1
00:00:00,033 --> 00:00:01,700
Hey, everyone.

2
00:00:01,767 --> 00:00:04,600
This year, we're doing
a series of public discussions

3
00:00:04,667 --> 00:00:08,400
on important social issues
facing the Internet.

4
00:00:08,467 --> 00:00:12,133
And today, we're here to talk
about governance issues

5
00:00:12,200 --> 00:00:15,767
around giving people a voice and content.

6
00:00:15,833 --> 00:00:20,200
And finding the right balance between
free expression and speech,

7
00:00:20,267 --> 00:00:24,967
and making sure that safety is respected,

8
00:00:25,033 --> 00:00:28,200
and harmful content can be taken down
from the Internet as well.

9
00:00:28,267 --> 00:00:31,100
And today, I'm here
with two folks,

10
00:00:31,167 --> 00:00:35,400
Jenny Martinez, the dean of Stanford Law,

11
00:00:35,467 --> 00:00:41,267
who is a scholar in International Studies
and Human Rights.

12
00:00:41,333 --> 00:00:45,600
And Noah Feldman,
who is a professor at Harvard Law,

13
00:00:45,700 --> 00:00:51,167
who has studied constitutional law
and a focus on free speech and expression.

14
00:00:51,467 --> 00:00:56,433
And to give a background
on how we're approaching this...

15
00:00:57,633 --> 00:01:00,367
I believe that
if we were writing the rules

16
00:01:00,433 --> 00:01:02,767
for the Internet today from scratch,

17
00:01:02,833 --> 00:01:05,033
I don't think that
we would want private companies

18
00:01:05,100 --> 00:01:09,367
to be making so many
fundamental decisions by themselves

19
00:01:09,433 --> 00:01:13,533
about important things like
what speech is acceptable,

20
00:01:13,600 --> 00:01:18,667
how to prevent election interference,
what the privacy rule should be.

21
00:01:19,467 --> 00:01:21,933
So, one of the things
I've been outspoken on

22
00:01:22,000 --> 00:01:26,733
is talking about the need
for more regulation of these areas,

23
00:01:26,800 --> 00:01:30,367
in some cases, by governments,
and through democratic process,

24
00:01:30,433 --> 00:01:33,767
in other case,
through independent industrial process.

25
00:01:35,033 --> 00:01:40,633
And at Facebook, we're not waiting
until that process goes through.

26
00:01:40,700 --> 00:01:43,533
We are trying to take
some steps independently

27
00:01:43,600 --> 00:01:48,233
to establish things like
this independent oversight board

28
00:01:48,300 --> 00:01:53,300
that will, for content,
which will give people in our community

29
00:01:53,367 --> 00:01:57,300
the ability to appeal content decisions
that we make.

30
00:01:57,367 --> 00:02:00,767
So, that way,
if Facebook takes a piece of content down

31
00:02:00,833 --> 00:02:04,867
and you think that it shouldn't be taken
down, then you'll be able to appeal it.

32
00:02:04,933 --> 00:02:08,867
And if you still disagree
with our treatment on appeal,

33
00:02:08,933 --> 00:02:12,267
you'll be able to appeal it
to this external, independent board,

34
00:02:12,333 --> 00:02:14,800
and its decision will be binding.

35
00:02:14,867 --> 00:02:19,133
So, if it says that
something needs to stay up,

36
00:02:19,200 --> 00:02:23,733
it will stay up, no matter what I think
or our teams at Facebook think.

37
00:02:23,800 --> 00:02:29,233
And so, this is an experiment in
independent governance around expression,

38
00:02:29,300 --> 00:02:33,367
which we've been working on
for more than a year now.

39
00:02:33,433 --> 00:02:36,700
I've just talked to a lot of
different academics around this.

40
00:02:36,767 --> 00:02:41,867
Noah was originally involved in helping
to think through what this idea would be.

41
00:02:41,933 --> 00:02:47,867
And we've taken the last year or so
to go run trials around the world, right.

42
00:02:47,933 --> 00:02:51,400
So, we've had almost 30 trials
in a lot of different cities,

43
00:02:51,467 --> 00:02:55,467
experimenting with different constructs
around how this board would work.

44
00:02:55,533 --> 00:03:02,433
And we've just published today a report
of all the feedback that we've gotten,

45
00:03:02,500 --> 00:03:05,433
which outlined some of what
we've learned from these trials,

46
00:03:05,500 --> 00:03:08,300
but also highlights all the tensions

47
00:03:08,367 --> 00:03:10,600
between different ways
that we could design this.

48
00:03:10,667 --> 00:03:14,200
And now, the next step
going forward is going to be

49
00:03:14,267 --> 00:03:15,867
to try to lock down
some of these decisions,

50
00:03:15,933 --> 00:03:18,400
make some of these decisions
over the next few months,

51
00:03:18,467 --> 00:03:22,433
and launch this independent
oversight board by the end of the year.

52
00:03:22,500 --> 00:03:26,067
So, there's a lot that
this board could eventually do,

53
00:03:26,133 --> 00:03:28,533
the goal is going to be to start narrowly

54
00:03:28,600 --> 00:03:31,267
and then eventually over time,
expand its scope

55
00:03:31,333 --> 00:03:34,967
and hopefully include
more folks in the industry as well.

56
00:03:35,033 --> 00:03:39,633
But this is a big experiment that we hope

57
00:03:39,700 --> 00:03:45,967
can pioneer a new model
for governance of speech on the Internet.

58
00:03:46,033 --> 00:03:49,167
So, thank you for joining me
to go through this.

59
00:03:50,700 --> 00:03:53,367
I think it might be useful to start off

60
00:03:53,433 --> 00:03:57,433
by discussing the historical context
that we exist in now.

61
00:03:57,500 --> 00:04:00,733
Because, clearly,
the Internet is not the first time

62
00:04:00,800 --> 00:04:04,300
that societies have struggled
with these questions around expression,

63
00:04:04,367 --> 00:04:10,400
and safety and privacy and decency
and the different values that we hold.

64
00:04:10,467 --> 00:04:12,300
So, I'd be curious just...

65
00:04:12,367 --> 00:04:17,133
In the different scholarship
that you've each done,

66
00:04:17,200 --> 00:04:21,600
how would you characterize the moment
that we're in and the historical tradition

67
00:04:21,667 --> 00:04:28,200
in which this struggle
and approach to governance fits?

68
00:04:29,633 --> 00:04:31,233
I think one interesting thing, Mark,

69
00:04:31,300 --> 00:04:34,533
that you teed up,
you began the conversation,

70
00:04:34,600 --> 00:04:37,967
was that people
have these concerns because

71
00:04:38,033 --> 00:04:42,167
companies like Facebook have, really,
very global power,

72
00:04:42,233 --> 00:04:45,400
and power that seems
a lot like what governments have

73
00:04:45,467 --> 00:04:50,167
in terms of the ability to have an impact
on people's lives or to control speech.

74
00:04:50,233 --> 00:04:52,267
And with more than two billion users,

75
00:04:52,333 --> 00:04:55,867
it's bigger than
any country on the planet, and so,

76
00:04:55,933 --> 00:04:59,100
the sort of concentration of power,
and then also in a company

77
00:04:59,167 --> 00:05:01,133
that isn't accountable in the way

78
00:05:01,200 --> 00:05:04,667
that a democratically-elected
government would be,

79
00:05:04,733 --> 00:05:07,400
is something that gives people pause
and makes them think,

80
00:05:07,467 --> 00:05:10,633
"How should we manage this?"
Especially on an international scale.

81
00:05:10,700 --> 00:05:13,867
I think one thing
that's interesting, historically, is that

82
00:05:13,933 --> 00:05:19,733
the distinction between government power
and private power that we have today

83
00:05:19,800 --> 00:05:23,100
is very sharp, where there's a clear line
between what the government is

84
00:05:23,167 --> 00:05:27,400
and what governments do and what
we think of as private forms of power.

85
00:05:27,467 --> 00:05:31,333
But if you look back even just
a couple hundred years in human history,

86
00:05:31,400 --> 00:05:34,333
to the 1600s, the 1700s,

87
00:05:34,400 --> 00:05:39,267
there were previous periods
where very large companies played a role

88
00:05:39,333 --> 00:05:42,933
that also straddled the divide
between public and private power.

89
00:05:43,000 --> 00:05:47,733
So, the British East India Company,
the Dutch East India Company,

90
00:05:47,800 --> 00:05:53,833
those companies, in the 1600s and 1700s,
had very wide range, geographically.

91
00:05:53,900 --> 00:05:58,167
They also engaged in powers that today
we would think of as government powers.

92
00:05:58,233 --> 00:06:02,400
So, they coined money, they raised armies,
they entered into treaties.

93
00:06:02,467 --> 00:06:06,433
Things that, today, we would think
as exclusively the province of states.

94
00:06:06,500 --> 00:06:08,900
At earlier points in history,
just that line between

95
00:06:08,967 --> 00:06:13,767
what states did and what other actors did
on the international stage was more fluid.

96
00:06:13,833 --> 00:06:17,733
And so, I don't think the period
today is exactly the same.

97
00:06:17,800 --> 00:06:20,100
There is huge differences in the world,

98
00:06:20,167 --> 00:06:22,833
and the economy,
and the role of the companies.

99
00:06:22,900 --> 00:06:26,200
But it does show that
the categories of what governments do

100
00:06:26,267 --> 00:06:29,200
and what private companies do,
are not so neat and clean

101
00:06:29,267 --> 00:06:32,133
as they might've seemed
for part of the 20th century.

102
00:06:32,200 --> 00:06:35,767
And so, it's a governance challenge,
globally, that which,

103
00:06:35,833 --> 00:06:39,767
I think changed the way that
international law itself works over time.

104
00:06:39,833 --> 00:06:43,600
Maybe I can pick up the historical story
where Jenny left it off.

105
00:06:43,667 --> 00:06:48,667
So, you get into the 1700s and you get
these colonies all over North America.

106
00:06:48,733 --> 00:06:51,367
Some of them are actually like companies.

107
00:06:51,433 --> 00:06:54,667
They get a charter from the crown,
they are in business,

108
00:06:54,733 --> 00:06:56,867
some of them make it, some of them don't.

109
00:06:56,933 --> 00:07:00,067
But, eventually, they emerge
as quasi-independent entities,

110
00:07:00,133 --> 00:07:02,700
and then they become states
with independence.

111
00:07:02,800 --> 00:07:05,133
And then they write
these state constitutions,

112
00:07:05,200 --> 00:07:07,800
and that's really the first time
that people start saying,

113
00:07:07,867 --> 00:07:12,633
"Maybe free speech needs
to be protected by a formal rule."

114
00:07:12,700 --> 00:07:16,767
Until then, there were traditions
of freedom of expression in some places,

115
00:07:16,833 --> 00:07:20,200
but most governments
before then just thought,

116
00:07:20,267 --> 00:07:23,633
"Well, if I don't like
what this speech is, I'll shut it down."

117
00:07:23,700 --> 00:07:26,500
You can either shut it down
before someone talks, by having a censor,

118
00:07:26,567 --> 00:07:29,433
who has to review everything
before it could be published,

119
00:07:29,500 --> 00:07:31,567
or you can punish people
after they've spoken

120
00:07:31,633 --> 00:07:33,267
if you don't like what they've done.

121
00:07:33,333 --> 00:07:36,233
And then, over the next
couple of hundred years,

122
00:07:36,300 --> 00:07:40,200
this constitutional practice
of defending free speech with the rule

123
00:07:40,267 --> 00:07:45,033
slowly developed mostly in America and
mostly without being applied in practice.

124
00:07:45,100 --> 00:07:47,133
And it wasn't really until
the beginning of the 20th century

125
00:07:47,200 --> 00:07:50,567
that even the U.S. Supreme Court
started saying, "Hey, wait a minute.

126
00:07:50,633 --> 00:07:53,233
"Government, you passed that law.
That violates the freedom of speech."

127
00:07:53,700 --> 00:07:55,967
And then that went global
after World War II.

128
00:07:56,700 --> 00:08:02,000
After World War II is where you suddenly
get countries rather realizing

129
00:08:02,067 --> 00:08:04,567
that instead of acting as censors,

130
00:08:04,633 --> 00:08:07,067
they're much better off
having a constitutional court

131
00:08:07,133 --> 00:08:10,267
that's quasi-independent,
and it's part of the government,

132
00:08:10,333 --> 00:08:13,300
but it's an independent decision-maker
within the government.

133
00:08:13,367 --> 00:08:17,033
And then giving that body the authority
to decide about speech issue.

134
00:08:17,100 --> 00:08:20,367
So, that kind of spinning off
separation of powers approach

135
00:08:20,433 --> 00:08:22,133
is really a product to the last 75 years.

136
00:08:23,233 --> 00:08:24,500
Yeah, very interesting.

137
00:08:24,567 --> 00:08:29,333
So, when I first started thinking
about this idea of how to get

138
00:08:29,400 --> 00:08:30,633
more of the community involved,

139
00:08:30,700 --> 00:08:33,433
how to build an independent process for...

140
00:08:34,400 --> 00:08:36,833
To get involved with
more of our content policies,

141
00:08:36,900 --> 00:08:41,433
because there's a strong feeling that,
you know, this is not something that

142
00:08:41,500 --> 00:08:44,167
a private company
should be just defining by itself.

143
00:08:46,033 --> 00:08:49,767
The question's about what people
can say in different contexts,

144
00:08:49,833 --> 00:08:52,300
if it affects people broadly,

145
00:08:52,367 --> 00:08:56,367
and is not something that
I think we would want

146
00:08:56,433 --> 00:08:59,633
any one set of people
deciding by themselves.

147
00:09:00,633 --> 00:09:05,367
The original idea
that I was thinking about

148
00:09:05,433 --> 00:09:12,100
was more to involve independent folks
in setting the policies directly.

149
00:09:12,933 --> 00:09:16,000
In some of our early conversations,

150
00:09:16,067 --> 00:09:20,333
you had basically proposed
and convinced me

151
00:09:20,400 --> 00:09:23,967
that it would make more sense to
start with more of an appeals-type analog

152
00:09:24,033 --> 00:09:28,733
rather than a legislative-type analog
or policy-setting analog.

153
00:09:28,800 --> 00:09:32,933
I'm curious to hear you talk through
in a bit more detail

154
00:09:33,000 --> 00:09:37,233
about why you think that that's
the right approach to protect speech

155
00:09:37,300 --> 00:09:41,067
and to engage in the kind of independence

156
00:09:41,133 --> 00:09:45,367
that we want to
provide for this community.

157
00:09:45,433 --> 00:09:48,000
There are really two reasons.
One is modesty.

158
00:09:48,733 --> 00:09:53,367
In the end, Facebook is a company,
not a country,

159
00:09:53,433 --> 00:09:57,867
and so, it doesn't have
a democratic base of citizens

160
00:09:57,933 --> 00:10:01,433
who could vote for elected representatives
in some obvious way,

161
00:10:01,500 --> 00:10:06,833
and then pass laws that would then be
enforced by a government apparatus.

162
00:10:06,900 --> 00:10:08,800
And so, it's hard, therefore, to say,

163
00:10:08,867 --> 00:10:12,800
"Let's draft out some new rules
for governance using this."

164
00:10:12,867 --> 00:10:14,500
But the second reason
has to do with the fact that,

165
00:10:14,567 --> 00:10:16,900
actually, even among democracies,

166
00:10:16,967 --> 00:10:20,467
no democracy thinks that it should
use its elected representatives

167
00:10:20,533 --> 00:10:23,300
to make the ultimate decisions
on their freedom of speech.

168
00:10:23,367 --> 00:10:27,533
No government says, "Let's not have
a constitutional court nowadays.

169
00:10:27,600 --> 00:10:31,067
"Let's just vote on what speech should
be allowed or what shouldn't be allowed."

170
00:10:31,133 --> 00:10:34,500
Because the experience that most countries
have had with that is a bad one.

171
00:10:34,567 --> 00:10:39,300
The public can very happily
shut down speech when it seems useful.

172
00:10:39,367 --> 00:10:41,233
The majority will
try to silence the minority.

173
00:10:41,300 --> 00:10:43,200
It's good common sense.

174
00:10:43,267 --> 00:10:46,833
And so, it seemed to me that,
logically, similarly,

175
00:10:46,900 --> 00:10:48,867
Facebook, if it's interested
in protecting voice,

176
00:10:48,933 --> 00:10:50,600
which seems to me
like you have to be...

177
00:10:50,667 --> 00:10:54,267
Yes. Yeah, that's what we stand for,
is giving people a voice.

178
00:10:54,333 --> 00:10:56,433
Yeah, no voice, no Facebook, right?

179
00:10:56,500 --> 00:10:58,467
People have to be able
to express themselves.

180
00:10:58,533 --> 00:11:02,200
Would similarly want to take advantage
of an approach that says,

181
00:11:02,267 --> 00:11:04,600
"Let's have an independent,
decision-making body

182
00:11:04,667 --> 00:11:07,367
"that will stand
for the principle of voice."

183
00:11:07,433 --> 00:11:09,767
And when there are circumstances
where we have to balance voice

184
00:11:09,833 --> 00:11:12,900
against people's safety
or against people's equality,

185
00:11:12,967 --> 00:11:16,633
that body can make those judgments
and can do it in an open way,

186
00:11:16,700 --> 00:11:18,967
transparently explaining the trade-offs,

187
00:11:19,033 --> 00:11:22,800
and ultimately,
defensively, to the world.

188
00:11:22,867 --> 00:11:24,767
That's where public legitimacy comes in.

189
00:11:24,833 --> 00:11:26,800
If a court explains
why it's doing what it's doing,

190
00:11:26,867 --> 00:11:28,667
if it says, "This is why
we're doing it this way",

191
00:11:28,733 --> 00:11:29,933
if it says, "Here's the balance",

192
00:11:30,000 --> 00:11:32,467
then people can say, "We agree",
or "We don't agree",

193
00:11:32,533 --> 00:11:34,067
and people can engage
in their own judgment.

194
00:11:34,133 --> 00:11:35,167
Yeah.

195
00:11:35,233 --> 00:11:37,767
Well, so you raise
this question around legitimacy,

196
00:11:37,833 --> 00:11:43,133
which I think is one of the core questions
in designing this kind of a system.

197
00:11:43,200 --> 00:11:46,867
For those who
have questions, rightfully, about,

198
00:11:46,933 --> 00:11:51,100
"Why should a private company
be making all of these decisions?"

199
00:11:52,167 --> 00:11:56,867
That then begs the question of any system
that you've set up that's independent,

200
00:11:56,933 --> 00:12:00,467
how does it derive its legitimacy
in a democratic process

201
00:12:00,533 --> 00:12:04,667
that's through voting, but here,
like you say... I mean, this is...

202
00:12:04,733 --> 00:12:06,967
It wouldn't make sense
for a number of reasons

203
00:12:07,033 --> 00:12:10,633
to have the community vote on people

204
00:12:10,700 --> 00:12:15,400
who are then deciding the speech policies.

205
00:12:16,467 --> 00:12:20,767
But I'm curious how you think about

206
00:12:20,833 --> 00:12:24,467
how this can derive its legitimacy

207
00:12:24,533 --> 00:12:28,467
as an independent appeals process,

208
00:12:28,533 --> 00:12:31,067
and where are the balances between

209
00:12:31,133 --> 00:12:34,533
something that industry
or we should be doing

210
00:12:34,600 --> 00:12:39,000
in standing up ourselves versus where
more democratic government regulation

211
00:12:39,067 --> 00:12:42,967
might be better
and more effective in this space?

212
00:12:43,467 --> 00:12:46,200
I think legitimacy can come
from a couple of different sources.

213
00:12:46,267 --> 00:12:50,467
One is the process itself,
and whether it's viewed as fair.

214
00:12:50,533 --> 00:12:51,933
Is it transparent?

215
00:12:52,000 --> 00:12:56,233
Are people viewed as being subject
to improper motivation or influence?

216
00:12:56,300 --> 00:12:57,933
Or are they seen as independent

217
00:12:58,000 --> 00:13:00,800
and trying to fairly
apply the rules that are there?

218
00:13:00,867 --> 00:13:03,733
I think there's also some legitimacy

219
00:13:03,800 --> 00:13:08,133
that comes from the perception
that the values that are being applied

220
00:13:08,200 --> 00:13:11,733
are one that are somehow accepted
by broader society.

221
00:13:11,800 --> 00:13:15,167
And one thing I think you've talked about
in some of the materials

222
00:13:15,233 --> 00:13:17,567
that I think is relevant to that,
is the question of

223
00:13:17,633 --> 00:13:22,333
whether you try to relate back
your values to some external standard

224
00:13:22,400 --> 00:13:24,900
that's been created
through a public process,

225
00:13:24,967 --> 00:13:26,967
like international human rights law,

226
00:13:27,033 --> 00:13:29,633
where you have countries coming together

227
00:13:29,700 --> 00:13:34,367
and saying, "Expression is valued
and protected",

228
00:13:34,433 --> 00:13:38,233
and then naming some of the other values
that society might have

229
00:13:38,300 --> 00:13:41,000
that you would,
in those hard cases, be balancing,

230
00:13:41,067 --> 00:13:43,733
whether it's public safety or equality.

231
00:13:43,800 --> 00:13:49,200
And I do think you can gain some
legitimacy by trying to tie things back

232
00:13:49,267 --> 00:13:53,067
to those public values into something
like international human rights law,

233
00:13:53,133 --> 00:13:57,400
which has been accepted by
most of the countries in the world.

234
00:13:57,467 --> 00:14:00,267
And it may not give all the answers
to hard questions,

235
00:14:00,333 --> 00:14:02,367
but it gives you at least some sense of

236
00:14:02,433 --> 00:14:04,233
"What are the values
you would be weighing?"

237
00:14:04,300 --> 00:14:08,400
And that's in a way more legitimate
than just saying these are our values,

238
00:14:08,467 --> 00:14:11,267
well, these are actually values
they've been adopted and endorsed

239
00:14:11,333 --> 00:14:14,867
by a more public process around the world.

240
00:14:14,933 --> 00:14:18,900
I think it makes a lot of sense,
and I know that within Facebook,

241
00:14:18,967 --> 00:14:22,500
you guys are working towards coming up
with a distinctive set of values.

242
00:14:22,567 --> 00:14:25,067
In my conversations
with people in Facebook,

243
00:14:25,133 --> 00:14:29,033
those always seem to be informed
very much, by international law.

244
00:14:29,100 --> 00:14:32,567
Very much reflecting those ideals
and those standards.

245
00:14:32,633 --> 00:14:35,567
They have to be tweaked a little bit,
because, again, Facebook, company,

246
00:14:35,633 --> 00:14:36,733
not a country.

247
00:14:36,800 --> 00:14:40,833
Most of those International Human Rights
standards are designed for a country.

248
00:14:40,900 --> 00:14:42,667
So it might not be exactly the same,

249
00:14:42,733 --> 00:14:45,567
but I think they're informed by it
and I think that is in people's minds.

250
00:14:45,633 --> 00:14:48,200
I think legitimacy will come
from seeing in action

251
00:14:48,267 --> 00:14:52,500
an independent body do some things
that you guys don't want them to do.

252
00:14:52,567 --> 00:14:54,200
Yeah. Yeah.

253
00:14:54,267 --> 00:14:59,667
Legitimacy will be real
when people see decisions

254
00:14:59,733 --> 00:15:03,267
that are different from what Facebook
would have otherwise decided to do.

255
00:15:03,333 --> 00:15:07,433
I think we live in
a very sophisticated world today

256
00:15:07,500 --> 00:15:09,100
where you couldn't create legitimacy
for something

257
00:15:09,167 --> 00:15:11,300
unless it was genuinely independent.

258
00:15:11,367 --> 00:15:14,233
People would want to see
the genuine independence in action.

259
00:15:14,300 --> 00:15:16,200
Yeah, that's my expectation,

260
00:15:16,267 --> 00:15:19,700
that we can say
this is an independent process,

261
00:15:19,767 --> 00:15:22,433
we can have a consultative approach

262
00:15:22,500 --> 00:15:25,733
to helping design it
and figure out who should be on it.

263
00:15:25,800 --> 00:15:29,033
But the trust in the institution
will need to be built up over time.

264
00:15:29,100 --> 00:15:31,567
I would imagine that the first time where

265
00:15:31,633 --> 00:15:35,500
our content review team
at Facebook makes a decision

266
00:15:37,033 --> 00:15:38,767
and I agree with that decision,

267
00:15:38,833 --> 00:15:42,300
but then this board says,
"No, it's going the other way,"

268
00:15:42,367 --> 00:15:43,667
and we listen.

269
00:15:45,333 --> 00:15:46,833
Our systems are hardwired

270
00:15:46,900 --> 00:15:50,200
that what it decides actually
gets reflected in the systems.

271
00:15:50,267 --> 00:15:53,600
Then we have to figure out how
the policies need to evolve to fit that.

272
00:15:54,633 --> 00:15:57,300
I kind of imagine
we will go through a number of those

273
00:15:57,367 --> 00:16:00,433
and that will increase trust
if this is an institution.

274
00:16:01,100 --> 00:16:02,533
Even if that creates other questions

275
00:16:02,600 --> 00:16:09,033
around the balance and the tension
between it and in our own operations.

276
00:16:10,700 --> 00:16:15,333
I think this is going to be a very
interesting one to see how it evolves.

277
00:16:16,033 --> 00:16:21,100
Jenny, you had mentioned some
of the different theories or frameworks

278
00:16:21,167 --> 00:16:25,400
for thinking about expression
of the international notion

279
00:16:25,467 --> 00:16:30,500
versus the American sense,
with the First Amendment

280
00:16:32,700 --> 00:16:35,767
that is much stronger
on protecting speech...

281
00:16:37,933 --> 00:16:40,067
rather than balancing it
against other values.

282
00:16:40,567 --> 00:16:45,400
I'm curious how you think
the standards on the Internet,

283
00:16:46,633 --> 00:16:48,900
I guess for our services, but broadly,

284
00:16:48,967 --> 00:16:54,067
the values fit within
the different frameworks

285
00:16:54,133 --> 00:16:57,200
that we've seen,
both the American one, European one,

286
00:16:57,267 --> 00:16:58,867
and others around the world as well.

287
00:16:58,933 --> 00:17:03,133
Even in the U.S.,
I think the court has recognized

288
00:17:03,200 --> 00:17:05,833
that the value of expression
has to be weighed,

289
00:17:05,900 --> 00:17:08,500
sometimes,
against other values in society.

290
00:17:08,567 --> 00:17:12,000
Famously, you can't shout "fire"
in a crowded theater.

291
00:17:12,067 --> 00:17:15,300
There are limits on free expression,
even in the United States,

292
00:17:15,367 --> 00:17:16,800
as the court has looked at,

293
00:17:16,867 --> 00:17:19,033
"What do you need
to have a society function?"

294
00:17:19,633 --> 00:17:20,767
Definitely the U.S.,

295
00:17:20,833 --> 00:17:22,567
if you were to line up
countries in the world,

296
00:17:22,633 --> 00:17:25,233
we are on the end
of protecting more speech.

297
00:17:25,300 --> 00:17:27,267
Honestly, that's where
I sort of come from,

298
00:17:27,333 --> 00:17:31,467
and I think you can't shake
your own legal background.

299
00:17:31,533 --> 00:17:33,533
As an American constitutional lawyer,

300
00:17:33,600 --> 00:17:36,200
I tend always towards
freedom of expression.

301
00:17:36,267 --> 00:17:39,300
When you look at, for example,
the European approach,

302
00:17:39,367 --> 00:17:42,633
each part of the world
has been shaped by its own history.

303
00:17:42,700 --> 00:17:45,367
When you see countries
like Germany or France

304
00:17:45,433 --> 00:17:47,933
having more restrictions on hate speech,

305
00:17:48,000 --> 00:17:49,567
or on Holocaust denial,

306
00:17:49,633 --> 00:17:53,300
it's really rooted in their
particular social and legal history,

307
00:17:53,367 --> 00:17:57,300
where the laws that they created
about freedom of expression

308
00:17:57,367 --> 00:17:59,933
in the late 1940s and '50s

309
00:18:00,000 --> 00:18:04,333
reacted to the experience they had
recently undergone in World War II,

310
00:18:04,400 --> 00:18:06,667
and the ways in which they saw

311
00:18:06,733 --> 00:18:11,100
that extreme hate speech
could lead to violence on a mass scale.

312
00:18:11,167 --> 00:18:15,133
One of the challenges
for an enterprise like this is

313
00:18:15,200 --> 00:18:18,667
how to balance the legitimate differences
among countries,

314
00:18:18,733 --> 00:18:20,933
in terms of their own history and culture,

315
00:18:21,000 --> 00:18:25,333
and how they want to balance
what are really hard choices at the edge.

316
00:18:25,400 --> 00:18:29,833
How to accommodate the approaches
of various parts of the world,

317
00:18:29,900 --> 00:18:34,833
while also trying to set
some standard of protection

318
00:18:34,900 --> 00:18:36,767
that you want to apply across the board.

319
00:18:36,833 --> 00:18:38,367
I think that's a real challenge.

320
00:18:38,433 --> 00:18:42,733
Mark, how have you thought about
the international dimension of this?

321
00:18:42,800 --> 00:18:43,933
You are an American,

322
00:18:44,000 --> 00:18:48,233
so you have the free expression sentiment
that we all are raised with.

323
00:18:48,300 --> 00:18:50,100
But you are running a global company.

324
00:18:50,933 --> 00:18:55,033
How have you traditionally found yourself
thinking about that challenge?

325
00:18:56,267 --> 00:18:59,367
So Facebook stands
for giving people a voice.

326
00:18:59,433 --> 00:19:02,033
That's such a fundamental part
of what we do.

327
00:19:02,100 --> 00:19:04,100
There are all these obstacles

328
00:19:04,167 --> 00:19:08,300
to giving people as much voice
as you would want.

329
00:19:08,367 --> 00:19:11,467
Some of them are these
legal or policy-oriented ones

330
00:19:11,533 --> 00:19:13,433
that we are talking about now.

331
00:19:13,500 --> 00:19:15,733
A lot of them are just technical.

332
00:19:15,800 --> 00:19:19,367
There are people don't have access
to the Internet or tools,

333
00:19:19,433 --> 00:19:23,533
or don't have access to as high fidelity
of tools to give people a voice.

334
00:19:23,600 --> 00:19:27,500
Those are less controversial
to break down those boundaries.

335
00:19:27,567 --> 00:19:32,167
But a lot of what I see us doing

336
00:19:32,233 --> 00:19:36,300
is basically trying
to move in the direction

337
00:19:36,367 --> 00:19:38,200
of giving more people a voice.

338
00:19:39,100 --> 00:19:42,400
When we talk about connecting people,

339
00:19:43,400 --> 00:19:47,267
the voice, and being able to choose
who you connect with and associate with,

340
00:19:47,333 --> 00:19:49,300
what you say to different people,

341
00:19:49,367 --> 00:19:52,900
is fundamental in our ability
to connect as people.

342
00:19:52,967 --> 00:19:55,600
When we talk about giving small businesses

343
00:19:55,667 --> 00:19:59,167
access to the same opportunities that,
traditionally, the big guys have had,

344
00:19:59,233 --> 00:20:01,633
whether it gets to finding customers

345
00:20:01,700 --> 00:20:04,033
or getting the word out
about what they are doing,

346
00:20:04,100 --> 00:20:05,567
that comes down to voice too.

347
00:20:05,633 --> 00:20:09,767
I think that voice
is incredibly empowering

348
00:20:09,833 --> 00:20:16,733
and in a lot of ways, is the primary value
and principle that we care about.

349
00:20:16,800 --> 00:20:19,800
And one of the things
I think we've struggled with

350
00:20:19,867 --> 00:20:23,867
is that there are
different principles and values.

351
00:20:25,267 --> 00:20:28,700
I think often, people don't acknowledge
the trade-offs between different things.

352
00:20:28,767 --> 00:20:31,867
But there are also trade-offs
between expression

353
00:20:31,933 --> 00:20:35,333
on the one hand and safety, or privacy.

354
00:20:35,400 --> 00:20:41,333
You should not have the ability
to go reveal people's private information

355
00:20:41,400 --> 00:20:43,767
that they didn't want to get revealed.

356
00:20:43,833 --> 00:20:46,700
You should not have the ability
to go bully people

357
00:20:46,767 --> 00:20:51,267
or use your voice
to go organize and incite violence, right?

358
00:20:51,333 --> 00:20:55,267
Those are things
we would all agree are bad.

359
00:20:56,300 --> 00:20:57,833
And then I think
these different traditions,

360
00:20:57,900 --> 00:21:01,033
whether it's the European one
or the American one

361
00:21:01,100 --> 00:21:03,367
have somewhat different ways
of balancing them.

362
00:21:03,433 --> 00:21:06,200
But I do think you have to acknowledge
that there are different values at play.

363
00:21:06,267 --> 00:21:08,867
Even if our goal is
to try to maximize voice,

364
00:21:08,933 --> 00:21:13,800
there are going to be these other values
that we really hold dear

365
00:21:13,867 --> 00:21:17,667
and feel a deep responsibility
to manage well for the community.

366
00:21:17,733 --> 00:21:21,633
One of the interesting challenges
in setting up this board is,

367
00:21:21,700 --> 00:21:25,700
I wouldn't want to set it up so that...

368
00:21:26,633 --> 00:21:31,133
I would worry, one of the issues
with expression overall

369
00:21:31,200 --> 00:21:34,567
is people are generally
for expression in the abstract.

370
00:21:34,633 --> 00:21:38,367
But in many individual cases
where something is controversial,

371
00:21:38,433 --> 00:21:42,033
the people lobbying
to take down a piece of content

372
00:21:42,100 --> 00:21:45,300
end up being a lot louder
than the people who, in the abstract,

373
00:21:45,367 --> 00:21:50,300
are for having it maintain
the principle of expression.

374
00:21:50,367 --> 00:21:52,800
So, the question is,
how do you design a system

375
00:21:52,867 --> 00:21:56,600
that can protect that more abstract

376
00:21:56,667 --> 00:21:59,767
but also very fundamental notion
of giving people a voice,

377
00:21:59,833 --> 00:22:02,500
while also acknowledging
there are these other values

378
00:22:02,567 --> 00:22:05,100
that are dear and are incredibly important

379
00:22:05,167 --> 00:22:09,300
to our responsibility to our community
that we need to uphold as well?

380
00:22:09,367 --> 00:22:13,267
That's kind of what I was hoping
to get out in the discussion

381
00:22:13,333 --> 00:22:16,367
between the kind of European approach

382
00:22:16,433 --> 00:22:19,567
or the International Human Rights frame
on this,

383
00:22:19,633 --> 00:22:22,000
compared to the American approach which,

384
00:22:22,067 --> 00:22:25,800
granted, the American approach
doesn't say that everything is allowed,

385
00:22:25,867 --> 00:22:30,500
but it certainly gives much more deference
to voice and expression.

386
00:22:30,567 --> 00:22:35,667
And I'm curious how you think that
we should balance that on the Internet

387
00:22:35,733 --> 00:22:41,400
in designing a system that fits
with the values that I am discussing here.

388
00:22:41,467 --> 00:22:45,233
Yeah, one thing
I wanted to ask Jenny about is,

389
00:22:45,300 --> 00:22:48,567
we are talking about a body that will be
made up of people from all over the world.

390
00:22:48,633 --> 00:22:50,133
Facebook's got users all over the world.

391
00:22:50,200 --> 00:22:52,767
And so, you have to have
members of an oversight board

392
00:22:52,833 --> 00:22:54,267
from all over the world.

393
00:22:54,333 --> 00:22:57,500
And then they will have a set of values
that are their common values

394
00:22:57,567 --> 00:23:00,000
that they will commit themselves
to upholding.

395
00:23:00,067 --> 00:23:01,067
And I'm wondering, Jenny...

396
00:23:01,133 --> 00:23:03,133
I remember after we clerked together
at the Supreme Court,

397
00:23:03,200 --> 00:23:05,533
you went off and clerked
for the International Court of Justice.

398
00:23:05,600 --> 00:23:08,300
And then you studied International law.

399
00:23:08,367 --> 00:23:13,367
How have bodies composed of people
from different places historically done?

400
00:23:13,433 --> 00:23:17,067
Have they done a good job when they
are supposed to converge on some values?

401
00:23:17,667 --> 00:23:19,767
I think
it's a really interesting question,

402
00:23:19,833 --> 00:23:21,833
and there have been
a bunch of international courts

403
00:23:21,900 --> 00:23:23,933
in the 20th and 21st centuries,

404
00:23:24,000 --> 00:23:25,900
and some have done
a better job than others

405
00:23:25,967 --> 00:23:30,233
in terms of how much convergence you have
between the different judges.

406
00:23:30,300 --> 00:23:34,600
But they do come from so many
different legal systems and cultures

407
00:23:34,667 --> 00:23:38,000
and sort of getting cohesion
in a set of people

408
00:23:38,067 --> 00:23:41,367
from different countries can be tough.

409
00:23:41,433 --> 00:23:43,500
Even just in the sort of the way

410
00:23:43,567 --> 00:23:46,000
they approach a legal problem
and analyzing it,

411
00:23:46,067 --> 00:23:48,200
you can find
that judges from different places

412
00:23:48,267 --> 00:23:50,667
have very different approaches.

413
00:23:50,733 --> 00:23:51,867
And so, I think...

414
00:23:51,933 --> 00:23:54,000
Some oversight board members
won't all be lawyers.

415
00:23:54,067 --> 00:23:55,367
They won't even have in common

416
00:23:55,433 --> 00:23:57,000
that they all went
to law school somewhere.

417
00:23:57,067 --> 00:23:58,867
Yeah, I think it'll be a huge challenge

418
00:23:58,933 --> 00:24:04,000
in building some sort of cohesion
and a sense of common mission.

419
00:24:04,067 --> 00:24:06,667
One thing that international courts do

420
00:24:06,733 --> 00:24:09,900
is try to come up with ways
of accommodating local difference

421
00:24:09,967 --> 00:24:13,400
with the floor of international standards.

422
00:24:13,467 --> 00:24:16,633
In the European Court of Human Rights,
for example,

423
00:24:16,700 --> 00:24:21,167
they have a sort of idea
called the "margin of appreciation,"

424
00:24:21,233 --> 00:24:22,367
they set a floor,

425
00:24:22,433 --> 00:24:25,067
all the countries in Europe have to have

426
00:24:25,133 --> 00:24:28,667
a minimum protection
for rights that is the floor.

427
00:24:28,733 --> 00:24:32,167
But in some countries
get a margin of appreciation

428
00:24:32,233 --> 00:24:36,300
of doing slightly more or slightly less
based on their local culture and needs.

429
00:24:36,367 --> 00:24:41,500
That's something that judges came up with
to try to accommodate those differences.

430
00:24:41,567 --> 00:24:43,900
But one of the things,
Mark, in your question,

431
00:24:43,967 --> 00:24:45,933
and I would love
to hear your thoughts about this,

432
00:24:46,000 --> 00:24:49,100
raised for me was,
we talked about the U.S.

433
00:24:49,167 --> 00:24:51,333
which has a very pro-free-speech approach.

434
00:24:51,400 --> 00:24:53,267
And Europe which does
a little more balancing

435
00:24:53,333 --> 00:24:55,533
on things like hate-speech or privacy.

436
00:24:55,600 --> 00:25:00,033
But there are many countries in the world
that are more authoritarian

437
00:25:00,100 --> 00:25:03,933
in terms of suppression of speech
and political speech.

438
00:25:04,000 --> 00:25:06,867
One of the things that
International Human Rights law tries to do

439
00:25:06,933 --> 00:25:10,567
is set a floor
and to push back against those countries

440
00:25:10,633 --> 00:25:13,900
that want to impose
greater restrictions on speech.

441
00:25:13,967 --> 00:25:15,467
You have a policy,

442
00:25:15,533 --> 00:25:18,367
and I understand the review board
will follow the policy

443
00:25:18,433 --> 00:25:20,033
of complying with local law.

444
00:25:20,100 --> 00:25:25,500
But I wonder, in terms of the value
of the tilt that you've described,

445
00:25:25,567 --> 00:25:28,600
towards voices wanting to suppress speech,

446
00:25:28,667 --> 00:25:32,967
how do you balance that accommodation
and respect for local law

447
00:25:33,033 --> 00:25:35,433
with the desire to protect free expression

448
00:25:35,500 --> 00:25:39,800
when you have a country that is engaged
in censorship that is problematic?

449
00:25:39,867 --> 00:25:45,700
This is a set of controversial decisions

450
00:25:45,767 --> 00:25:47,867
that we spent
a lot of time thinking about.

451
00:25:47,933 --> 00:25:52,867
It goes back to this principle of trying
to give people the most voice we can.

452
00:25:52,933 --> 00:25:57,167
And this idea that giving people a voice
is something you can expand on,

453
00:25:57,233 --> 00:26:00,033
if you give people better technical tools,

454
00:26:00,100 --> 00:26:03,667
if you give people the ability
to get their ideas and experience out,

455
00:26:03,733 --> 00:26:05,433
you are expanding people's voice.

456
00:26:05,500 --> 00:26:08,833
If you push on the boundaries of what
people are allowed to communicate,

457
00:26:08,900 --> 00:26:10,167
you're expanding people's voice.

458
00:26:11,533 --> 00:26:12,900
In some of these cases,

459
00:26:12,967 --> 00:26:16,367
there are certainly laws in countries
that I disagree with.

460
00:26:16,433 --> 00:26:22,500
That I wish weren't there
and I wish we did not have to follow.

461
00:26:24,333 --> 00:26:31,100
But if the trade-off
is having Facebook not available

462
00:26:31,167 --> 00:26:35,300
so that 99.9% of people in the country

463
00:26:35,367 --> 00:26:39,067
who are using these tools
to express things on a day-to-day basis

464
00:26:39,133 --> 00:26:44,100
that are important to them,
important to their social lives,

465
00:26:44,167 --> 00:26:45,667
in a lot of cases,

466
00:26:45,733 --> 00:26:48,333
for the tens of millions of people
who run small businesses,

467
00:26:48,400 --> 00:26:51,567
it's important to their well-being
and the local economy.

468
00:26:51,633 --> 00:26:55,467
To say that we are going
to deprive those people of the tool

469
00:26:55,533 --> 00:27:01,067
because 0.01% of the people in a country

470
00:27:01,133 --> 00:27:05,333
are trying to express something
that a local law forbids,

471
00:27:05,400 --> 00:27:08,433
doesn't feel like the right calculation.

472
00:27:08,500 --> 00:27:13,000
Even though I understand it.

473
00:27:14,900 --> 00:27:16,867
Even if we disagree with the law.

474
00:27:16,933 --> 00:27:20,667
So we've made the decision that,
in most of these cases,

475
00:27:20,733 --> 00:27:24,467
we are going
to follow local laws on speech.

476
00:27:24,533 --> 00:27:26,833
That doesn't mean that...

477
00:27:26,900 --> 00:27:30,967
There are lines here
that I think are incredibly important.

478
00:27:30,967 --> 00:27:35,400
you do have this question all the time

479
00:27:35,467 --> 00:27:38,867
where countries are trying to impose
their local standards

480
00:27:38,933 --> 00:27:40,667
on people outside of their country.

481
00:27:41,267 --> 00:27:43,500
That's obviously not going to be
something that we go for.

482
00:27:43,567 --> 00:27:48,467
At the same time,
we're very sensitive on the privacy side,

483
00:27:48,533 --> 00:27:52,967
not to store people's data in countries

484
00:27:53,033 --> 00:27:56,033
that we think aren't going to respect
people's human rights.

485
00:27:56,867 --> 00:28:00,100
And that's a little bit different
from a question around censorship.

486
00:28:00,167 --> 00:28:03,500
But the notion there is,
"Okay, if people can't express

487
00:28:03,567 --> 00:28:04,933
"every single thing they would want

488
00:28:05,000 --> 00:28:08,367
"because a local, repressive government
says that you can't,

489
00:28:08,433 --> 00:28:11,367
"then that's limiting your voice
and that's a problem."

490
00:28:11,433 --> 00:28:15,000
But putting data in a place
could really cause real-world harm,

491
00:28:15,067 --> 00:28:16,733
where the government goes after people

492
00:28:16,800 --> 00:28:19,633
and imprisons them or hurts them
or their family.

493
00:28:19,700 --> 00:28:24,200
So, that's a clear line
that we don't cross,

494
00:28:24,267 --> 00:28:27,333
and that I would encourage
more folks in the industry

495
00:28:27,400 --> 00:28:31,333
to kind of uphold as a red line as well.

496
00:28:31,400 --> 00:28:32,900
In fact, it's something...
Yeah, I'm sorry.

497
00:28:32,967 --> 00:28:35,033
I was just going to ask
one follow-up question, which is,

498
00:28:35,100 --> 00:28:38,400
that all makes sense in terms
of preserving the access of people

499
00:28:38,467 --> 00:28:42,333
to the service, when otherwise,
the country might block you entirely.

500
00:28:42,400 --> 00:28:45,100
But with this new content review board,

501
00:28:45,167 --> 00:28:47,933
even if you say
you're going to comply with local law,

502
00:28:48,000 --> 00:28:51,400
at the end, why not let the board say,

503
00:28:51,467 --> 00:28:54,800
"Okay, this decision, you're gonna do it
because you comply with local law,

504
00:28:54,867 --> 00:28:57,533
"but in an advisory opinion kind of way,

505
00:28:57,600 --> 00:29:00,833
"we think this particular
censorship decision

506
00:29:00,900 --> 00:29:04,800
"violates Facebook's values
or International Human Rights standards.

507
00:29:04,867 --> 00:29:07,267
"We understand the company is gonna
go along with the local law,

508
00:29:07,333 --> 00:29:11,767
"but we think it's a sort of
inappropriate suppression of speech."

509
00:29:11,833 --> 00:29:13,600
I would be supportive of that.

510
00:29:13,667 --> 00:29:15,767
Yeah, I think it's a great idea.

511
00:29:15,833 --> 00:29:17,700
I think we do anticipate exactly that.
Yeah.

512
00:29:17,767 --> 00:29:23,133
And one of the fundamental properties
of the board

513
00:29:23,200 --> 00:29:26,433
is going to be reason-giving.

514
00:29:26,500 --> 00:29:29,167
Right, so it's not just that
it's a group of people

515
00:29:29,233 --> 00:29:30,533
that go off and make a decision

516
00:29:30,600 --> 00:29:33,867
and then the decision is done,
it's opaque.

517
00:29:33,933 --> 00:29:37,900
A lot of the point is to build up

518
00:29:37,967 --> 00:29:41,633
almost a case-law analog

519
00:29:41,700 --> 00:29:47,433
in precedence and rationale
for why certain decisions are being made

520
00:29:48,167 --> 00:29:53,667
that, hopefully, over time,
will not just influence our systems here

521
00:29:53,733 --> 00:29:56,067
and the different services that we run,

522
00:29:56,133 --> 00:29:58,433
but the way people think about this
across the industry.

523
00:29:58,500 --> 00:30:00,867
We'll start narrow
right in terms of the scope.

524
00:30:00,933 --> 00:30:03,600
But, I mean, over time,
if this works well,

525
00:30:03,667 --> 00:30:06,833
you could see this expanding
to be something

526
00:30:06,900 --> 00:30:10,667
where other companies
directly want to use a board like this

527
00:30:10,733 --> 00:30:13,500
to help adjudicate and deal with appeals

528
00:30:13,567 --> 00:30:17,233
on some of their most complex issues
that they face as well.

529
00:30:18,233 --> 00:30:22,633
And hopefully, the work
that gets done here is serious thinking

530
00:30:22,700 --> 00:30:28,100
that can also, over time,
influence the way that other scholars

531
00:30:28,167 --> 00:30:31,767
and folks who are putting
laws and regulations in place

532
00:30:31,833 --> 00:30:34,133
in different countries around the world,
think about the way

533
00:30:34,200 --> 00:30:36,400
that expression should work
on the Internet as well.

534
00:30:36,467 --> 00:30:38,333
If I could say
something optimistic about this.

535
00:30:38,400 --> 00:30:40,067
I mean, in that scenario
that Mark describes,

536
00:30:40,133 --> 00:30:42,633
where the board does a good job
and it issues opinions,

537
00:30:42,700 --> 00:30:43,933
and people listen to them,

538
00:30:44,000 --> 00:30:47,400
I think courts in a lot of countries
will take that seriously.

539
00:30:47,467 --> 00:30:50,600
You know, the same way that courts,
you know, sometimes listen

540
00:30:50,667 --> 00:30:52,933
to the opinions of courts
from different countries.

541
00:30:53,000 --> 00:30:54,933
It doesn't mean that
they're forced to follow them, obviously.

542
00:30:55,000 --> 00:30:56,333
Just means they think
"Oh, that's interesting,

543
00:30:56,400 --> 00:30:59,200
"that's a reasonable way
of going about the particular problem."

544
00:30:59,267 --> 00:31:02,067
And I think, especially in countries

545
00:31:02,133 --> 00:31:04,500
where the legal pressure
they're putting on Facebook

546
00:31:04,567 --> 00:31:07,967
is sort of at the boundary
of what their own law actually requires.

547
00:31:08,033 --> 00:31:10,467
Sometimes countries say, "You have
to take this down, it's illegal."

548
00:31:10,533 --> 00:31:13,600
And then, Facebook can push back and say,
"It's not really illegal under your laws,

549
00:31:13,667 --> 00:31:15,400
"you just don't really like it,
in this case."

550
00:31:15,467 --> 00:31:17,467
The board could help in that effort

551
00:31:17,533 --> 00:31:20,400
by saying, "Well, you know,
we have done the balancing.

552
00:31:20,467 --> 00:31:23,633
"We've made a thoughtful judgment.
That should influence you."

553
00:31:23,700 --> 00:31:25,533
Again, that's an optimistic picture.

554
00:31:27,600 --> 00:31:30,567
Yeah. I mean,
that's an interesting question

555
00:31:30,633 --> 00:31:33,333
about how that set of things plays out

556
00:31:33,400 --> 00:31:36,833
around the handling of, kind of,

557
00:31:36,900 --> 00:31:39,700
incoming requests
from different governments.

558
00:31:39,767 --> 00:31:41,567
It's certainly not why
we're setting it up,

559
00:31:41,633 --> 00:31:44,633
but I think it'll have
an interesting effect over time.

560
00:31:44,700 --> 00:31:46,333
Yeah. And especially
when you're starting out,

561
00:31:46,400 --> 00:31:50,033
to the extent that you have a commitment
to follow the law of local countries,

562
00:31:50,100 --> 00:31:51,800
you know, if your legal department says,

563
00:31:51,867 --> 00:31:54,267
"Well, this violates the law
of the local country",

564
00:31:54,333 --> 00:31:56,567
I don't think it makes sense
to send it to the board,

565
00:31:56,633 --> 00:31:58,267
and put the board in a position
of having to say,

566
00:31:58,333 --> 00:32:01,567
"Well, if we were deciding this..."
You know, but you could do that.

567
00:32:01,633 --> 00:32:03,767
You could have sort of
the effectively advice

568
00:32:03,833 --> 00:32:05,467
from the body under those circumstances.

569
00:32:05,533 --> 00:32:07,633
-Yeah. Chose to. Yeah.
-Yeah.

570
00:32:07,700 --> 00:32:13,667
But going back to this question
about how to manage a global platform

571
00:32:13,733 --> 00:32:18,100
with the needs of what are very different
cultures in different places,

572
00:32:18,167 --> 00:32:22,400
I mean, this has been one of the points
that the report that we published

573
00:32:22,467 --> 00:32:25,367
highlights in terms of different tensions
and design decisions.

574
00:32:26,300 --> 00:32:31,200
I'm curious how you would think about this
in terms of how to have a group

575
00:32:31,267 --> 00:32:33,867
that has sufficient cohesion,

576
00:32:35,967 --> 00:32:39,733
that it can help make these decisions
for the platform around the world

577
00:32:39,800 --> 00:32:43,767
where deference needs to be given
to local cultures

578
00:32:43,833 --> 00:32:46,367
and how you would
just think about that overall.

579
00:32:46,433 --> 00:32:50,300
I mean, for me, because Facebook
is committed to universal values

580
00:32:50,367 --> 00:32:53,367
and is also committed
to having community standards

581
00:32:53,433 --> 00:32:56,500
on its platform
that are the same everywhere,

582
00:32:56,567 --> 00:32:59,233
what you want
is the international participants

583
00:32:59,300 --> 00:33:01,000
to give you local understanding.

584
00:33:01,067 --> 00:33:02,833
To give you cultural knowledge.

585
00:33:02,900 --> 00:33:08,167
So, if you are trying to figure out
if a particular post is ironic and funny,

586
00:33:08,233 --> 00:33:11,000
or is, in fact, you know,
intended to do harm to people,

587
00:33:11,067 --> 00:33:15,100
you need somebody with a sophistication
to know what people think in the culture.

588
00:33:15,167 --> 00:33:17,767
And you can rely on experts for that
up to a certain point.

589
00:33:17,833 --> 00:33:19,967
But there's no substitute
for having someone on the board

590
00:33:20,033 --> 00:33:21,467
who actually can help
explain those things.

591
00:33:21,533 --> 00:33:23,633
I think that's hugely valuable.

592
00:33:23,700 --> 00:33:26,367
But then the body also has to be aware
that they're going to be some things.

593
00:33:26,433 --> 00:33:28,533
Like, nudity policies is a great example.

594
00:33:28,600 --> 00:33:30,700
Where, you know, there's a huge range

595
00:33:30,767 --> 00:33:34,900
of different cultural choices
all over the world on nudity.

596
00:33:34,967 --> 00:33:37,400
You know, in some cultures,
it's normal to wear very few clothes,

597
00:33:37,467 --> 00:33:39,900
and in some cultures, it's normal
to wear a whole lot of clothes.

598
00:33:39,967 --> 00:33:42,667
And I think if you're going to have,
as Facebook does have,

599
00:33:42,733 --> 00:33:45,000
a single standard on nudity,

600
00:33:45,067 --> 00:33:48,000
then the body has to be made up,
the oversight board has to be made up

601
00:33:48,067 --> 00:33:50,867
of people who say, "We get it.
We get that there's cultural difference,

602
00:33:50,933 --> 00:33:54,100
"but we also get that
this is a platform that operates globally,

603
00:33:54,167 --> 00:33:58,200
"and we've had to converge on something
that more or less works acknowledging

604
00:33:58,267 --> 00:34:00,933
"that it's not a perfect fit
for the south of France,

605
00:34:01,000 --> 00:34:03,000
"and it's not a perfect fit
for Saudi Arabia."

606
00:34:04,967 --> 00:34:08,033
I think one challenge, in respect to that,

607
00:34:08,100 --> 00:34:10,900
is figuring out, sort of,
what's the optimal number of people

608
00:34:10,967 --> 00:34:13,633
to cover the geographic diversity
of the world

609
00:34:13,700 --> 00:34:17,600
while still having cohesion
among the people who are on the board,

610
00:34:17,667 --> 00:34:20,667
so that there's consistency in approach
and a kind of understanding.

611
00:34:20,733 --> 00:34:25,400
I think that to get that, you can't
realistically represent all the countries

612
00:34:25,467 --> 00:34:29,800
or even significant portions of it
in a manageable-sized group.

613
00:34:29,867 --> 00:34:32,333
So you have to have some way
of picking people

614
00:34:32,400 --> 00:34:35,933
who are going to be aware
of their own cultural blind spots,

615
00:34:36,000 --> 00:34:40,333
and then also have the ability
to talk to experts,

616
00:34:40,400 --> 00:34:45,500
to talk to people from a particular
geographic space to get more information.

617
00:34:45,567 --> 00:34:48,333
But I think that's part of the struggle
you will have

618
00:34:48,400 --> 00:34:52,033
in trying to find
the right number of people.

619
00:34:52,100 --> 00:34:55,800
In our conversations,
we've been hovering around 40,

620
00:34:55,867 --> 00:34:58,900
with the goal of
trying to maximize participation

621
00:34:58,967 --> 00:35:00,067
from different parts in the world,

622
00:35:00,133 --> 00:35:02,033
while also having a body
that can still function.

623
00:35:02,100 --> 00:35:05,233
You know, probably the 40 aren't going to
sit on any one decision together,

624
00:35:05,300 --> 00:35:07,000
-they'll have to sit in sub-panels.
-Yeah.

625
00:35:07,067 --> 00:35:11,367
But that seems like ballpark,
to me, roughly the right number.

626
00:35:11,900 --> 00:35:13,733
Yeah, that's something
the report discusses.

627
00:35:13,800 --> 00:35:15,767
It's basically the range...

628
00:35:15,833 --> 00:35:19,000
I think there is the discussion about,
"Should it be closer to 20?

629
00:35:19,067 --> 00:35:20,600
-"Should it be closer to 100?"
-Yeah.

630
00:35:20,667 --> 00:35:22,567
There are different pros and cons
around cohesion,

631
00:35:22,633 --> 00:35:26,067
the group ability to handle the workload.

632
00:35:27,233 --> 00:35:29,333
-If the workload gets too big...
-Yeah.

633
00:35:29,400 --> 00:35:31,433
It's already going to function in this way

634
00:35:31,500 --> 00:35:35,400
where if you are in the community
and you want to appeal a decision,

635
00:35:35,467 --> 00:35:38,133
your first appeal goes to
a set of appealers,

636
00:35:38,200 --> 00:35:43,767
people who can look at those appeals
inside Facebook.

637
00:35:43,833 --> 00:35:46,767
But then, afterwards, if you still
aren't satisfied with that,

638
00:35:46,833 --> 00:35:49,333
you'll be able to appeal
to this independent board,

639
00:35:49,400 --> 00:35:52,833
but it will have to choose
which cases it takes.

640
00:35:52,900 --> 00:35:54,300
And there's a question of

641
00:35:54,367 --> 00:35:59,633
do we want it to handle 100 cases a year
or 50 or 1,000...

642
00:35:59,700 --> 00:36:02,667
And that's an important piece
to this, as well.

643
00:36:03,933 --> 00:36:07,400
The other question
that I think is a really big deal around

644
00:36:07,467 --> 00:36:10,300
how we set this up in terms of appointment

645
00:36:10,367 --> 00:36:13,267
is not just about the size,
but how do we do the appointment?

646
00:36:13,333 --> 00:36:16,233
To get to this question around
legitimacy and independence,

647
00:36:16,300 --> 00:36:21,000
which is, of course,
we achieve some independence

648
00:36:21,067 --> 00:36:23,733
even if Facebook picks the people.

649
00:36:23,800 --> 00:36:27,033
But then they are independent
and they are not paid by us,

650
00:36:27,100 --> 00:36:29,667
and the decisions are binding,

651
00:36:29,733 --> 00:36:32,400
but we don't have the ability
to remove them.

652
00:36:33,933 --> 00:36:36,033
That achieves a level of independence,

653
00:36:36,100 --> 00:36:40,433
but I don't know if that would be
as much as we would want,

654
00:36:40,500 --> 00:36:44,633
to really maximize the legitimacy
and independence of the board.

655
00:36:44,700 --> 00:36:47,967
So, I'm curious how you think about that

656
00:36:48,033 --> 00:36:51,800
and the trade-off space there,
between how this can go.

657
00:36:53,200 --> 00:36:55,200
I think people agree
that it would be better

658
00:36:55,267 --> 00:36:58,233
if we weren't appointing these people
all ourselves.

659
00:36:58,300 --> 00:37:02,833
But it's been hard to come to consensus
on any other approach

660
00:37:02,900 --> 00:37:05,700
that would make a whole lot of more sense.

661
00:37:05,767 --> 00:37:07,000
I mean, clearly, I think people don't want

662
00:37:07,067 --> 00:37:11,133
governments to be appointing the people,
for example, in a lot of countries.

663
00:37:11,200 --> 00:37:12,867
But I'm curious, what balance,

664
00:37:12,933 --> 00:37:15,100
where do you think
we should end up on this?

665
00:37:15,933 --> 00:37:19,200
Well, one mechanism you sometimes see
in international bodies

666
00:37:19,267 --> 00:37:25,933
is to have one step in between
selecting the selectors.

667
00:37:26,000 --> 00:37:29,133
So, for the International Court
of Justice, for example,

668
00:37:29,200 --> 00:37:32,633
states appoint representatives
to the permanent court of arbitration,

669
00:37:32,700 --> 00:37:34,133
which is a different body,

670
00:37:34,200 --> 00:37:37,700
and then, those people then suggest
candidates for the ICJ,

671
00:37:37,767 --> 00:37:39,033
in addition to other things they do

672
00:37:39,100 --> 00:37:42,067
and then, in that case,
it's the UN General Assembly

673
00:37:42,133 --> 00:37:44,567
and Security Council
that make the final decision.

674
00:37:44,633 --> 00:37:47,833
But there's a step
between the interested parties,

675
00:37:47,900 --> 00:37:48,900
that is the states,

676
00:37:48,967 --> 00:37:52,633
and then the actual people who are chosen.

677
00:37:52,700 --> 00:37:56,233
So one possibility is,
sort of, selecting people

678
00:37:56,300 --> 00:37:59,867
who then go on and select
the actual members of the board.

679
00:37:59,933 --> 00:38:01,600
So there's a little bit of remove.

680
00:38:01,667 --> 00:38:04,200
I've really struggled with this
in all of our conversations.

681
00:38:04,267 --> 00:38:06,100
I mean, I like Jenny's idea,

682
00:38:06,167 --> 00:38:08,167
and I like the idea
of some intervening entity.

683
00:38:08,233 --> 00:38:10,733
But I think that from
a public perspective, people will say,

684
00:38:10,800 --> 00:38:12,900
"Wait, you picked the people
who picked the people,

685
00:38:12,967 --> 00:38:14,967
"so how did you choose
that first set of people?"

686
00:38:15,033 --> 00:38:18,200
I think we get very quickly
into a regress problem.

687
00:38:18,267 --> 00:38:22,867
And so, maybe there's a hybrid solution
where we can choose some of the people,

688
00:38:22,933 --> 00:38:26,400
and then those people could participate
alongside Facebook

689
00:38:26,467 --> 00:38:29,033
and external input on choosing
the next set of people.

690
00:38:29,100 --> 00:38:33,000
It might be possible to arrange
a kind of middle ground solution there

691
00:38:33,067 --> 00:38:35,067
without making the process so complicated.

692
00:38:35,133 --> 00:38:37,400
I mean, one thing about
these big international entities

693
00:38:37,467 --> 00:38:39,067
is that they do a lot of other stuff.

694
00:38:39,133 --> 00:38:41,667
The people who do the picking,
that's not the only thing they do.

695
00:38:41,733 --> 00:38:42,767
Yeah.

696
00:38:42,833 --> 00:38:44,267
And that would not necessarily
be the case here

697
00:38:44,333 --> 00:38:46,933
because we are not working against
the backdrop of a big infrastructure

698
00:38:47,000 --> 00:38:49,367
since this is an experiment
that you are doing.

699
00:38:50,067 --> 00:38:51,233
Yeah.

700
00:38:51,300 --> 00:38:54,300
I think somewhere in the space
of what you're saying, though,

701
00:38:54,367 --> 00:38:59,233
of having some selection group

702
00:38:59,300 --> 00:39:02,533
that we'll arrive at
through our consultative process overall,

703
00:39:02,600 --> 00:39:06,633
so if Facebook blesses that initial group,

704
00:39:06,700 --> 00:39:08,600
it won't be something that

705
00:39:08,667 --> 00:39:11,900
we just sit in a room
and come up with by ourselves.

706
00:39:11,967 --> 00:39:15,200
Whether that group ends up
being on the board,

707
00:39:15,267 --> 00:39:17,400
as well as picking the other people,

708
00:39:17,467 --> 00:39:21,367
or ends up
just selecting the board by itself...

709
00:39:22,533 --> 00:39:23,867
There are pros and cons to those,

710
00:39:23,933 --> 00:39:26,800
but I think something in that space is
probably what makes the most sense to me,

711
00:39:26,867 --> 00:39:29,200
about where we will want to
end up on this.

712
00:39:29,267 --> 00:39:33,300
And in the end,
people are going to see this body

713
00:39:33,367 --> 00:39:34,733
as an independent decision-making body

714
00:39:34,800 --> 00:39:36,233
when it makes independent decisions.

715
00:39:36,300 --> 00:39:39,133
I really believe
that its public legitimacy will come

716
00:39:39,200 --> 00:39:41,367
from actually doing a good job

717
00:39:41,433 --> 00:39:44,633
rather than
purely from the initial mode of selection.

718
00:39:44,700 --> 00:39:45,767
Yeah.

719
00:39:45,833 --> 00:39:49,867
I think another big challenge that
you will have that will make or break this

720
00:39:49,933 --> 00:39:53,900
is how you manage
the volume of potential cases

721
00:39:53,967 --> 00:39:55,900
and the case selection process.

722
00:39:55,967 --> 00:39:58,367
Because I think if people see

723
00:39:58,433 --> 00:40:03,300
either because of the volume of content
that there could be to review,

724
00:40:03,367 --> 00:40:06,533
that you are not reviewing enough of it
or the right ones,

725
00:40:06,600 --> 00:40:10,100
or the timing is too slow,
where people think that,

726
00:40:10,167 --> 00:40:12,333
"This is content
that we are really worried about now,

727
00:40:12,400 --> 00:40:13,967
"but by the the time
this board gets to it,

728
00:40:14,033 --> 00:40:17,667
"it's months down the road,
the ship has already sailed.

729
00:40:17,733 --> 00:40:19,767
"The riots have happened in this city,

730
00:40:19,833 --> 00:40:22,067
"or whatever harm has occurred,
has happened."

731
00:40:22,133 --> 00:40:25,733
So, I think that's a huge challenge
that it will make or break this

732
00:40:25,800 --> 00:40:29,400
in terms of whether people view it as
legitimate and actually doing something.

733
00:40:29,467 --> 00:40:33,667
So I wonder how you think about,
both of you, the volume of cases

734
00:40:33,733 --> 00:40:37,700
and how you can possibly both resolve
things expeditiously enough,

735
00:40:37,767 --> 00:40:41,800
that the public thinks this is actually
weighing in at a time when it's useful,

736
00:40:41,867 --> 00:40:43,700
and then, hitting the right cases

737
00:40:43,767 --> 00:40:46,100
out of this huge number of

738
00:40:46,167 --> 00:40:48,200
content-moderation decisions
you are making.

739
00:40:48,267 --> 00:40:49,333
Yeah.

740
00:40:49,400 --> 00:40:50,600
This is a really important point.

741
00:40:50,667 --> 00:40:53,767
I think we envision
two different tracks.

742
00:40:53,833 --> 00:40:57,633
So there's one set of things

743
00:40:57,700 --> 00:41:01,500
that are more deliberative,
precedent-setting,

744
00:41:01,567 --> 00:41:06,767
longer-term decisions
where the importance is less about,

745
00:41:08,167 --> 00:41:11,033
"Okay, this specific case,
did we take the piece of content down

746
00:41:11,100 --> 00:41:12,733
"or leave it up today?"

747
00:41:12,800 --> 00:41:17,000
But more about influencing
the long-term direction of the policies

748
00:41:17,067 --> 00:41:19,233
and the way that the platform evolves.

749
00:41:19,300 --> 00:41:23,133
There's another set of issues
which very much are urgent.

750
00:41:23,200 --> 00:41:27,133
And there, we need a mechanism
to basically be able to refer issues

751
00:41:27,200 --> 00:41:29,533
that we see to the board

752
00:41:29,600 --> 00:41:32,333
and have a relatively rapid turnaround

753
00:41:32,400 --> 00:41:36,333
that could affect whether
something could have an imminent impact.

754
00:41:36,400 --> 00:41:40,167
And that, I think, also reflects how
the issues will get raised to the board.

755
00:41:40,233 --> 00:41:43,267
On one hand, there will be
an appeals process

756
00:41:43,333 --> 00:41:45,800
where anyone in the community
will be able to get an appeal

757
00:41:45,867 --> 00:41:47,900
which will first go to Facebook.

758
00:41:47,967 --> 00:41:49,267
And then, a second appeal,

759
00:41:49,333 --> 00:41:51,067
they will be able
to at least raise it to the board.

760
00:41:51,133 --> 00:41:53,167
And then the board
can consider hearing it.

761
00:41:54,300 --> 00:41:56,567
And I imagine that there will also be
a staff for the board.

762
00:41:56,633 --> 00:41:58,167
There will be a permanent staff to help

763
00:41:58,233 --> 00:42:01,333
review and prioritize
some of this, as well.

764
00:42:01,400 --> 00:42:04,400
And at the same time,
I think there also needs to be a track

765
00:42:04,467 --> 00:42:07,200
where Facebook, and our systems directly,

766
00:42:07,267 --> 00:42:12,000
can say, "Hey, we see that this piece
of content is about to go viral.

767
00:42:12,067 --> 00:42:16,767
"And we are making a decision
on how we are going to handle this

768
00:42:16,833 --> 00:42:18,433
"because that is our responsibility."

769
00:42:18,500 --> 00:42:22,567
But we also want to make sure,
we think this is a very important matter,

770
00:42:22,633 --> 00:42:26,233
so we want to refer this to the board
and put it on your radar now

771
00:42:26,300 --> 00:42:28,600
that you want to make a decision
on this now

772
00:42:28,667 --> 00:42:31,033
and rule one whether
our handling on this was right.

773
00:42:31,100 --> 00:42:34,433
And that will create
additional accountability

774
00:42:34,500 --> 00:42:38,767
and oversight of what we do internally,
which I think is important.

775
00:42:38,833 --> 00:42:43,167
It's also a way for us
to execute on our responsibility

776
00:42:43,233 --> 00:42:45,467
for managing the content
as well as possible.

777
00:42:45,533 --> 00:42:49,000
It's not that we are ever going to be
out of the business

778
00:42:49,067 --> 00:42:51,967
of having to make these decisions
ourselves, internally.

779
00:42:52,033 --> 00:42:55,933
But we just think that it's so important
that these decisions get made well,

780
00:42:56,000 --> 00:42:59,367
that in addition to trying to build out
robust internal systems,

781
00:42:59,433 --> 00:43:01,967
we are also going to build out
this whole independent system

782
00:43:02,033 --> 00:43:05,067
for oversight and accountability
of the way that we are making those,

783
00:43:05,133 --> 00:43:08,100
and that we can
refer stuff to it ourselves,

784
00:43:08,167 --> 00:43:11,200
in addition to the way
that the community can appeal it.

785
00:43:11,267 --> 00:43:14,400
And just one point that
very much builds on what Mark was saying,

786
00:43:14,467 --> 00:43:18,700
one huge advantage of reason-giving
of explanation for the decision

787
00:43:18,767 --> 00:43:21,333
is that it drives you in life, if you have
to give a reason for what you do,

788
00:43:21,400 --> 00:43:23,900
to think it through clearly
and to state it clearly.

789
00:43:23,967 --> 00:43:27,133
And especially for high-pressure,
short-run decisions,

790
00:43:27,200 --> 00:43:30,733
we are all human, it's almost impossible
to make those decisions perfectly.

791
00:43:30,800 --> 00:43:34,500
But if you know that there's going to be
a review of that decision,

792
00:43:34,567 --> 00:43:36,600
you're going to have to come up
with a good reason for it.

793
00:43:36,667 --> 00:43:38,433
And then,
when the body is doing the review,

794
00:43:38,500 --> 00:43:39,767
it's going to give clear reasons for it.

795
00:43:39,833 --> 00:43:41,300
And then, I think, for the general public

796
00:43:41,367 --> 00:43:43,600
that's trying to figure out
was this the right decision,

797
00:43:43,667 --> 00:43:46,067
they will have a framework
for making their own judgment.

798
00:43:46,133 --> 00:43:47,833
Instead of just instinctively thinking,

799
00:43:47,900 --> 00:43:49,933
"Well, if Facebook did this,
it must've been the right decision,"

800
00:43:50,000 --> 00:43:52,867
or, "If Facebook did this,
it must've been the wrong decision."

801
00:43:52,933 --> 00:43:53,967
Yeah.

802
00:43:54,833 --> 00:43:57,867
One interesting thing is that you have,
at your hands,

803
00:43:57,933 --> 00:43:59,700
many more tools than governments have.

804
00:43:59,767 --> 00:44:02,833
Governments have, sort of,
an on-off switch about speech.

805
00:44:02,900 --> 00:44:06,667
They can punish people for speech,
they can order you to take it down.

806
00:44:06,733 --> 00:44:07,967
But you have more control

807
00:44:08,033 --> 00:44:11,767
in terms of how many people see
particular kinds of speech.

808
00:44:11,833 --> 00:44:17,333
And so, that is very wrapped up
in the algorithms

809
00:44:17,400 --> 00:44:19,633
and the way that
you've designed your product.

810
00:44:19,700 --> 00:44:22,667
But I wonder how you think
about the question

811
00:44:22,733 --> 00:44:25,400
of the on/off switch for speech

812
00:44:25,467 --> 00:44:31,767
versus that more textured approach
people can get at the speech,

813
00:44:31,833 --> 00:44:34,267
but should it be pushed,
should it be reaching,

814
00:44:34,333 --> 00:44:36,000
should it be allowed to go viral.

815
00:44:36,467 --> 00:44:38,567
How, if at all, would this board

816
00:44:38,633 --> 00:44:40,533
eventually look
at those kind of questions?

817
00:44:41,067 --> 00:44:44,933
Yeah, so let's talk about
the V1 of it first,

818
00:44:45,000 --> 00:44:47,233
and then, over time...

819
00:44:48,633 --> 00:44:52,133
You know, I think
this is such an ambitious

820
00:44:52,200 --> 00:44:54,700
and unusual project, in general,

821
00:44:54,767 --> 00:44:57,167
for a company to take on.

822
00:44:57,233 --> 00:45:00,333
One of the things I've tried
to be careful about

823
00:45:00,400 --> 00:45:03,500
is making sure that the scope
is clear in the beginning,

824
00:45:03,600 --> 00:45:06,167
so that way, it doesn't just collapse
under its own weight.

825
00:45:06,233 --> 00:45:08,133
So, my hope... I have a lot of hope

826
00:45:08,200 --> 00:45:10,967
that this can expand over time
and cover more scope,

827
00:45:11,567 --> 00:45:13,733
can include more focus in the industry,

828
00:45:13,800 --> 00:45:15,733
can become a part of a broader process,

829
00:45:15,800 --> 00:45:17,867
but I also just want to be careful

830
00:45:17,933 --> 00:45:21,033
about not trying to do too much
at once such that...

831
00:45:21,500 --> 00:45:25,433
I think it's going to be hard to pull off
what we're trying to, here.

832
00:45:27,167 --> 00:45:33,033
But that said, I do think the point
that you're raising is an important one.

833
00:45:33,100 --> 00:45:36,567
It's not just about, "Yes/no.
Is this on the platform or not?"

834
00:45:36,633 --> 00:45:37,833
It's not...

835
00:45:37,900 --> 00:45:39,767
There aren't an infinite set of options.

836
00:45:39,833 --> 00:45:46,033
I mean, the content generally falls
into a number of clear categories.

837
00:45:46,100 --> 00:45:48,133
I mean, the vast majority is,

838
00:45:48,200 --> 00:45:50,767
"Okay, this is fine. People can share it
as they want.

839
00:45:50,833 --> 00:45:54,133
"It can spread as people want."

840
00:45:54,633 --> 00:45:59,100
There's another category which is maybe
one step over from that

841
00:45:59,267 --> 00:46:01,367
which is...

842
00:46:01,433 --> 00:46:04,800
"Okay, maybe you're saying something
that's factually incorrect."

843
00:46:05,900 --> 00:46:08,200
We don't believe that

844
00:46:08,267 --> 00:46:10,867
you shouldn't be able to say things
that are wrong to your friends,

845
00:46:10,933 --> 00:46:16,067
but we don't want things
that are hoaxes to be going viral.

846
00:46:16,133 --> 00:46:20,800
There, we don't block the content
from being on our services,

847
00:46:20,867 --> 00:46:23,633
but we do reduce its distribution
so it doesn't go viral.

848
00:46:23,700 --> 00:46:25,633
We put up a warning that says,

849
00:46:25,700 --> 00:46:28,033
"Independent fact checkers
have said that this is false."

850
00:46:28,100 --> 00:46:29,833
That's kind of another discrete bucket.

851
00:46:29,900 --> 00:46:32,467
Then you get into another category

852
00:46:32,533 --> 00:46:34,500
where it's like,
"Okay, this content is bad.

853
00:46:34,567 --> 00:46:38,600
"It's harmful. It's going to cause
some imminent damage,

854
00:46:38,967 --> 00:46:40,900
"so we need to take this content down."

855
00:46:40,967 --> 00:46:43,567
And then you get into things
that are even more extreme of...

856
00:46:43,633 --> 00:46:48,033
"This content... Either you're
a repeated actor who's spreading

857
00:46:48,100 --> 00:46:50,400
"things that could be dangerous,

858
00:46:51,433 --> 00:46:53,833
"or what you're doing
is so harmful by itself,

859
00:46:53,900 --> 00:46:57,467
"that this is not only gonna have
an impact on this piece of content,

860
00:46:57,533 --> 00:46:59,900
"but on your standing
as a member of the community

861
00:46:59,967 --> 00:47:01,700
"and your ability to use the service,

862
00:47:01,767 --> 00:47:03,867
"either for some period of time
or permanently."

863
00:47:03,933 --> 00:47:06,533
And there are those categories,
and I do think

864
00:47:07,567 --> 00:47:11,133
the board will be able
to not only weigh in

865
00:47:11,200 --> 00:47:15,067
on the treatment
on a specific piece of content,

866
00:47:15,133 --> 00:47:18,400
but like any appeals process overall,

867
00:47:18,467 --> 00:47:22,167
we'll need to be able to influence
the underlying policies

868
00:47:22,233 --> 00:47:27,233
and how they get set,
and how they affect,

869
00:47:27,300 --> 00:47:30,833
how we think about which content
should be in which of those buckets

870
00:47:30,900 --> 00:47:32,500
and how these things should be treated.

871
00:47:32,567 --> 00:47:35,000
And I just think
that's an incredibly important part

872
00:47:35,067 --> 00:47:37,833
of how the system needs to work over time.

873
00:47:37,900 --> 00:47:39,667
And this is going to be
a year's long project

874
00:47:39,733 --> 00:47:43,067
of evolving it to increase the scope
to get there.

875
00:47:44,767 --> 00:47:47,533
And as Mark said,
"First, we gotta walk."

876
00:47:48,100 --> 00:47:49,767
But I think that's exactly right.

877
00:47:49,833 --> 00:47:51,967
One thing that makes it
so interesting and exciting

878
00:47:52,033 --> 00:47:58,733
is that there's a very different mode
of distributing speech on social media.

879
00:47:58,800 --> 00:48:02,800
And that creates
multiple avenues and mechanisms

880
00:48:02,867 --> 00:48:04,567
for learning things,

881
00:48:04,633 --> 00:48:07,767
and also multiple avenues
and mechanisms for harm,

882
00:48:07,833 --> 00:48:09,433
and last but not least,

883
00:48:09,500 --> 00:48:11,833
multiple avenues for solutions.

884
00:48:11,900 --> 00:48:13,867
So, it does seem like, over time,

885
00:48:13,933 --> 00:48:17,000
if this board can stand up and do its job
the way we want it to

886
00:48:17,067 --> 00:48:18,967
and develop public legitimacy,

887
00:48:19,033 --> 00:48:22,233
then we should be able to explore
all those things that Mark's discussing.

888
00:48:22,300 --> 00:48:24,333
Yeah, so I know we're a bit short on time,

889
00:48:24,400 --> 00:48:27,267
but there are two more topics
that I wanted to make sure we covered.

890
00:48:27,333 --> 00:48:29,167
So, one is about regulation more broadly.

891
00:48:29,233 --> 00:48:31,967
Today, most of
the conversation's been focused

892
00:48:32,033 --> 00:48:35,667
on how we would stand up
this independent process.

893
00:48:35,733 --> 00:48:38,967
It's still largely something
that we're initiating.

894
00:48:39,033 --> 00:48:43,800
We're doing a consultive process
with a lot of folks to get input on it,

895
00:48:44,767 --> 00:48:47,933
but it's not a...

896
00:48:48,433 --> 00:48:51,733
It's not a government-imposed regulation.

897
00:48:52,133 --> 00:48:57,033
And I think that a model like this
probably makes the most sense

898
00:48:57,100 --> 00:48:59,900
in countries,
especially like the United States,

899
00:48:59,967 --> 00:49:03,300
that have such a strong protection
under the First Amendment

900
00:49:03,367 --> 00:49:08,667
that there's unlikely to be
much significant content regulation.

901
00:49:09,233 --> 00:49:12,167
But I'm curious how you think about
how this plays out across the world,

902
00:49:12,233 --> 00:49:17,700
and where a model like this
can intersect with regulation,

903
00:49:17,767 --> 00:49:21,333
and where that can be productive,
or what we should look out for,

904
00:49:21,400 --> 00:49:25,933
or where we should be encouraging
more regulation

905
00:49:26,000 --> 00:49:28,400
that might be more effective
than this in certain places,

906
00:49:28,467 --> 00:49:30,767
or just how you think about that overall.

907
00:49:32,300 --> 00:49:34,367
Well, I think
it's an interesting question,

908
00:49:34,433 --> 00:49:35,667
because it's not clear

909
00:49:35,733 --> 00:49:37,933
that more government regulation here
would be useful.

910
00:49:38,000 --> 00:49:39,167
On the one hand,

911
00:49:39,233 --> 00:49:41,733
governments are more accountable
to the general public.

912
00:49:41,800 --> 00:49:45,367
But on the other hand, there's a danger
that regulation could tilt

913
00:49:45,433 --> 00:49:47,433
too much towards suppression of speech,

914
00:49:47,900 --> 00:49:52,600
and, also, that it could inhibit
competition among companies,

915
00:49:52,667 --> 00:49:56,133
because complying with regulations
can be expensive,

916
00:49:56,200 --> 00:49:59,433
and if governments put
really onerous regulations on,

917
00:49:59,500 --> 00:50:02,567
then that will make it harder
for other companies to enter,

918
00:50:02,633 --> 00:50:05,633
and that, in turn, could worsen
whatever problems there are

919
00:50:05,700 --> 00:50:09,233
in terms of people feeling that
there's too much concentrated power.

920
00:50:09,300 --> 00:50:13,033
So it's not clear that
government regulation would be better

921
00:50:13,100 --> 00:50:16,100
even though it would be more...
For at least some of the...

922
00:50:16,167 --> 00:50:18,600
Maybe for a lot of other things,
regulation would be useful,

923
00:50:18,667 --> 00:50:22,400
but on the sort of
most expression part of this,

924
00:50:22,467 --> 00:50:26,033
versus, you know, campaign speech
and finance disclosure,

925
00:50:26,100 --> 00:50:32,600
or privacy issues where regulation fits
more neatly into what might be desirable.

926
00:50:32,667 --> 00:50:36,033
I do think it's interesting,
internationally, to think about

927
00:50:36,100 --> 00:50:38,100
what industry and governments

928
00:50:38,167 --> 00:50:40,567
and civil society have done
in some other areas.

929
00:50:40,967 --> 00:50:42,267
In international law,

930
00:50:42,333 --> 00:50:45,067
people call it
multi-stakeholder initiatives.

931
00:50:45,133 --> 00:50:48,267
So, for example, with respect to diamonds

932
00:50:48,333 --> 00:50:50,600
being mined in areas
where there's armed conflict,

933
00:50:50,667 --> 00:50:52,533
there's something called
The Kimberley Process

934
00:50:52,600 --> 00:50:55,033
that involved governments,
it involved companies,

935
00:50:55,100 --> 00:50:58,367
it involved non-governmental organizations
and human rights groups

936
00:50:58,433 --> 00:51:00,567
to try to come up with some standards.

937
00:51:00,900 --> 00:51:03,733
And so, here, I do think there's
the potential in the long run.

938
00:51:03,800 --> 00:51:06,500
It's not a quick and dirty solution,

939
00:51:06,567 --> 00:51:09,800
but in the long run,
to have greater involvement

940
00:51:09,867 --> 00:51:12,867
from the industry as a whole

941
00:51:12,933 --> 00:51:15,667
from non-governmental organizations,

942
00:51:15,733 --> 00:51:18,600
inter-governmental organizations
like the UN,

943
00:51:18,833 --> 00:51:21,800
to have more global standard setting,

944
00:51:21,867 --> 00:51:23,500
even in a non-binding way

945
00:51:23,567 --> 00:51:27,467
where you have some kind of
international process to set standards

946
00:51:27,533 --> 00:51:31,200
that industry participates in,
that civil society groups participate in,

947
00:51:31,267 --> 00:51:33,300
and maybe they're not binding
at the end of the day,

948
00:51:33,367 --> 00:51:36,533
but that they represent
a kind of deliberative process

949
00:51:36,600 --> 00:51:38,400
involving more stake holders,

950
00:51:38,467 --> 00:51:41,933
and so that's something we've seen
internationally in a lot of other areas,

951
00:51:42,000 --> 00:51:44,767
and I think it's something
you could see develop here.

952
00:51:44,833 --> 00:51:46,700
To me, that makes a lot of sense. I mean,

953
00:51:46,767 --> 00:51:48,900
I think when you think
about governments regulating,

954
00:51:48,967 --> 00:51:50,700
I would like to see governments say,

955
00:51:50,767 --> 00:51:53,967
among those countries where they are
gonna demand some form of regulation,

956
00:51:54,567 --> 00:51:56,833
"Look, we want you to have a system

957
00:51:56,900 --> 00:51:59,567
"where you state your values
and explain what you're doing."

958
00:51:59,933 --> 00:52:01,833
And then, you know, Facebook could say,

959
00:52:01,900 --> 00:52:04,333
"Well, look, we have that system.
Here it is."

960
00:52:04,400 --> 00:52:07,333
And then, in a sense, this could
eventually, again, if it's successful,

961
00:52:07,400 --> 00:52:09,400
be a kind of best practices model

962
00:52:09,467 --> 00:52:11,200
for governments that say...

963
00:52:11,267 --> 00:52:14,300
Because, I think, most governments
don't actually want to engage

964
00:52:14,367 --> 00:52:17,567
in the review of billions of pieces
of content every day.

965
00:52:17,633 --> 00:52:20,933
They just want to make sure
that it's being done responsibly.

966
00:52:21,000 --> 00:52:23,000
And that seems to be very sensible.

967
00:52:23,067 --> 00:52:26,267
And I think, similarly, if other companies
want to get involved over time,

968
00:52:26,333 --> 00:52:28,400
and there could be
common industry-wide phenomenon,

969
00:52:28,467 --> 00:52:30,167
that too, I think, fits this goal.

970
00:52:30,233 --> 00:52:31,967
I don't think governments should want to.

971
00:52:32,033 --> 00:52:35,400
And I would not love the idea
of governments going into the details

972
00:52:35,467 --> 00:52:39,600
of what content goes up
and what content comes down.

973
00:52:39,667 --> 00:52:42,167
But I think they could possibly say,
"Have a system."

974
00:52:42,233 --> 00:52:43,267
Yeah.

975
00:52:43,333 --> 00:52:45,133
And that seems
like a logical mode of regulation.

976
00:52:45,200 --> 00:52:47,567
-It protects freedom of expression.
-That makes a lot of sense to me.

977
00:52:47,633 --> 00:52:51,467
We've been engaged
with the French government,

978
00:52:51,533 --> 00:52:54,133
which is trying to figure out
its regulatory approach

979
00:52:54,200 --> 00:52:59,667
to the Internet and expression
and balancing these different values.

980
00:53:01,400 --> 00:53:04,533
And they've been very engaged in this,

981
00:53:04,600 --> 00:53:07,067
and very sophisticated in thinking through

982
00:53:07,133 --> 00:53:11,000
and understanding all of the systems
that exist out there today,

983
00:53:11,067 --> 00:53:14,067
and where they should be improved,
but where they are good.

984
00:53:14,133 --> 00:53:18,333
And it seems to me like,
where they are getting to,

985
00:53:18,400 --> 00:53:20,700
which feels right,

986
00:53:20,767 --> 00:53:25,133
is more on the lens
of requiring systems in place

987
00:53:25,200 --> 00:53:27,900
that are adequate
and continuously improving,

988
00:53:27,967 --> 00:53:29,833
where there is the right transparency

989
00:53:29,900 --> 00:53:32,900
so that people and the governments
can understand

990
00:53:32,967 --> 00:53:36,633
what the systems are and what
the companies are doing to maintain them.

991
00:53:36,700 --> 00:53:40,900
And it's quite possible over time,
if this approach works well,

992
00:53:40,967 --> 00:53:46,167
I would hope that one thing
we could potentially consider is

993
00:53:46,233 --> 00:53:50,667
having governments consider
this approach of an independent board

994
00:53:50,733 --> 00:53:56,300
as part of the overall system of content
governance for Internet services online.

995
00:53:57,467 --> 00:53:59,300
And I think that has also the advantage

996
00:53:59,367 --> 00:54:01,700
that you wouldn't
then have the balkanization

997
00:54:01,767 --> 00:54:07,900
of 170 sum different standards
if there is a consensus internationally,

998
00:54:07,967 --> 00:54:10,833
that there should be a process,
and that it's a process regulation

999
00:54:10,900 --> 00:54:14,233
versus every country coming up
with a different set of rules

1000
00:54:14,300 --> 00:54:18,267
that makes it more feasible to have
the kind of openness that we have.

1001
00:54:18,333 --> 00:54:22,267
That's another reason for the appeal of
a kind of oversight board that's modeled

1002
00:54:22,333 --> 00:54:24,833
in some way on constitutional courts
that we're familiar with.

1003
00:54:24,900 --> 00:54:27,333
Because almost all countries,
not all, but almost all,

1004
00:54:27,400 --> 00:54:30,600
think that's roughly the right way to
go about treating these kinds of issues.

1005
00:54:30,667 --> 00:54:31,833
Yeah, that makes sense.

1006
00:54:31,900 --> 00:54:35,367
All right, so, the last issue
I want to make sure we touch on,

1007
00:54:35,433 --> 00:54:37,967
because we've spent
all this time today talking about

1008
00:54:38,033 --> 00:54:41,767
the independent board and voice.

1009
00:54:42,867 --> 00:54:44,133
But this isn't the only place

1010
00:54:44,200 --> 00:54:47,200
where we're trying to focus on
more independent governments.

1011
00:54:47,267 --> 00:54:48,767
Just last week,

1012
00:54:50,200 --> 00:54:54,333
we announced that
we were joining with 27 other companies

1013
00:54:54,400 --> 00:54:57,767
to create the Libra Association

1014
00:54:57,833 --> 00:55:03,967
to help stand up a new currency
and financial infrastructure

1015
00:55:04,033 --> 00:55:06,933
to empower billions of people
around the world

1016
00:55:07,000 --> 00:55:09,600
to be able to participate
in the financial system

1017
00:55:09,667 --> 00:55:13,500
and do affordable
or even free transactions

1018
00:55:13,567 --> 00:55:16,233
that can unlock a whole lot of
different use cases, if this works.

1019
00:55:16,300 --> 00:55:19,633
But clearly,
to build something like that,

1020
00:55:19,700 --> 00:55:23,433
that's not the type of thing that
any one company can do by themselves

1021
00:55:23,500 --> 00:55:27,433
or that other participants
in a system like that

1022
00:55:27,500 --> 00:55:30,300
would want any one company
to have control over.

1023
00:55:30,367 --> 00:55:34,133
So, we're very deliberate
in making sure that,

1024
00:55:34,200 --> 00:55:37,400
as we worked with a lot of
other companies to start this,

1025
00:55:37,467 --> 00:55:41,700
that Facebook gets one vote
in the Association.

1026
00:55:41,767 --> 00:55:44,200
We don't get to control it.

1027
00:55:44,267 --> 00:55:47,700
And we want this to be
as broad of an area as possible.

1028
00:55:47,767 --> 00:55:51,333
And I'm curious, how you think about,

1029
00:55:52,833 --> 00:55:55,200
what do you think
are the important lessons

1030
00:55:55,267 --> 00:55:59,400
from the experience of thinking through
these issues on expression

1031
00:55:59,467 --> 00:56:01,667
on Facebook that we should carry
towards other areas?

1032
00:56:01,733 --> 00:56:06,967
But also, where are the other areas that,
over time, we can potentially

1033
00:56:07,033 --> 00:56:10,200
create this independent governance
to build trust

1034
00:56:10,267 --> 00:56:12,767
and have more independent processes

1035
00:56:12,833 --> 00:56:15,933
for handling
really socially-important parts

1036
00:56:16,000 --> 00:56:18,533
of what's going on on the Internet?

1037
00:56:19,900 --> 00:56:23,933
Well, it's interesting,
another private/public question,

1038
00:56:24,000 --> 00:56:27,100
where money is traditionally
the province of governments

1039
00:56:27,167 --> 00:56:31,167
and international financial systems
are under the control of states,

1040
00:56:31,233 --> 00:56:34,867
either through their own currencies
or through international bodies.

1041
00:56:34,933 --> 00:56:39,733
I think one challenge with this is
not only having a balance of

1042
00:56:39,800 --> 00:56:43,067
different, private entities
that might be involved in setting this up,

1043
00:56:43,133 --> 00:56:46,533
but thinking about the public values

1044
00:56:46,600 --> 00:56:50,467
of protection against
money laundering and terrorist financing,

1045
00:56:50,533 --> 00:56:52,100
all those kind of things.

1046
00:56:52,167 --> 00:56:53,700
I think those are things
you've thought about,

1047
00:56:53,767 --> 00:56:55,567
but imagine
from the perspective of states,

1048
00:56:55,633 --> 00:56:56,933
those are really big concerns.

1049
00:56:57,000 --> 00:57:01,167
Having something that is global in reach,

1050
00:57:01,233 --> 00:57:04,533
but that's outside the control
of any one state,

1051
00:57:04,600 --> 00:57:08,200
it presents an international
governance problem in the same way

1052
00:57:08,267 --> 00:57:11,100
that some of the things we've been
talking about with expression do.

1053
00:57:11,167 --> 00:57:15,733
And I do think it is a shift
in the way international law

1054
00:57:15,800 --> 00:57:20,167
will work over time
when you've got these kind of functions

1055
00:57:20,233 --> 00:57:23,467
being carried out in this private sector
on a global scale.

1056
00:57:23,533 --> 00:57:27,800
Yeah. And just to add one piece to this,

1057
00:57:27,867 --> 00:57:32,800
the points on AML and KYC,
or anti-money laundering

1058
00:57:32,867 --> 00:57:36,233
and you're accustomed to a lot of
the financial regulations...

1059
00:57:36,300 --> 00:57:40,100
One of the things that we hoped
to push forward here

1060
00:57:40,167 --> 00:57:44,667
in our contribution to Libra is that

1061
00:57:45,500 --> 00:57:49,767
a lot of the cryptocurrency space today
is still trying to navigate

1062
00:57:49,833 --> 00:57:53,767
where a lot of those utilities
are going to fit,

1063
00:57:53,833 --> 00:57:58,033
on how they follow those rules
and how they work with governments.

1064
00:57:58,100 --> 00:58:00,267
We're trying to take an approach

1065
00:58:00,333 --> 00:58:05,300
where we are very compliant
and safe on that

1066
00:58:05,367 --> 00:58:10,700
and using the tools that we have
as a large, established company,

1067
00:58:10,767 --> 00:58:13,033
to bring some of that to bear

1068
00:58:14,000 --> 00:58:18,000
in making a version
of this digital currency

1069
00:58:18,067 --> 00:58:20,100
where those protections do exist.

1070
00:58:20,167 --> 00:58:24,833
Unlike a lot of other areas
in the crypto space today.

1071
00:58:24,900 --> 00:58:28,400
That's one of the unique things
we can help bring,

1072
00:58:28,467 --> 00:58:32,767
and also why we're partnering with folks
like Visa, and MasterCard, and PayPal,

1073
00:58:32,833 --> 00:58:36,267
and a lot of these financial
and payments companies

1074
00:58:36,333 --> 00:58:38,733
who have a specific expertise in that.

1075
00:58:38,800 --> 00:58:44,367
And we want to make sure that the
Libra work is pushing in that direction.

1076
00:58:44,767 --> 00:58:48,533
For me, the big take away
from the oversight board thinking,

1077
00:58:48,600 --> 00:58:51,767
that's very much relevant,
I think, to the Libra project and others

1078
00:58:51,833 --> 00:58:55,500
is that companies like Facebook
have to do two things at the same time.

1079
00:58:55,567 --> 00:58:57,133
They have to take responsibility

1080
00:58:57,200 --> 00:59:01,733
for the things that people correctly feel
they do in the world,

1081
00:59:01,800 --> 00:59:05,867
and at the same time, they have to
devolve some power and decision-making

1082
00:59:06,367 --> 00:59:10,533
to transparent, independent,
third-party entities.

1083
00:59:10,600 --> 00:59:14,633
And it's a delicate balance,
to get both of those things right.

1084
00:59:14,700 --> 00:59:17,100
People want Facebook to be
responsible for its content,

1085
00:59:17,167 --> 00:59:20,067
but they also think,
based on our consultations,

1086
00:59:20,167 --> 00:59:22,667
that there are a lot of decisions
that Facebook shouldn't be making

1087
00:59:22,733 --> 00:59:24,067
and therefore, should be devolved

1088
00:59:24,133 --> 00:59:26,133
to an independent decision-making body.

1089
00:59:26,200 --> 00:59:29,900
Similarly, in the context
of a currency like this,

1090
00:59:29,967 --> 00:59:32,533
unless Facebook takes the leading role,
as you've done,

1091
00:59:32,600 --> 00:59:34,600
it's unlikely that this will emerge.

1092
00:59:35,567 --> 00:59:37,867
It's a bold experiment in that regard.

1093
00:59:37,933 --> 00:59:39,000
At the same time,

1094
00:59:39,067 --> 00:59:43,933
you are also creating independent
institutions that will be transparent

1095
00:59:44,000 --> 00:59:47,067
and that will be subject
to public discussion and analysis

1096
00:59:47,133 --> 00:59:48,233
in their decision-making.

1097
00:59:48,300 --> 00:59:52,567
They will do a lot of the ultimate, really
important decision-making going forward.

1098
00:59:52,633 --> 00:59:55,233
And that, I think, to me,
is the path forward

1099
00:59:55,300 --> 00:59:57,967
at least at this historical moment.

1100
00:59:58,600 --> 01:00:00,367
Companies do a lot of things

1101
01:00:00,433 --> 01:00:02,367
and the public wants them
to do those things.

1102
01:00:02,433 --> 01:00:04,933
And, in fact, thinks that the companies
do those things pretty well

1103
01:00:05,000 --> 01:00:06,167
in many instances.

1104
01:00:06,233 --> 01:00:08,967
At the same time, the public gets

1105
01:00:09,033 --> 01:00:11,700
that when companies do
so many things well, there are dangers.

1106
01:00:11,767 --> 01:00:15,600
And one potential solution to that,
maybe the strongest solution,

1107
01:00:15,667 --> 01:00:17,033
is public scrutiny,

1108
01:00:17,100 --> 01:00:20,133
and the devolving of some power
away from you guys.

1109
01:00:20,200 --> 01:00:22,433
And that, to me, is a consistent theme.

1110
01:00:22,500 --> 01:00:25,533
And I hope it's a theme that we'll see
repeated in other contexts too.

1111
01:00:25,967 --> 01:00:27,000
I think that's right, and, I think,

1112
01:00:27,067 --> 01:00:28,900
another issue where you see that
is privacy

1113
01:00:28,967 --> 01:00:30,933
which overlaps with expression sometimes.

1114
01:00:31,000 --> 01:00:34,433
When you're disclosing private content,
then there's an intersection.

1115
01:00:34,500 --> 01:00:37,733
But there's also a whole realm of privacy
that isn't so much about speech,

1116
01:00:37,800 --> 01:00:40,233
but about what's being done
with people's data.

1117
01:00:40,300 --> 01:00:42,567
And that's another one where I think

1118
01:00:42,633 --> 01:00:45,233
these concerns about the balance between

1119
01:00:45,300 --> 01:00:49,200
voluntary actions by industry
and public regulation

1120
01:00:49,267 --> 01:00:52,567
plays out on the global scale and where...

1121
01:00:52,633 --> 01:00:56,133
Maybe the balance might be different
in terms of more government regulation

1122
01:00:56,200 --> 01:00:57,467
versus voluntary action.

1123
01:00:57,533 --> 01:00:59,967
But I think you see
that same set of trade-offs

1124
01:01:00,033 --> 01:01:04,100
of people worried about these decisions
that can have really big impacts

1125
01:01:04,167 --> 01:01:05,800
on people's lives

1126
01:01:05,867 --> 01:01:08,433
that aren't necessarily being held
to uniform standards

1127
01:01:08,500 --> 01:01:11,933
or being applied in the same way
in different parts of the world.

1128
01:01:12,000 --> 01:01:14,567
So, I think that's another analogous area.

1129
01:01:14,633 --> 01:01:16,100
Yeah, that makes sense.

1130
01:01:16,167 --> 01:01:19,800
Of the major issues that I see,

1131
01:01:21,767 --> 01:01:24,167
that we are at the center of
or that faced the Internet,

1132
01:01:24,233 --> 01:01:25,867
I do think it's...

1133
01:01:25,933 --> 01:01:30,633
Speech and how to handle harmful content
is one of them.

1134
01:01:30,700 --> 01:01:35,500
Which I think this independent approach
will hopefully be one of the tools

1135
01:01:35,567 --> 01:01:38,400
for helping to manage that better.

1136
01:01:39,700 --> 01:01:41,700
There's a whole separate set of things
around elections

1137
01:01:41,767 --> 01:01:44,533
and preventing election interference,

1138
01:01:44,600 --> 01:01:47,067
which may overlap with us in some places

1139
01:01:47,133 --> 01:01:50,433
around what kinds of content you want
on the services,

1140
01:01:50,500 --> 01:01:52,767
but that has...

1141
01:01:52,833 --> 01:01:56,867
I think that that scenario
where more regulation,

1142
01:01:58,267 --> 01:02:00,133
where countries should decide
for themselves

1143
01:02:00,200 --> 01:02:03,100
who they want spending money to advertise
in their elections,

1144
01:02:03,167 --> 01:02:06,933
what they define
as acceptable advertising.

1145
01:02:07,767 --> 01:02:09,767
And there's probably more room
for more regulation there

1146
01:02:09,833 --> 01:02:13,200
and there's more room
for collaboration with governments,

1147
01:02:13,267 --> 01:02:15,133
especially on the intelligence side,

1148
01:02:15,200 --> 01:02:21,600
to share signals to be able to find people
who are trying to interfere in elections.

1149
01:02:21,667 --> 01:02:24,300
Then the third issue is privacy

1150
01:02:24,367 --> 01:02:26,967
where I do think that that's one,

1151
01:02:27,033 --> 01:02:29,167
what we've seen with GDPR,

1152
01:02:29,233 --> 01:02:34,467
that there can be regulation
that, I think, can be effective

1153
01:02:34,533 --> 01:02:38,967
at encoding the rights
that people care about.

1154
01:02:39,033 --> 01:02:40,800
But that, also...

1155
01:02:41,767 --> 01:02:45,900
That give enough space for companies

1156
01:02:45,967 --> 01:02:49,500
to have principles

1157
01:02:49,567 --> 01:02:53,367
like, "We're not gonna put
our data centers in countries,

1158
01:02:53,433 --> 01:02:57,367
"and localized data, in ways that are
going to put people in harm's way."

1159
01:02:58,633 --> 01:03:00,967
To me, one of the most important
parts of GDPR

1160
01:03:01,033 --> 01:03:05,533
was that it didn't spell out
the data localization requirement

1161
01:03:05,600 --> 01:03:07,033
and I think that
there's probably more room

1162
01:03:07,100 --> 01:03:09,267
for additional legislation and rules

1163
01:03:09,333 --> 01:03:12,100
that make it even clearer that
that's not gonna be the direction

1164
01:03:12,167 --> 01:03:14,800
that we want the Internet to go in.

1165
01:03:15,867 --> 01:03:17,100
So, if that's the third issue,

1166
01:03:17,167 --> 01:03:20,100
then the last one,
I think, is almost the...

1167
01:03:20,167 --> 01:03:24,900
It really is the tension with privacy
which is around data portability

1168
01:03:24,967 --> 01:03:28,133
and people's ability to move data
from one place to another, which...

1169
01:03:29,500 --> 01:03:31,433
It doesn't need to be in tension.

1170
01:03:31,500 --> 01:03:34,800
We should be able
to have privacy and portability,

1171
01:03:34,867 --> 01:03:37,633
but if we, as a society,

1172
01:03:37,700 --> 01:03:43,000
are mostly focused on,
or talking about issues around privacy,

1173
01:03:43,600 --> 01:03:45,600
then the incentive for all companies

1174
01:03:45,667 --> 01:03:47,933
is to make sure
that all the data's locked down,

1175
01:03:49,033 --> 01:03:52,100
in that, you're not making it
as easy as possible for people

1176
01:03:52,167 --> 01:03:54,333
to bring data out
that could potentially be risky

1177
01:03:54,400 --> 01:03:57,567
and create issues on the privacy side.

1178
01:03:57,633 --> 01:04:01,367
So, I think that we need to balance that
with good rules on data portability

1179
01:04:01,433 --> 01:04:04,400
that also require that companies act
in certain ways

1180
01:04:04,467 --> 01:04:08,600
to always enable people to bring the data

1181
01:04:08,667 --> 01:04:12,500
that they should have access to use
in other places as well.

1182
01:04:12,567 --> 01:04:14,533
And there the hard questions are about,

1183
01:04:14,600 --> 01:04:16,467
"Okay, well, what's the definition
and what's your data

1184
01:04:16,533 --> 01:04:19,400
"and what's someone else's data,

1185
01:04:19,467 --> 01:04:20,733
"what should you have access
to be able to bring that?"

1186
01:04:20,800 --> 01:04:24,600
We could spend a whole hour on,
on just that alone.

1187
01:04:25,767 --> 01:04:28,533
But, yeah, I do think some of these
will lend themselves

1188
01:04:28,600 --> 01:04:31,267
to independent governance

1189
01:04:31,333 --> 01:04:35,367
and, in some ways, even self-regulation
among an industry.

1190
01:04:36,367 --> 01:04:39,267
Some of them will be places

1191
01:04:39,333 --> 01:04:43,267
where, I think, governments should require
that companies have systems in place.

1192
01:04:43,333 --> 01:04:45,200
Like we were talking about on content.

1193
01:04:46,000 --> 01:04:47,867
And others will probably lend themselves

1194
01:04:47,933 --> 01:04:51,367
to more specific rules
about how companies need to operate.

1195
01:04:51,433 --> 01:04:55,500
But I do think that this is one
of the most important sets of questions

1196
01:04:55,567 --> 01:04:58,367
facing the Internet and all of us now,

1197
01:04:58,433 --> 01:05:01,333
because there's not just one issue here.

1198
01:05:01,400 --> 01:05:04,667
I think it's at least
these four different issues

1199
01:05:04,733 --> 01:05:06,800
which are all meaningful issues

1200
01:05:06,867 --> 01:05:11,533
that we need to figure out a way
that the public can trust

1201
01:05:11,600 --> 01:05:15,100
that companies are operating in a way
that they're happy with

1202
01:05:15,167 --> 01:05:20,133
in order for the Internet to reach
the potential that everyone wants it to.

1203
01:05:22,200 --> 01:05:25,700
-So, thank you for joining for this.
-Thanks for having us. Fun.

1204
01:05:26,467 --> 01:05:29,967
Yeah, we're still in the middle
of the process on this,

1205
01:05:31,067 --> 01:05:35,267
but the public consultation
in getting people's views

1206
01:05:35,333 --> 01:05:40,000
on how they think that we should
be moving forward on this,

1207
01:05:40,067 --> 01:05:44,033
including the almost 30 different trials
that we've done around the world.

1208
01:05:44,100 --> 01:05:46,900
It's been a very interesting process
so far.

1209
01:05:46,967 --> 01:05:48,700
And if you're interested in learning more,

1210
01:05:48,767 --> 01:05:52,100
I encourage you to read the report
that we've published.

1211
01:05:53,400 --> 01:05:56,633
It's long,
but it has a lot of detail in there

1212
01:05:56,700 --> 01:05:58,167
around a lot of the different feedback

1213
01:05:58,233 --> 01:05:59,833
and some of the trade-offs
and the principles

1214
01:05:59,900 --> 01:06:03,233
of how we can design
this board and process

1215
01:06:03,300 --> 01:06:09,167
to basically make the different decisions
to implement this.

1216
01:06:10,367 --> 01:06:13,033
But I'm looking forward to,
over the next six months,

1217
01:06:13,100 --> 01:06:16,100
putting this into place

1218
01:06:16,167 --> 01:06:21,700
and hopefully evolving it over time
to take on a more important role

1219
01:06:21,767 --> 01:06:25,200
in how we manage content
across our services

1220
01:06:25,267 --> 01:06:29,167
and hopefully, eventually,
for more places on the Internet.

1221
01:06:29,233 --> 01:06:31,767
So, thank you for joining for this.

1222
01:06:31,833 --> 01:06:34,533
You've got a really challenging task,
so good luck.

1223
01:06:35,733 --> 01:06:36,733
Thanks. Thank you.

1224
01:06:40,800 --> 01:06:41,800
All right.