly white group of voters
1
3000
3000
多数の白人有権者が アフリカ系アメリカ人の
00:21
voted for an African-American candidate for President.
2
6000
3000
大統領候補に投票したというのは 初めてのことでした
00:24
And in fact Barack Obama did quite well.
3
9000
2000
実際オバマはかなりよくやりました
00:26
He won 375 electoral votes.
4
11000
2000
選挙人投票で３７５票を獲得し
00:28
And he won about 70 million popular votes
5
13000
3000
そして一般票では約７０００万票という
00:31
more than any other presidential candidate --
6
16000
2000
人種や政党にかかわらず
00:33
of any race, of any party -- in history.
7
18000
3000
史上最多の得票をしたのです
00:36
If you compare how Obama did against how John Kerry had done four years earlier --
8
21000
4000
オバマの状況を４年前のジョン・ケリーと比べるなら
00:40
Democrats really like seeing this transition here,
9
25000
3000
この推移に見てとれるように
00:43
where almost every state becomes bluer, becomes more democratic --
10
28000
4000
ほとんどの州がより青く より民主党寄りになっています
00:47
even states Obama lost, like out west,
11
32000
2000
西部のようなオバマが落とした州でさえそうです
00:49
those states became more blue.
12
34000
2000
南部でも北東部でもほとんど全ての州で
00:51
In the south, in the northeast, almost everywhere
13
36000
3000
青くなっています
00:54
but with a couple of exceptions here and there.
14
39000
3000
そこここにある いくつかの例外を除いて
00:57
One exception is in Massachusetts.
15
42000
2000
例外の一つはマサチューセッツ州ですが
00:59
That was John Kerry's home state.
16
44000
2000
ジョン・ケリーの本拠地なので
01:01
No big surprise, Obama couldn't do better than Kerry there.
17
46000
2000
オバマが下でも驚くことはありません
01:03
Or in Arizona, which is John McCain's home,
18
48000
2000
それにマケインの本拠のアリゾナ州でも
01:05
Obama didn't have much improvement.
19
50000
2000
オバマは票を増やせませんでした
01:07
But there is also this part of the country, kind of in the middle region here.
20
52000
2000
他方で この国の中央に位置する
01:09
This kind of Arkansas, Tennessee, Oklahoma, West Virginia region.
21
54000
4000
アーカンソー テネシー オクラホマ ウェストヴァージニアの辺りも駄目でした
01:13
Now if you look at '96, Bill Clinton --
22
58000
2000
９６年のビル・クリントンは
01:15
the last Democrat to actually win -- how he did in '96,
23
60000
3000
民主党が前回勝利した時ですが
01:18
you see real big differences in this part of the country right here,
24
63000
3000
まさにこの地域で 実に大きな違いが見られます
01:21
the kind of Appalachians, Ozarks, highlands region, as I call it:
25
66000
4000
アパラチアやオザークといった 高地地域です
01:25
20 or 30 point swings
26
70000
2000
９６年のクリントンと ０８年のオバマの間で
01:27
from how Bill Clinton did in '96 to how Obama did
27
72000
2000
２０から３０ポイントも
01:29
in 2008.
28
74000
2000
下がりました
01:31
Yes Bill Clinton was from Arkansas, but these are very, very profound differences.
29
76000
5000
確かにクリントンはアーカンソー出身でしたが それにしても大きな変化です
01:36
So, when we think about parts of the country like Arkansas, you know.
30
81000
2000
アーカンソーのような地域について考えるとき
01:38
There is a book written called, "What's the Matter with Kansas?"
31
83000
3000
「カンザスはどうなっているのか？」という本がありますが
01:41
But really the question here -- Obama did relatively well in Kansas.
32
86000
3000
オバマはカンザスでは割合上手くやっています
01:44
He lost badly but every Democrat does.
33
89000
2000
負けましたが 民主党はここではいつも負けています
01:46
He lost no worse than most people do.
34
91000
2000
他の人ほど ひどくは負けませんでした
01:48
But yeah, what's the matter with Arkansas?
35
93000
4000
それにしてもアーカンソーはどうなっているのか？
01:52
(Laughter)
36
97000
1000
（笑）
01:53
And when we think of Arkansas we tend to have pretty negative connotations.
37
98000
3000
アーカンソーについてはネガティブなイメージがあります
01:56
We think of a bunch of rednecks, quote, unquote, with guns.
38
101000
3000
銃を持った多数のいわゆる「頑固な南部人」
01:59
And we think people like this probably don't want to vote
39
104000
3000
彼らが投票しないだろうとは想像できます
02:02
for people who look like this and are named Barack Obama.
40
107000
3000
こんな見た目をした バラク・オバマなどという名前の人間には
02:05
We think it's a matter of race. And is this fair?
41
110000
3000
人種問題です これはフェアな見方でしょうか？
02:08
Are we kind of stigmatizing people from Arkansas, and this part of the country?
42
113000
3000
アーカンソーあたりの人に対する偏見でしょうか？
02:11
And the answer is: it is at least partially fair.
43
116000
3000
少なくともある部分フェアな見方です
02:14
We know that race was a factor, and the reason why we know that
44
119000
2000
人種という要因が確かにあるとわかるのは
02:16
is because we asked those people.
45
121000
2000
彼らに尋ねたからです
02:18
Actually we didn't ask them, but when they conducted
46
123000
2000
直接尋ねた訳ではありませんが
02:20
exit polls in every state,
47
125000
2000
５０州中の３７州で行われた
02:22
in 37 states, out of the 50,
48
127000
2000
出口調査で
02:24
they asked a question, that was pretty direct, about race.
49
129000
3000
人種について直接的な質問がありました
02:27
They asked this question.
50
132000
2000
こう質問したのです
02:29
In deciding your vote for President today, was the race
51
134000
2000
“今日の大統領選の投票先を決めるにあたって
02:31
of the candidate a factor?
52
136000
2000
候補者の人種はその要因になりましたか？”
02:33
We're looking for people that said, "Yes, race was a factor;
53
138000
3000
この質問に対し “はい 人種は要因であるのみならず
02:36
moreover it was an important factor, in my decision,"
54
141000
2000
決める上で重要な点でした”と答えた人や
02:38
and people who voted for John McCain
55
143000
3000
恐らく他の要因との兼ね合いか
02:41
as a result of that factor,
56
146000
2000
この要因自体のために 結果として
02:43
maybe in combination with other factors, and maybe alone.
57
148000
2000
マケインに投票した人は どれほどいたのか？
02:45
We're looking for this behavior among white voters
58
150000
3000
そのような振る舞いは 白人投票者の間で
02:48
or, really, non-black voters.
59
153000
3000
あるいは黒人以外の投票者の間で どれくらいあったのか？
02:51
So you see big differences in different parts
60
156000
2000
ご覧のように 結果は地域によって
02:53
of the country on this question.
61
158000
2000
大きな違いが見られます
02:55
In Louisiana, about one in five white voters
62
160000
3000
ルイジアナでは約５人に１人の白人投票者が
02:58
said, "Yes, one of the big reasons why I voted against Barack Obama
63
163000
3000
“はい 私がオバマに投票しなかった大きな理由は
03:01
is because he was an African-American."
64
166000
2000
アフリカ系アメリカ人だからです”と答えています
03:03
If those people had voted for Obama,
65
168000
2000
仮にそれらの人々の半分でも
03:05
even half of them, Obama would have won Louisiana safely.
66
170000
4000
オバマに投票していたなら 彼はルイジアナで難なく勝てたでしょう
03:09
Same is true with, I think, all of these states you see on the top of the list.
67
174000
2000
このリストの上位の州すべてにそう言えます
03:11
Meanwhile, California, New York, we can say, "Oh we're enlightened"
68
176000
4000
一方カリフォルニアやニューヨークの人々は “我々は啓蒙されている”
03:15
but you know, certainly a much lower incidence of this
69
180000
2000
と言うでしょうが
03:17
admitted, I suppose,
70
182000
2000
人種に基づく投票というのは
03:19
manifestation of racially-based voting.
71
184000
3000
確かにずっと少なくなっています
03:22
Here is the same data on a map.
72
187000
2000
これは同じデータを地図で表したものです
03:24
You kind of see the relationship between
73
189000
2000
この図で赤くなっているのは
03:26
the redder states of where more people responded and said,
74
191000
2000
“オバマの人種は 私には問題でした”
03:28
"Yes, Barack Obama's race was a problem for me."
75
193000
3000
と答えた人の多かった州です
03:31
You see, comparing the map to '96, you see an overlap here.
76
196000
3000
９６年と比較した地図と この地図の間には共通点が見られます
03:34
This really seems to explain
77
199000
2000
これはまさにオバマが
03:36
why Barack Obama did worse
78
201000
2000
何故この地域でひどい結果だったのかを
03:38
in this one part of the country.
79
203000
2000
説明しているように見えます
03:40
So we have to ask why.
80
205000
2000
我々は問うべきでしょう
03:42
Is racism predictable in some way?
81
207000
2000
人種差別は何らかの方法で予測できるのか？
03:44
Is there something driving this?
82
209000
2000
人種差別を引き起こすものが何かあるのか？
03:46
Is it just about some weird stuff that goes on in Arkansas
83
211000
2000
それは我々が理解していないアーカンソーや
03:48
that we don't understand, and Kentucky?
84
213000
2000
ケンタッキーの特異性にすぎないのか？
03:50
Or are there more systematic factors at work?
85
215000
2000
それとも何か構造的な要因が働いているのか？
03:52
And so we can look at a bunch of different variables.
86
217000
2000
見る事のできる変数は他にもたくさんあります
03:54
These are things that economists and political scientists look at all the time --
87
219000
3000
経済学者や政治科学者が常々観察してきた
03:57
things like income, and religion, education.
88
222000
3000
収入や宗教や教育といった要因は
04:00
Which of these seem to drive
89
225000
2000
我々が１１月４日に行った
04:02
this manifestation of racism
90
227000
2000
この国家的大実験における
04:04
in this big national experiment we had on November 4th?
91
229000
3000
人種差別の表出を引き起こしたものなのでしょうか？
04:07
And there are a couple of these that have
92
232000
2000
実際予測を可能にするような
04:09
strong predictive relationships,
93
234000
2000
強い相関を持つ要素が2つあります
04:11
one of which is education,
94
236000
3000
教育がその一つです
04:14
where you see the states with the fewest years of schooling
95
239000
2000
成人あたりの修学期間の短い州が
04:16
per adult are in red,
96
241000
2000
赤く表示されています
04:18
and you see this part of the country, the kind of Appalachians region,
97
243000
3000
ご覧の通りアパラチア地域は教育水準が低くなっています
04:21
is less educated. It's just a fact.
98
246000
2000
これは単に事実です
04:23
And you see the relationship there
99
248000
2000
ここに人種差別に基づく投票傾向との
04:25
with the racially-based voting patterns.
100
250000
3000
関連を見る事ができます
04:28
The other variable that's important is
101
253000
2000
もう一つの重要な変数は
04:30
the type of neighborhood that you live in.
102
255000
3000
近隣住民のタイプです
04:33
States that are more rural --
103
258000
2000
より田舎の州では
04:35
even to some extent of the states like New Hampshire and Maine --
104
260000
2000
ニューハンプシャーやメインでさえ
04:37
they exhibit a little bit of
105
262000
2000
人種差別によるオバマへの
04:39
this racially-based voting against Barack Obama.
106
264000
3000
反対投票が見受けられます
04:42
So it's the combination of these two things: it's education
107
267000
2000
ですからこの２つの要素の組み合わせなのです
04:44
and the type of neighbors that you have,
108
269000
2000
教育と近隣住民のタイプです
04:46
which we'll talk about more in a moment.
109
271000
2000
これについてはすぐ後で話しますが
04:48
And the thing about states like Arkansas and Tennessee
110
273000
2000
アーカンソーやテネシーのような州の問題は
04:50
is that they're both very rural,
111
275000
2000
非常に田舎であり
04:52
and they are educationally impoverished.
112
277000
4000
かつ教育的にも貧しいという事です
04:56
So yes, racism is predictable.
113
281000
2000
だから人種差別は予測可能なのです
04:58
These things, among maybe other variables,
114
283000
2000
恐らく他にも要因があるでしょうが
05:00
but these things seem to predict it.
115
285000
2000
これは人種差別を予見しているように見えます
05:02
We're going to drill down a little bit more now,
116
287000
2000
この「総合的社会調査」というのを
05:04
into something called the General Social Survey.
117
289000
2000
もう少し掘り下げてみましょう
05:06
This is conducted by the University of Chicago
118
291000
2000
これはシカゴ大学によって一年おきに
05:08
every other year.
119
293000
2000
行われているもので
05:10
And they ask a series of really interesting questions.
120
295000
2000
彼らは実に興味深い一連の質問をしています
05:12
In 2000 they had particularly interesting questions
121
297000
2000
２０００年には人種差別的意見に関する
05:14
about racial attitudes.
122
299000
2000
特に興味深い質問があります
05:16
One simple question they asked is,
123
301000
2000
一例として挙げると
05:18
"Does anyone of the opposite race live in your neighborhood?"
124
303000
4000
“近隣にあなたと異なるの人種の人は誰か住んでいますか？”
05:22
We can see in different types of communities that the results are quite different.
125
307000
3000
コミュニティのタイプによって結果が大きく異なるのがわかります
05:25
In cites, about 80 percent of people
126
310000
3000
都市部では約８０%の人が
05:28
have someone whom they consider a neighbor of another race,
127
313000
3000
近隣に別の人種が住んでいると答えています
05:31
but in rural communities, only about 30 percent.
128
316000
3000
しかし田舎のコミュニティでは僅か３０%です
05:34
Probably because if you live on a farm, you might not have a lot of neighbors, period.
129
319000
3000
農場暮らしなら隣人は少ないというだけかもしれません
05:37
But nevertheless, you're not having a lot of interaction with people
130
322000
3000
しかし いずれにしても 自分と違った人との交流は
05:40
who are unlike you.
131
325000
2000
多くないのです
05:42
So what we're going to do now is take the white people in the survey
132
327000
3000
そこで この調査から白人を抽出し
05:45
and split them between those who have black neighbors --
133
330000
3000
黒人や別の人種の隣人がいる人と
05:48
or, really, some neighbor of another race --
134
333000
2000
白人の隣人しかいない人とに
05:50
and people who have only white neighbors.
135
335000
3000
分けてみることにしましょう
05:53
And we see in some variables
136
338000
2000
政治的意見に関しては
05:55
in terms of political attitudes, not a lot of difference.
137
340000
2000
大きな違いは見られません
05:57
This was eight years ago, some people were more Republican back then.
138
342000
3000
これは８年前で 当時は共和党寄りの人が多くいました
06:00
But you see Democrats versus Republican,
139
345000
2000
しかし民主党支持か共和党支持かは
06:02
not a big difference based on who your neighbors are.
140
347000
3000
隣人がどうであるかによって さほど変わりません
06:05
And even some questions about race -- for example
141
350000
2000
人種に関する質問でさえそうです 例えば
06:07
affirmative action, which is kind of a political question,
142
352000
2000
マイノリティ優遇措置という
06:09
a policy question about race, if you will --
143
354000
2000
人種に関する政策の問題についても
06:11
not much difference here.
144
356000
2000
ここに大きな違いはありません
06:13
Affirmative action is not very popular frankly, with white voters, period.
145
358000
3000
マイノリティ優遇措置は白人投票者に人気がない点で共通していますが
06:16
But people with black neighbors and people with mono-racial neighborhoods
146
361000
3000
黒人の隣人がいる人といない人との間に
06:19
feel no differently about it really.
147
364000
3000
この問題に対する感覚の違いは全くありません
06:22
But if you probe a bit deeper and get a bit more personal if you will,
148
367000
4000
しかしもう少し深く 個人的なことを聞いてみるとどうでしょう
06:26
"Do you favor a law banning interracial marriage?"
149
371000
2000
“異人種間の結婚は法的に禁止すべきか？”
06:28
There is a big difference.
150
373000
2000
この場合大きな違いが見られます
06:30
People who don't have neighbors of a different race
151
375000
2000
異なる人種の隣人がいない人は
06:32
are about twice as likely
152
377000
2000
いる人と比べ２倍近くが
06:34
to oppose interracial marriage as people who do.
153
379000
3000
異人種間の結婚に否定的です
06:37
Just based on who lives in your immediate neighborhood around you.
154
382000
3000
単に周囲に住む隣人によって変わるのです
06:40
And likewise they asked, not in 2000, but in the same survey in 1996,
155
385000
4000
２０００年版ではなく１９９６年版ですが 同じ調査でこんな質問をしています
06:44
"Would you not vote for a qualified black president?"
156
389000
4000
“優れた大統領候補なら黒人でも投票しますか？”
06:48
You see people without neighbors who are African-American who
157
393000
2000
隣人にアフリカ系アメリカ人がいない人は
06:50
were much more likely to say, "That would give me a problem."
158
395000
3000
“それは私には問題だ”と答える傾向が強くなっています
06:53
So it's really not even about urban versus rural.
159
398000
2000
だからこれは都会か田舎かという話ではなく
06:55
It's about who you live with.
160
400000
2000
誰が周りに暮らしているかということなのです
06:57
Racism is predictable. And it's predicted by
161
402000
2000
人種差別は予測可能であり それは
06:59
interaction or lack thereof with people unlike you, people of other races.
162
404000
4000
異人種との交流の有無によって予見できるのです
07:03
So if you want to address it,
163
408000
2000
だからこの問題に取り組みたいなら
07:05
the goal is to facilitate interaction with people of other races.
164
410000
3000
人種間の交流促進を目指すべきでしょう
07:08
I have a couple of very obvious, I suppose,
165
413000
2000
それを行うための
07:10
ideas for maybe how to do that.
166
415000
3000
ごく自明なアイデアがいくつかあります
07:13
I'm a big fan of cities.
167
418000
2000
私は都会が大好きです
07:15
Especially if we have cites that are diverse and sustainable,
168
420000
3000
特にその都市が持続可能で多様であり
07:18
and can support people of different ethnicities and different income groups.
169
423000
3000
異なる民族性や収入帯の集団を支えられるなら素晴らしいです
07:21
I think cities facilitate more of the kind of networking,
170
426000
3000
都会は田舎よりも日常的に
07:24
the kind of casual interaction than you might have on a daily basis.
171
429000
3000
人の繋がりや気軽な交流を促進すると思います
07:27
But also not everyone wants to live in a city, certainly not a city like New York.
172
432000
3000
しかし誰もがニューヨークのような都会で暮らしたいわけではありません
07:30
So we can think more about things like street grids.
173
435000
3000
そこで考えるべきなのは道路網のようなものについてです
07:33
This is the neighborhood where I grew up in East Lansing, Michigan.
174
438000
2000
これは私が育ったミシガン州イーストランシングの町並みです
07:35
It's a traditional Midwestern community, which means you have real grid.
175
440000
3000
伝統的な中西部の田舎で 道が格子状になっています
07:38
You have real neighborhoods and real trees, and real streets you can walk on.
176
443000
3000
本当のご近所や 木や 歩くことが出来る道路があり
07:41
And you interact a lot with your neighbors --
177
446000
3000
そして近所にいる 好きな人や 知らない人たちと
07:44
people you like, people you might not know.
178
449000
2000
交流することになります
07:46
And as a result it's a very tolerant community,
179
451000
3000
結果としてそれはとても寛容なコミュニティとなり
07:49
which is different, I think, than something like this,
180
454000
2000
次のような所とは異なると私は考えます
07:51
which is in Schaumburg, Illinois,
181
456000
2000
このイリノイ州のシャンバーグでは
07:53
where every little set of houses has their own cul-de-sac
182
458000
3000
袋小路になった住宅群や ドライブスルーのスターバックスが
07:56
and drive-through Starbucks and stuff like that.
183
461000
2000
このように詰め込まれています
07:58
I think that actually this type of urban design,
184
463000
3000
私はこのタイプの都市計画が
08:01
which became more prevalent in the 1970s and 1980s --
185
466000
3000
１９７０年代から８０年代に流行したことと
08:04
I think there is a relationship between that and the country becoming
186
469000
3000
レーガン政権下で国が保守化したことの間には
08:07
more conservative under Ronald Reagan.
187
472000
2000
関連があると考えています
08:09
But also here is another idea we have --
188
474000
3000
他にもアイデアがあります
08:12
is an intercollegiate exchange program
189
477000
2000
ニューヨークと外国とで学生を交換する
08:14
where you have students going from New York abroad.
190
479000
3000
交換留学制度がありますが
08:17
But frankly there are enough differences within the country now
191
482000
2000
率直に言って米国内だけでも十分な差異があります
08:19
where maybe you can take a bunch of kids from NYU,
192
484000
3000
たとえばニューヨーク大学の学生を
08:22
have them go study for a semester at the University of Arkansas,
193
487000
2000
一学期間アーカンソー大学で学ばせ
08:24
and vice versa. Do it at the high school level.
194
489000
3000
逆も同様にするとか それを高校の段階でやるのはどうでしょう
08:27
Literally there are people who might be in school in Arkansas or Tennessee
195
492000
3000
実際アーカンソーやテネシーの学校にいる人は
08:30
and might never interact in a positive affirmative way
196
495000
3000
恐らく別の州や別の人種の人たちと
08:33
with someone from another part of the country, or of another racial group.
197
498000
4000
積極的に交流する機会を持たないでしょう
08:37
I think part of the education variable we talked about before
198
502000
3000
先程話した教育という要因の一部は
08:40
is the networking experience you get when you go to college
199
505000
2000
大学に行くことで得られる
08:42
where you do get a mix of people that you might not interact with otherwise.
200
507000
4000
他では交流する事がなかった人達と混じり合う経験のためだと思います
08:46
But the point is, this is all good news,
201
511000
2000
重要なのはこれが全て良いニュースだという事です
08:48
because when something is predictable,
202
513000
3000
なぜなら 何かが予測可能であるなら
08:51
it is what I call designable.
203
516000
2000
それはデザイン可能でもあるからです
08:53
You can start thinking about solutions to solving that problem,
204
518000
2000
問題の解決策を考える事ができます
08:55
even if the problem is pernicious and as intractable as racism.
205
520000
3000
たとえその問題が人種差別のように悪質で扱い難いものであっても
08:58
If we understand the root causes of the behavior
206
523000
2000
その行動の根本原因や
09:00
and where it manifests itself and where it doesn't,
207
525000
2000
どういう時に起きるかがわかっているなら
09:02
we can start to design solutions to it.
208
527000
3000
その解決策をデザインする事ができるのです
09:05
So that's all I have to say. Thank you very much.
209
530000
2000
これが私の言うべき全てです ありがとうございました
09:07
(Applause)
210
532000
1000
（拍手）
ABOUT THE SPEAKER
Nate Silver - StatisticianMath whiz and baseball fan Nate Silver was mainly known for predicting outcomes in fantasy ballgames -- until his technique hit a home run calling the outcome of the 2008 election primaries.
Why you should listen
In the 2008 election season's closing weeks, throngs of wonks and laypeople alike were glued to FiveThirtyEight.com, a habitforming political blog. Red and blue bar charts crowded the scrollbars as the pulse of exit polls crept along the site's latest projections. It seemed almost miraculous: In a year of acute turns of favor, the site's owner and mouthpiece, Nate Silver (who blogged anonymously as "Poblano" until outing himself on May 30, 2008, as a baseball numberhead), managed to predict the winners of every U.S. Senate contest -- and the general Presidential election.
Besides being just-damn-fascinating, Silver's analysis is a decidedly contrarian gauntlet thrown before an unrepentant, spectacle-driven media. The up-and-coming pundit, who cut his teeth forecasting the performance of Major League Baseball players, has a fairly direct explanation of why most projections fail: "Polls are cherry-picked based on their brand name or shock value rather than their track record of accuracy."
Silver's considerable smarts are already helping local campaigns build constituencies and strategize. He is the author of The Signal and the Noise: Why So Many Predictions Fail - but Some Don't.
Nate Silver | Speaker | TED.com