1
00:00:00,542 --> 00:00:02,726
Who doesn't love efficiency?

2
00:00:02,750 --> 00:00:04,018
I do.

3
00:00:04,042 --> 00:00:07,393
Efficiency means more for less.

4
00:00:07,417 --> 00:00:11,309
More miles per gallon,
more light per watt,

5
00:00:11,333 --> 00:00:12,792
more words per minute.

6
00:00:13,917 --> 00:00:16,393
More for less is the next best thing

7
00:00:16,417 --> 00:00:17,958
to something for nothing.

8
00:00:19,000 --> 00:00:24,059
Algorithms, big data, the cloud
are giving us more for less.

9
00:00:24,083 --> 00:00:27,726
Are we heading toward
a friction-free utopia

10
00:00:27,750 --> 00:00:30,559
or toward a nightmare of surveillance?

11
00:00:30,583 --> 00:00:31,851
I don't know.

12
00:00:31,875 --> 00:00:33,375
My interest is in the present.

13
00:00:34,208 --> 00:00:36,309
And I'd like to show you

14
00:00:36,333 --> 00:00:40,309
how the past can help us
understand the present.

15
00:00:40,333 --> 00:00:42,976
There's nothing that summarizes

16
00:00:43,000 --> 00:00:45,768
both the promise
and the danger of efficiency

17
00:00:45,792 --> 00:00:47,768
like the humble potato.

18
00:00:47,792 --> 00:00:50,309
The potato originated in the Andes

19
00:00:50,333 --> 00:00:55,476
and it spread to Europe
from the ancient Inca.

20
00:00:55,500 --> 00:01:00,250
The potato is a masterpiece
of balanced nutrition.

21
00:01:01,458 --> 00:01:03,917
And it had some very powerful friends.

22
00:01:04,792 --> 00:01:07,476
King Frederick the Great of Prussia

23
00:01:07,500 --> 00:01:10,643
was the first enthusiast.

24
00:01:10,667 --> 00:01:13,018
He believed that the potato could help

25
00:01:13,042 --> 00:01:16,143
increase the population
of healthy Prussians.

26
00:01:16,167 --> 00:01:19,434
And the more healthy Prussians,

27
00:01:19,458 --> 00:01:22,184
the more healthy Prussian soldiers.

28
00:01:22,208 --> 00:01:24,518
And some of those
healthy Prussian soldiers

29
00:01:24,542 --> 00:01:28,333
captured a French military
pharmacist named Parmentier.

30
00:01:29,292 --> 00:01:31,809
Parmentier, at first, was appalled

31
00:01:31,833 --> 00:01:35,601
by the morning, noon and night diet

32
00:01:35,625 --> 00:01:39,184
fed to POWs of potatoes,

33
00:01:39,208 --> 00:01:41,184
but he came to enjoy it.

34
00:01:41,208 --> 00:01:44,518
He thought they were making
him a healthier person.

35
00:01:44,542 --> 00:01:46,309
And so, when he was released,

36
00:01:46,333 --> 00:01:51,309
he took it on himself
to spread the potato to France.

37
00:01:51,333 --> 00:01:53,809
And he had some powerful friends.

38
00:01:53,833 --> 00:01:58,643
Benjamin Franklin
advised him to hold a banquet,

39
00:01:58,667 --> 00:02:01,917
at which every dish included potatoes.

40
00:02:03,083 --> 00:02:05,500
And Franklin was a guest of honor.

41
00:02:06,250 --> 00:02:08,143
Even the king and queen of France

42
00:02:08,167 --> 00:02:12,309
were persuaded to wear potatoes,

43
00:02:12,333 --> 00:02:13,976
potato flowers, pardon me.

44
00:02:14,000 --> 00:02:15,268
(Laughter)

45
00:02:15,292 --> 00:02:19,768
The king wore a potato
flower in his lapel,

46
00:02:19,792 --> 00:02:23,768
and the queen wore
a potato flower in her hair.

47
00:02:23,792 --> 00:02:27,643
That was a truly great
public relations idea.

48
00:02:27,667 --> 00:02:29,476
But there was a catch.

49
00:02:29,500 --> 00:02:33,059
The potato was too efficient
for Europe's good.

50
00:02:33,083 --> 00:02:35,393
In Ireland, it seemed a miracle.

51
00:02:35,417 --> 00:02:38,559
Potatoes flourished, the population grew.

52
00:02:38,583 --> 00:02:40,518
But there was a hidden risk.

53
00:02:40,542 --> 00:02:43,851
Ireland's potatoes
were genetically identical.

54
00:02:43,875 --> 00:02:47,643
They were a very efficient breed,
called the Lumper.

55
00:02:47,667 --> 00:02:50,018
And the problem with the Lumper

56
00:02:50,042 --> 00:02:53,893
was that a blight from South America

57
00:02:53,917 --> 00:02:55,601
that affected one potato

58
00:02:55,625 --> 00:02:57,000
would affect them all.

59
00:02:57,958 --> 00:03:01,434
Britain's exploitation
and callousness played a role,

60
00:03:01,458 --> 00:03:04,809
but it was because of this monoculture

61
00:03:04,833 --> 00:03:07,184
that a million people died

62
00:03:07,208 --> 00:03:10,750
and another two million
were forced to emigrate.

63
00:03:11,542 --> 00:03:14,059
A plant that was supposed to end famine

64
00:03:14,083 --> 00:03:16,625
created one of the most tragic ones.

65
00:03:17,750 --> 00:03:19,643
The problems of efficiency today

66
00:03:19,667 --> 00:03:22,184
are less drastic but more chronic.

67
00:03:22,208 --> 00:03:24,184
They can also prolong the evils

68
00:03:24,208 --> 00:03:25,976
that they were intended to solve.

69
00:03:26,000 --> 00:03:28,268
Take the electronic medical records.

70
00:03:28,292 --> 00:03:32,351
It seemed to be the answer
to the problem of doctors' handwriting,

71
00:03:32,375 --> 00:03:33,934
and it had the benefit

72
00:03:33,958 --> 00:03:37,268
of providing much better data
for treatments.

73
00:03:37,292 --> 00:03:39,393
In practice, instead, it has meant

74
00:03:39,417 --> 00:03:41,226
much more electronic paperwork

75
00:03:41,250 --> 00:03:44,518
and physicians are now complaining
that they have less,

76
00:03:44,542 --> 00:03:48,875
rather than more time
to see patients individually.

77
00:03:49,750 --> 00:03:53,542
The obsession with efficiency
can actually make us less efficient.

78
00:03:54,292 --> 00:03:57,518
Efficiency also bites back
with false positives.

79
00:03:57,542 --> 00:04:01,226
Hospitals have hundreds
of devices registering alarms.

80
00:04:01,250 --> 00:04:03,309
Too often, they're crying wolf.

81
00:04:03,333 --> 00:04:05,268
It takes time to rule those out.

82
00:04:05,292 --> 00:04:10,184
And that time results in fatigue,
stress and, once more,

83
00:04:10,208 --> 00:04:13,792
the neglect of the problems
of real patients.

84
00:04:14,625 --> 00:04:17,976
There are also false positives
in pattern recognition.

85
00:04:18,000 --> 00:04:20,143
A school bus, viewed from the wrong angle,

86
00:04:20,167 --> 00:04:22,292
can resemble a punching bag.

87
00:04:23,083 --> 00:04:25,309
So precious time is required

88
00:04:25,333 --> 00:04:28,268
to eliminate misidentification.

89
00:04:28,292 --> 00:04:30,893
False negatives are a problem, too.

90
00:04:30,917 --> 00:04:33,684
Algorithms can learn a lot -- fast.

91
00:04:33,708 --> 00:04:36,476
But they can tell us only about the past.

92
00:04:36,500 --> 00:04:41,601
So many future classics
get bad reviews, like "Moby Dick,"

93
00:04:41,625 --> 00:04:43,809
or are turned down
by multiple publishers,

94
00:04:43,833 --> 00:04:45,643
like the "Harry Potter" series.

95
00:04:45,667 --> 00:04:50,018
It can be wasteful
to try to avoid all waste.

96
00:04:50,042 --> 00:04:53,726
Efficiency is also a trap
when the opposition copies it.

97
00:04:53,750 --> 00:04:55,893
Take the late 19th-century

98
00:04:55,917 --> 00:04:58,976
French 75-millimeter artillery piece.

99
00:04:59,000 --> 00:05:01,934
It was a masterpiece of lethal design.

100
00:05:01,958 --> 00:05:06,184
This piece could fire a shell
every four seconds.

101
00:05:06,208 --> 00:05:07,559
But that wasn't so unusual.

102
00:05:07,583 --> 00:05:11,726
What was really brilliant
was that because of the recoil mechanism,

103
00:05:11,750 --> 00:05:14,601
it could return to the exact same position

104
00:05:14,625 --> 00:05:16,143
without having to be reaimed.

105
00:05:16,167 --> 00:05:20,018
So the effective rate of firing
was drastically increased.

106
00:05:20,042 --> 00:05:22,559
Now, this seemed to be a way for France

107
00:05:22,583 --> 00:05:26,333
to defeat Germany
the next time they fought.

108
00:05:27,042 --> 00:05:30,393
But, predictably, the Germans were working

109
00:05:30,417 --> 00:05:32,059
on something very similar.

110
00:05:32,083 --> 00:05:34,393
So when the First World War broke out,

111
00:05:34,417 --> 00:05:37,351
the result was the trench warfare

112
00:05:37,375 --> 00:05:40,809
that lasted longer
than anybody had expected.

113
00:05:40,833 --> 00:05:45,042
A technology that was designed
to shorten the war, prolonged it.

114
00:05:46,125 --> 00:05:49,726
The biggest cost of all
may be missed opportunities.

115
00:05:49,750 --> 00:05:53,393
The platform economy
connecting buyers and sellers

116
00:05:53,417 --> 00:05:54,893
can be a great investment,

117
00:05:54,917 --> 00:05:57,851
and we have seen that
in the last few weeks.

118
00:05:57,875 --> 00:06:01,434
Companies that are still losing
hundreds of millions of dollars

119
00:06:01,458 --> 00:06:05,208
may be creating billionaires
with initial public offerings.

120
00:06:06,875 --> 00:06:10,143
But the really difficult inventions

121
00:06:10,167 --> 00:06:12,726
are the physical and chemical ones.

122
00:06:12,750 --> 00:06:14,976
They mean bigger risks.

123
00:06:15,000 --> 00:06:18,309
They may be losing out,
because hardware is hard.

124
00:06:18,333 --> 00:06:22,518
It's much harder to scale up
a physical or chemical invention

125
00:06:22,542 --> 00:06:25,250
than it is a software-based invention.

126
00:06:26,000 --> 00:06:27,476
Think of batteries.

127
00:06:27,500 --> 00:06:31,643
Lithium-ion batteries
in portable devices and electric cars

128
00:06:31,667 --> 00:06:34,333
are based on a 30-year-old principle.

129
00:06:35,042 --> 00:06:37,476
How many smartphone batteries today

130
00:06:37,500 --> 00:06:40,792
will last a full day on a single charge?

131
00:06:41,542 --> 00:06:43,167
Yes, hardware is hard.

132
00:06:44,583 --> 00:06:47,559
It took over 20 years for the patent

133
00:06:47,583 --> 00:06:49,934
on the principle of dry photocopying,

134
00:06:49,958 --> 00:06:53,143
by Chester Carlson in 1938,

135
00:06:53,167 --> 00:07:00,083
to result in the Xerox 914 copier
introduced in 1959.

136
00:07:00,958 --> 00:07:05,934
The small, brave company,
Haloid in Rochester, NY

137
00:07:05,958 --> 00:07:10,893
had to go through what most corporations
would never have tolerated.

138
00:07:10,917 --> 00:07:12,726
There was one failure after another,

139
00:07:12,750 --> 00:07:16,018
and one of the special problems was fire.

140
00:07:16,042 --> 00:07:19,851
In fact, when the 914
was finally released,

141
00:07:19,875 --> 00:07:24,768
it still had a device
that was called a scorch eliminator

142
00:07:24,792 --> 00:07:29,083
but actually it was
a small fire extinguisher built in.

143
00:07:30,167 --> 00:07:34,809
My answer to all these questions is:
inspired inefficiency.

144
00:07:34,833 --> 00:07:38,893
Data and measurement are essential,
but they're not enough.

145
00:07:38,917 --> 00:07:42,851
Let's leave room for human intuition
and human skills.

146
00:07:42,875 --> 00:07:46,684
There are seven facets
of inspired inefficiency.

147
00:07:46,708 --> 00:07:50,476
First, take the scenic route,
say yes to serendipity.

148
00:07:50,500 --> 00:07:53,059
Wrong turns can be productive.

149
00:07:53,083 --> 00:07:56,184
Once, when I was exploring
the east bank of the Mississippi,

150
00:07:56,208 --> 00:07:58,184
I took the wrong turn.

151
00:07:58,208 --> 00:08:01,893
I was approaching a toll bridge
crossing the great river,

152
00:08:01,917 --> 00:08:05,393
and the toll collector
said I could not turn back.

153
00:08:05,417 --> 00:08:08,851
So I paid my 50 cents --
that's all it was at the time --

154
00:08:08,875 --> 00:08:12,184
and I was in Muscatine, Iowa.

155
00:08:12,208 --> 00:08:14,101
I had barely heard of Muscatine,

156
00:08:14,125 --> 00:08:17,268
but it proved to be a fascinating place.

157
00:08:17,292 --> 00:08:21,601
Muscatine had some
of the world's richest mussel beds.

158
00:08:21,625 --> 00:08:24,726
A century ago,
a third of the world's buttons

159
00:08:24,750 --> 00:08:26,559
were produced in Muscatine,

160
00:08:26,583 --> 00:08:29,143
1.5 billion a year.

161
00:08:29,167 --> 00:08:31,434
The last plants have closed now,

162
00:08:31,458 --> 00:08:34,976
but there is still a museum
of the pearl button industry

163
00:08:35,000 --> 00:08:37,976
that's one of the most
unusual in the world.

164
00:08:38,000 --> 00:08:39,809
But buttons were only the beginning.

165
00:08:39,833 --> 00:08:41,726
This is the house in Muscatine

166
00:08:41,750 --> 00:08:44,809
where China's future
president stayed in 1986,

167
00:08:44,833 --> 00:08:47,518
as a member of an agricultural delegation.

168
00:08:47,542 --> 00:08:50,643
It is now the Sino-US Friendship House,

169
00:08:50,667 --> 00:08:53,768
and it's a pilgrimage site
for Chinese tourists.

170
00:08:53,792 --> 00:08:55,518
How could I have foreseen that?

171
00:08:55,542 --> 00:08:56,792
(Laughter)

172
00:08:57,958 --> 00:09:01,601
Second, get up from the couch.

173
00:09:01,625 --> 00:09:03,809
Sometimes it can be more efficient

174
00:09:03,833 --> 00:09:05,393
to do things the hard way.

175
00:09:05,417 --> 00:09:07,351
Consider the internet of things.

176
00:09:07,375 --> 00:09:10,018
It's wonderful
to be able to control lights,

177
00:09:10,042 --> 00:09:13,018
set the thermostat, even vacuum the room

178
00:09:13,042 --> 00:09:14,893
without leaving one's seat.

179
00:09:14,917 --> 00:09:17,143
But medical research has shown

180
00:09:17,167 --> 00:09:20,393
that actually fidgeting,
getting up, walking around

181
00:09:20,417 --> 00:09:23,601
is one of the best things
you can do for your heart.

182
00:09:23,625 --> 00:09:25,708
It's good for the heart and the waistline.

183
00:09:26,917 --> 00:09:29,934
Third, monetize your mistakes.

184
00:09:29,958 --> 00:09:31,684
Great forms can be created

185
00:09:31,708 --> 00:09:34,851
by imaginative development of accidents.

186
00:09:34,875 --> 00:09:37,809
Tad Leski, an architect
of the Metropolitan Opera

187
00:09:37,833 --> 00:09:39,101
at Lincoln Center,

188
00:09:39,125 --> 00:09:43,042
was working on a sketch
and some white ink fell on the drawing.

189
00:09:44,125 --> 00:09:47,059
Other people might just
have thrown it away,

190
00:09:47,083 --> 00:09:51,893
but Leski was inspired
to produce a starburst chandelier

191
00:09:51,917 --> 00:09:56,417
that was probably the most notable
of its kind of the 20th century.

192
00:09:57,292 --> 00:10:00,184
Fourth, sometimes try the hard way.

193
00:10:00,208 --> 00:10:03,559
It can be more efficient
to be less fluent.

194
00:10:03,583 --> 00:10:06,393
Psychologists call this
desirable difficulty.

195
00:10:06,417 --> 00:10:08,851
Taking detailed notes with a keyboard

196
00:10:08,875 --> 00:10:12,309
would seem to be the best way
to grasp what a lecturer is saying,

197
00:10:12,333 --> 00:10:15,476
to be able to review it verbatim.

198
00:10:15,500 --> 00:10:19,476
However, studies have shown
that when we have to abbreviate,

199
00:10:19,500 --> 00:10:22,559
when we have to summarize
what a speaker is saying,

200
00:10:22,583 --> 00:10:26,976
when we're taking notes
with a pen or a pencil on paper,

201
00:10:27,000 --> 00:10:28,976
we're processing that information.

202
00:10:29,000 --> 00:10:30,851
We're making that our own,

203
00:10:30,875 --> 00:10:33,184
and we are learning much more actively

204
00:10:33,208 --> 00:10:36,934
than when we were just transcribing

205
00:10:36,958 --> 00:10:38,809
what was being said.

206
00:10:38,833 --> 00:10:41,559
Fifth, get security through diversity.

207
00:10:41,583 --> 00:10:43,101
Monoculture can be deadly.

208
00:10:43,125 --> 00:10:44,559
Remember the potato?

209
00:10:44,583 --> 00:10:46,643
It was efficient until it wasn't.

210
00:10:46,667 --> 00:10:49,208
Diversity applies to organizations, too.

211
00:10:50,042 --> 00:10:55,726
Software can tell what has made people
in an organization succeed in the past.

212
00:10:55,750 --> 00:10:58,893
And it's useful, sometimes,
in screening employees.

213
00:10:58,917 --> 00:11:03,018
But remember, the environment
is constantly changing,

214
00:11:03,042 --> 00:11:06,976
and software, screening software,
has no way to tell,

215
00:11:07,000 --> 00:11:08,434
and we have no way to tell,

216
00:11:08,458 --> 00:11:11,934
who is going to be useful in the future.

217
00:11:11,958 --> 00:11:16,809
So, we need to supplement
whatever the algorithm tells us

218
00:11:16,833 --> 00:11:19,726
by an intuition and by looking for people

219
00:11:19,750 --> 00:11:23,226
with various backgrounds
and various outlooks.

220
00:11:23,250 --> 00:11:27,684
Sixth, achieve safety
through redundancy and human skills.

221
00:11:27,708 --> 00:11:31,268
Why did two 737 Max aircraft crash?

222
00:11:31,292 --> 00:11:33,393
We still don't know the full story,

223
00:11:33,417 --> 00:11:36,226
but we know how to
prevent future tragedies.

224
00:11:36,250 --> 00:11:38,851
We need multiple independent systems.

225
00:11:38,875 --> 00:11:42,893
If one fails, then the others
can override it.

226
00:11:42,917 --> 00:11:46,351
We also need skilled operators
to come to the rescue

227
00:11:46,375 --> 00:11:48,934
and that means constant training.

228
00:11:48,958 --> 00:11:52,143
Seventh, be rationally extravagant.

229
00:11:52,167 --> 00:11:54,851
Thomas Edison was a pioneer
of the film industry,

230
00:11:54,875 --> 00:11:57,268
as well as of camera technology.

231
00:11:57,292 --> 00:12:01,268
Nobody has done more
for efficiency than Thomas Edison.

232
00:12:01,292 --> 00:12:03,125
But his cost cutting broke down.

233
00:12:03,958 --> 00:12:06,934
His manager hired
a so-called efficiency engineer,

234
00:12:06,958 --> 00:12:09,268
who advised him to save money

235
00:12:09,292 --> 00:12:12,851
by using more of the film stock
that he'd shot,

236
00:12:12,875 --> 00:12:14,559
having fewer retakes.

237
00:12:14,583 --> 00:12:15,851
Well, Edison was a genius,

238
00:12:15,875 --> 00:12:19,309
but he didn't understand
the new rules of feature films

239
00:12:19,333 --> 00:12:24,684
and the fact that failure
was becoming the price of success.

240
00:12:24,708 --> 00:12:28,101
On the other hand, some great directors,
like Erich Von Stroheim,

241
00:12:28,125 --> 00:12:29,393
were the opposite.

242
00:12:29,417 --> 00:12:30,809
They were superb dramatists,

243
00:12:30,833 --> 00:12:34,101
and Stroheim was also a memorable actor.

244
00:12:34,125 --> 00:12:36,268
But they couldn't live
within their budgets.

245
00:12:36,292 --> 00:12:38,042
So that was not sustainable.

246
00:12:39,208 --> 00:12:43,184
It was Irving Thalberg,
a former secretary with intuitive genius,

247
00:12:43,208 --> 00:12:45,559
who achieved rational extravagance.

248
00:12:45,583 --> 00:12:48,601
First at Universal, and then at MGM,

249
00:12:48,625 --> 00:12:52,809
becoming the ideal
of the Hollywood producer.

250
00:12:52,833 --> 00:12:56,101
Summing up, to be truly efficient,

251
00:12:56,125 --> 00:12:58,643
we need optimal inefficiency.

252
00:12:58,667 --> 00:13:01,268
The shortest path may be a curve

253
00:13:01,292 --> 00:13:03,434
rather than a straight line.

254
00:13:03,458 --> 00:13:05,643
Charles Darwin understood that.

255
00:13:05,667 --> 00:13:07,434
When he encountered a tough problem,

256
00:13:07,458 --> 00:13:09,226
he made a circuit of a trail,

257
00:13:09,250 --> 00:13:12,125
the sandwalk that he'd built
behind his house.

258
00:13:12,875 --> 00:13:16,101
A productive path
can be physical, like Darwin's,

259
00:13:16,125 --> 00:13:19,559
or a virtual one, or an unforeseen detour

260
00:13:19,583 --> 00:13:22,059
from a path we had laid out.

261
00:13:22,083 --> 00:13:24,518
Too much efficiency can weaken itself.

262
00:13:24,542 --> 00:13:28,976
But a bit of inspired inefficiency
can strengthen it.

263
00:13:29,000 --> 00:13:32,976
Sometimes, the best way to move forward

264
00:13:33,000 --> 00:13:34,851
is to follow a circle.

265
00:13:34,875 --> 00:13:36,184
Thank you.

266
00:13:36,208 --> 00:13:39,042
(Applause)