[Joint House and Senate Hearing, 115 Congress]
[From the U.S. Government Publishing Office]
115th Congress } Printed for the use of the
2nd Session } Commission on Security and Cooperation in Europe
======================================================================
Lies, Bots, and Social Media:
What is Computational Propaganda
and How Do We Defeat It?
[GRAPHIC NOT AVAILABLE IN TIFF FORMAT]
November 29, 2018
Briefing of the
Commission on Security and Cooperation in Europe
------------------------------------------------------------------------
Washington: 2019
Commission on Security and Cooperation in Europe
234 Ford House Office Building
Washington, DC 20515
202-225-1901
[email protected]
http://www.csce.gov
@HelsinkiComm
HOUSE SENATE
CHRISTOPHER H. SMITH, New Jersey ROGER WICKER, Mississippi,
Co-Chairman Chairman
ALCEE L. HASTINGS, Florida BENJAMIN L. CARDIN. Maryland
ROBERT B. ADERHOLT, Alabama JOHN BOOZMAN, Arkansas
MICHAEL C. BURGESS, Texas CORY GARDNER, Colorado
STEVE COHEN, Tennessee MARCO RUBIO, Florida
RICHARD HUDSON, North Carolina JEANNE SHAHEEN, New Hampshire
RANDY HULTGREN, Illinois THOM TILLIS, North Carolina
SHEILA JACKSON LEE, Texas TOM UDALL, New Mexico
GWEN MOORE, Wisconsin SHELDON WHITEHOUSE, Rhode Island
Executive Branch Commissioners
DEPARTMENT OF STATE
DEPARTMENT OF DEFENSE
DEPARTMENT OF COMMERCE
[II]
The Helsinki process, formally titled the Conference on Security
and Cooperation in Europe, traces its origin to the signing of the
Helsinki Final Act in Finland on August 1, 1975, by the leaders of 33
European countries, the United States and Canada. As of January 1,
1995, the Helsinki process was renamed the Organization for Security
and Cooperation in Europe (OSCE). The membership of the OSCE has
expanded to 56 participating States, reflecting the breakup of the
Soviet Union, Czechoslovakia, and Yugoslavia.
The OSCE Secretariat is in Vienna, Austria, where weekly meetings
of the participating States' permanent representatives are held. In
addition, specialized seminars and meetings are convened in various
locations. Periodic consultations are held among Senior Officials,
Ministers and Heads of State or Government.
Although the OSCE continues to engage in standard setting in the
fields of military security, economic and environmental cooperation,
and human rights and humanitarian concerns, the Organization is
primarily focused on initiatives designed to prevent, manage and
resolve conflict within and among the participating States. The
Organization deploys numerous missions and field activities located in
Southeastern and Eastern Europe, the Caucasus, and Central Asia. The
website of the OSCE is: .
The Commission on Security and Cooperation in Europe, also known as
the Helsinki Commission, is a U.S. Government agency created in 1976 to
monitor and encourage compliance by the participating States with their
OSCE commitments, with a particular emphasis on human rights.
The Commission consists of nine members from the United States
Senate, nine members from the House of Representatives, and one member
each from the Departments of State, Defense and Commerce. The positions
of Chair and Co-Chair rotate between the Senate and House every two
years, when a new Congress convenes. A professional staff assists the
Commissioners in their work.
In fulfilling its mandate, the Commission gathers and disseminates
relevant information to the U.S. Congress and the public by convening
hearings, issuing reports that reflect the views of Members of the
Commission and/or its staff, and providing details about the activities
of the Helsinki process and developments in OSCE participating States.
The Commission also contributes to the formulation and execution of
U.S. policy regarding the OSCE, including through Member and staff
participation on U.S. Delegations to OSCE meetings. Members of the
Commission have regular contact with parliamentarians, government
officials, representatives of non-governmental organizations, and
private individuals from participating States. The website of the
Commission is: .
Lies, Bots, and Social Media: What is Computational
Propaganda and How Do We Defeat It?
November 29, 2018
Page
PARTICIPANTS
Mark Toner, State Department Senior Advisor, Commission on Security
and Cooperation in Europe 1
Paul Massaro, Policy Advisor, Commission on Security and
Cooperation in Europe 16
Matt Chessen, Acting Deputy Science and Technology Advisor to the
Secretary of State, U.S. Department of State 2
Ambassador Karen Kornbluh, Senior Fellow and Director, Technology
Policy Program, The German Marshall Fund of the United States 7
Nina Jankowicz, Global Fellow at the Woodrow Wilson International
Center for Scholars' Kennan Institute 11
Lies, Bots, and Social Media: What is Computational.
Propaganda and How Do We Defeat It?
----------
November 29, 2018
The briefing was held at 10:33 a.m. in Room 562, Dirksen Senate
Office Building, Washington, DC, Mark Toner, State Department Senior
Advisor, Commission on Security and Cooperation in Europe, presiding.
Panelists present: Mark Toner, State Department Senior Advisor,
Commission on Security and Cooperation in Europe; Paul Massaro, Policy
Advisor, Commission on Security and Cooperation in Europe; Matt
Chessen, Acting Deputy Science and Technology Advisor to the Secretary
of State, U.S. Department of State; Ambassador Karen Kornbluh, Senior
Fellow and Director, Technology Policy Program, The German Marshall
Fund of the United States; and Nina Jankowicz, Global Fellow at the
Woodrow Wilson International Center for Scholars' Kennan Institute.
Mr. Toner. Good morning, and thanks to all of you for joining us
for what I hope is just the start of an ongoing discussion on a topic
that is increasingly relevant to all of us in the digital age, and that
is computational propaganda and, more broadly, disinformation spread
via digital platforms. My name is Mark Toner and I'm joined today by my
colleague Paul Massaro.
On behalf of the Commission on Security and Cooperation in Europe,
a.k.a. the Helsinki Commission, and the commission's chairman, Senator
Roger Wicker, I wanted to thank our three panelists today, each of whom
are experts on the issue of computational propaganda and, more broadly,
online disinformation--what it is, where it originates, how it works,
and what we can do to stop it or at least better manage it, all, of
course, while preserving our bedrock freedom of expression.
It's been said we live in a post-truth world. Indeed, information
technology has moved at such a pace that many of us have lost our
ability to delineate between fact and fiction, a vulnerability that
adversaries, either individuals or political groups or corporations or
even, God forbid, foreign governments have sought to exploit. It's my
hope that our discussion today will unpack not only why this is
happening but also how we can best try to fix it.
It's a conversation, of course, that has to include all the
different players--that's industry, civil society, and government, of
course--and employ technology and policy in new and creative ways. And
while there's certainly plenty of blame to go around in our slow
response to this threat, there are, I hope, real solutions that we can
pursue that can return the internet to the promise of its early days.
Our first speaker, Matt Chessen, is a career U.S. diplomat,
technologist, and author who is currently serving as the acting deputy
advisor in the Office of Science and Technology to the secretary of
state. It's his job to connect State Department foreign policy
priorities to research, development, and discoveries emerging from the
high tech and private sectors. Matt also manages the development of
policy portfolios for artificial intelligence and computational
propaganda, and Matt's written and spoken extensively on the problem of
computational propaganda and how to address it through AI [artificial
intelligence] and other technology.
Our second panelist is Karen Kornbluh, who is currently the senior
fellow and director for the technology policy program at the German
Marshall Fund of the United States. Karen was previously with the
Council on Foreign Relations, where she was the senior fellow for
digital policy. She also served as U.S. Ambassador to the OECD
[Organisation for Economic Cooperation and Development] from 2009 to
2012, where she spearheaded the development of the first internet
policymaking principle.
And our third panelist today is Nina Jankowicz, a global fellow at
the Woodrow Wilson Center's Kennan Institute, who is currently working
on a book about the evolution of modern Russian influence campaigns in
Eastern Europe. She's also advised the Ukrainian Government on
strategic communications, and her op-eds have been published in The New
York Times and Washington Post, among others. She's also a frequent
commentator on disinformation in Russian and Eastern European affairs.
Each of our panelists will give brief remarks before we open it up
to questions. They'll talk about different aspects of computational
propaganda. Matt will give us a bit of history about why this is such a
threat to our core democratic values, provide an overview of
computational propaganda, bots, and AI, and then talk a bit about left-
brain right-brain solutions to the problem.
Karen will give us more historical context--recent historical
context on how we've gotten to this point, from the internet's early
promise to today's algorithm-driven platforms. And Nina will talk about
some of the platforms--what some of the platforms have done to address
the problem, why it's not enough, and also offer some possible
solutions. And then, of course, we'll have a discussion. We'll open it
up to your questions.
So, Matt, do you want to start us off?
Mr. Chessen. Sure. Thank you, Mark, for the excellent overview and
introduction.
As Mark said, I'm going to start off by talking about the
civilizational context for these issues. So we really need to ask
ourselves, why do we care about computational propaganda, weaponized
narratives. Why do we care about the possibility of a post-truth world?
And I really like the answer that's provided by Arizona State
University and it's also mentioned actually by General Michael Hayden
quite a bit, and then it's these--the possibility of a post-truth world
actually directly undermines the enlightenment ideals of the search for
truth through reason.
And so if you go back to the founding of our country, the Founding
Fathers were Enlightenment Era thinkers. The Constitution is probably
the most important Enlightenment Era document. And so the ideas in the
Constitution set the foundation for modern democracy. These democracies
are based on rule of law systems where the empirical thinking is really
core to their functioning. And so facts and evidence are really
critical for everything from administrative due process to judicial
processes and evidence.
So if we actually start getting into this post-truth world where a
fact is just whatever you can convince people of, then facts don't
matter and this is a direct threat to the evidence-based rule of law
system that modern democracy is based upon. So I always tell people
it's very important that we push back on this idea that we're in a
post-truth world. There is an objective reality. Facts do matter.
Expertise matters and evidence matters. If we do concede that we're in
a post-truth world, then countries, organizations, or even people with
very strong information operation capabilities and a casual
relationship with the truth can hold inordinate amounts of power.
So if you take the principle that democracy is the superior system
and, based on evidence, people would come to that conclusion but you're
in a post-truth world, then countries like Russia and China can
convince their own populations and the populations of other countries
that democracy is inferior to authoritarian rule.
Some people speculate that the challenges we're facing with
weaponized narrative and computational propaganda and disinformation
spell the end of the enlightenment and I actually want to push back on
that idea. We can go into this in more detail in the Q&A if everybody
is interested but, really, the enlightenment was about the elites using
truth and reason to push back on the established institutions of the
time.
What you're seeing now is really hyper-empowered individuals using
technology to push back on the elites and push back on the institutions
that they've been running for a long period of time. And so we've hyper
empowered these individuals with really powerful informational tools
but we don't yet have the systems and frameworks to ensure
accountability and responsibility. We really don't have a lot of trust
in the system.
And so we've had over 200 years to create these checks and balances
in everything from the government use of power to peer-reviewed
research. But we've only had a few decades to create structures for the
internet and social media. And so to illustrate how this works, I'd
like to talk about what some people call the collective intelligence
system.
The collective intelligence system is the way that a society
determines truth from fiction. In the mid-20th century, our collective
intelligence system consisted of several major national newspapers, a
rich local media ecosystem, a few national TV channels, government,
academia, and the church, and that's how people basically filtered
what's true from what's not true.
But as the information channels have sort of exploded, confidence
in these institutions have gone down and some of this is natural
evolution just from the diversity of information outlets and some of it
is actually manufactured outrage by some of the malicious actors out
there. But the net result is that our collective intelligence system of
how our society determines truth from fiction really is broken right
now.
Into that gap, malicious actors have stepped in and are using new
technologies to create their own collective intelligence system.
They're creating malicious collective intelligence systems, and what
they do is they attract people into these systems with emotionally
pleasing disinformation and then they keep them on an emotional hook
and never let them go. And so this is what leads us to this idea of
computational propaganda.
So let's talk a little bit about what computational propaganda is
and give sort of an overview of some of the techniques. The technical
term for this is really online information operations. This comes from
sort of a military term. But I like computational propaganda because
it's good shorthand. Basically, what it refers to is the use of
information and communication technologies to manipulate people's
perceptions and influence their behavior.
And so computational propaganda uses a variety of different
technologies including big data. You have social media, you have
autonomous agents like bots and, increasingly, you're starting to see
artificial intelligence applications used. People like to focus on the
bots but we really need to keep in mind that there's human beings
behind all of this and there's human beings being affected by all of
this, right.
So the humans create the content. They circulate the content.
They're driving the information campaigns, and that's what's really at
the core in computational propaganda. It's this coordinated
manipulative activity. And so bots are one tool but it's not the only
tool that's used. But since it is something that people talk about a
lot we'll talk about the bots.
So bots have some advantages over humans that are based both in the
technology and based in human psychology. Most of the bots that are out
there are what I call dumb bots in that they are not artificial
intelligence and, basically, what they do is they post content either
on a schedule or in response to some sort of trigger, and this could be
a news key word or it could be some sort of prominent person tweeting
and then the bots will go and immediately retweet back.
Generally, there's three different types of bots out there. So
there's propaganda bots, and these are just trying to push content out
in high volume. You've got follower bots, and the follower bots are
trying to do astroturfing, and this is what's basically faking
grassroots support. They are liking people or content or following
people or content to try and push them up and game the algorithms that
are determining trending topics, trending people, trending content.
Then you have roadblock bots, which are really trying to undermine
free speech, undermine free expression. A lot of this really started
out with spamming hashtags where, you know, a group is trying to
organize using a hashtag. Some other group will come in and basically
just spam that hashtag with garbage so that they can't actually find
the content they're interested in. But it also is used for mass
intimidation, typically in combination with doxxing--releasing personal
information--and then sort of targeted harassment. This happens a lot
in authoritarian countries where an independent journalist will be
reporting on something and then they'll get threats from it looks like
tens of thousands of people who are saying, We know where your kids go
to school, we know where you live, we're going to come kill them or
kidnap them, and it basically pressures them into self-censoring their
speech.
So why do bots work? This is all rooted in human psychology, right?
Bots are actually sitting behind social media accounts that typically
look like the target audience they are trying to influence, and just
from human psychology, people tend to believe people who look like
them, act like them, or are sort of the same as them. And so if they
see a bunch of bots that look like actual people, like people in their
community, they tend to believe them more.
People are also proven to believe ideas if they're endorsed in
volume, right? So that's why astroturfing and gaming these algorithms
is persuasive. People like and believe what they think other people
like and believe. Bots can also undermine expertise. There are some
studies that show that in information-rich environments ideals that are
endorsed en masse by a group are more persuasive than ideas that are
pushed out by experts.
So there's lots of other techniques that computational propaganda
plays off of, and I talk about this because we can't just talk about
the technology. We actually have to talk about the science behind these
things as well.
So what's coming up next? We really need to think about what's
coming up with artificial intelligence because there's a number of
emerging technologies that are really going to significantly impact the
information space.
And I want to preface this with a statement that there's nothing
inherently wrong about these technologies. There's nothing inherently
bad about them. Technologies are neutral. It's really the malicious
actors, the malicious intent, and the malicious effects that we need to
focus on. So let's go through some of these technologies.
So, AI chat bots--if you've never used a chat bot I'd encourage you
to try one out. There' s a great one that's put out by Microsoft called
Zo. There's been one that's been out in Mandarin for a year. It's
called Xiaoice. Chat bots are basically bots that are able to have very
human-like conversations and a lot of research shows that people
develop very close emotional attachments to some of these chat bots.
With Xiaoice, people used to proclaim love for Xiaoice. They'd say,
Xiaoice, I wish you were a woman who--or a real girl who I could
actually marry. They'd say, Xiaoice, you're my only friend who's
available all the time, and people tend to have very high levels of
engagement with these chat bots.
There is also the emergence of these affective computing systems.
And so these are systems that both can portray human emotions very
accurately in either sort of avatars or just through text. But they can
also detect human emotions. And so if one of these AI systems is
interacting with you online, they can detect your emotional state and
then target content based on that particular emotional state.
Psychometric profiling is something that a lot of people know about
now because of the Cambridge Analytica scandals. So this is the idea
that with, really, relatively few pieces of data you can develop very
sort of solid profiles of people and determine their political
preferences, determine their personality, their sexuality. You can
determine skin color from data. And so every American has somewhere
between 2[,000] and 5,000 pieces of data that's just for sale through
data brokers, and then we give the data away all the time on social
media. And these can be used to, basically, build profiles that then
can be used for very personalized manipulative content.
Dynamic content creation systems are also an emerging phenomenon.
AI systems are already being used to write news articles. If you've
ever read a Minor League Baseball article it was probably generated by
an AI from a box score from the game. There's also AI systems that are
basically generating articles on company financial reports just from
the SEC filings. You have AIs that are writing books and screenplays,
which are not very good yet. [Laughter.] But they are actually
producing some really good visual art that's starting to win contests
and also some classical music and electronic music that's actually
pretty good. You can hear that on SoundCloud if you want to.
In combination with this sort of dynamic content generation is
probably the phenomenon that I'm most concerned about, which is the
audio and video manipulation tools, right. So right now, you can
actually clone someone's voice and you can actually take existing video
and make subtle manipulations to it that look very realistic and, you
know, it's very easy to see how you could take an existing video that's
out there on YouTube of a president or some foreign leader, subtly
change the video, subtly change the wording in there, and completely
change the meaning and cause some sort of crisis.
A lot of people are familiar with sort of the deep fakes
phenomenon. This was something where people basically created programs
for mapping actresses' faces to pornographic actresses and then, as a
lot of things do on the internet, it started out in pornography and
then was being used in sort of other entertainment applications.
People were using it to map Nicholas Cage's face to a bunch of
different movies that he was never in. I think someone actually put
Harrison Ford's face into the ``Solo'' movie, which was kind of an
interesting one that I just saw, and then there's videos of actually--
that are dynamically generated where they've taken a bunch of photos of
President Obama and an audio track of one of his speeches and then they
can dynamically generate as many different versions of that speech with
different vocal intonations, different backgrounds behind it.
And so the ability to actually create this sort of pliable and
artificial reality is really becoming sort of profound. I like to tell
people, you know, that the machines are coming and they want to have a
word with us, and we're entering into this era where a lot of the
content and a lot of the speech online could be machines talking to
people but also machines talking to machines, because those machines
are going to be trying to influence people.
But they're not going to be able to tell which accounts have
machines behind them, and so you're going to have machines talking to,
persuading, arguing with machines, and we could see the social spaces
on the internet really sort of overwhelmed with this machine-driven
speech and communication. So that could have profound impacts on sort
of the democratic spaces for speech and conversation online.
I do also want to just give an overview of how I like to think
about some of the solutions to these types of problems. There's a lot
of discussion out there about sort of the left--what I call the left
brain, or the rational, solutions. And so these are things like media
literacy training, teaching critical thinking, fact-checking tools, all
these technology tools that are really around, really giving you more
information about your information, right.
And so I think these are all incredibly valuable and they're
incredibly necessary, but we can't neglect the human element and the
emotional element, and the reason that a lot of people consume
disinformation is because it resonates with their identity. It's
emotionally pleasing to them. And the example I like to use is this
idea of what's going on with obesity.
So we have an obesity epidemic in this country. Everybody knows how
to lose weight, right, so you eat better and you exercise more, and we
give people a huge amount of data and information about their food. If
you go into a supermarket and look at any sort of package of food,
there's a lot of data and information there. But it doesn't help the
problem, right? It's not an informational problem. It has more to do
with people's cognition, their psychology, and their emotion.
So I think that's something that's really neglected in this space
is really thinking about that. A lot of people just think, Oh, we'll
just give people more information about their information and then
they'll make the right decisions. A lot of this is tied into people's
identity and a lot of it--I think some of the solutions to this of
getting people to believe the truth is actually trying to communicate
the truth to them in a way that doesn't throw up their identity-based
or their value-based defenses.
We need to combine these left-brain and right-brain solutions
together and I think we need to build new institutions of democracy for
the 21st century. So unlike these malicious collective intelligence
systems, these would be positive democratic collective intelligence
systems. These would be open, transparent. They'd include
accountability. They'd be based on logic and reason. They'd be bottom-
up so they'd be democratically generated. But they also take into
account the diversity of values, identities, and beliefs.
I also want to remind people that this is actually critically
important. You've got countries like China that are building social
credit systems. They're integrating this into their computational
propaganda systems. They're integrating this into their surveillance
systems, which are increasingly being enabled by AI, and these systems
are very effective at exercising top-down control over people.
If they start exporting these systems around the world, you may get
sort of serious competition that's highly enabled by technology that
really enables this top-down authoritarian social control. And so we
don't want that to become the de facto standard for the world. We need
to think about what is the democratic alternative of that.
So whether you call it weaponized narratives, computational
propaganda, information operations, or simply disinformation, I would
say that we are facing a serious challenge to democratic civilization.
Now, I'm optimistic because I've been working on this for several years
now and in the last couple years I've seen a state change in the amount
of awareness, energy, and resources focused on these issues.
But I am concerned and the problem is very serious. We don't want
to underestimate the task at hand. We are engaged in a struggle for
truth and reason that leads back to the very founding of our country
and the creation of modern democracy.
Thank you.
Mr. Toner. That's quite an ending but----
Amb. Kornbluh. That was great. That was great. Thank you.
Mr. Toner. Over to you, Karen.
Amb. Kornbluh. Thanks for having me.
Mr. Toner. Sure.
Amb. Kornbluh. This is really a great conversation. So I'm going to
talk to you a little bit about some things that I wrote in an article
for Foreign Affairs. And the reason I wrote the article is because I
was being asked as somebody's who's been around the table for decades
as internet policy was being made and was a big booster--and still am a
booster of the internet--how could I be saying that we needed to have
more regulations--how could I be critical? And I wanted to go back and
look at what had been our beliefs at the beginning. What did we do to
set up the policy framework of the original internet? And why are we in
a different place now?
So I'm going to start by just telling you a little story. So
there's this University of Wisconsin researcher and she--Young Mie
Kim--and she collected ads in the 2016 election. She had people tape
what they were seeing--they were being micro targeted--and send it in
to her, and she was combing through it after the election and she just
found these horrific, horrific ads. And she would try to trace them
back and figure out who had sponsored them, had they reported to the
FEC, did they have a website, had they filed with the IRS, and a bunch
of them had just disappeared.
She found one set. They pretended to be travel spots and they would
show you--I wish I had a clip of it--they would show you Paris and
purport that Paris was now under Sharia law, and they showed the Mona
Lisa covered up and, you know, the Eiffel Tower with a big sign in
Arabic on it. And it looked like a travel ad and it just had a little
clip that it was sponsored and she found out that it was sponsored by a
group called Secure America Now, and that was a dead end. And it was
only because another group, a watchdog group, Open [Secrets]--
opensecrets.org contacted the accountant for the group who mistakenly--
he didn't have to do this--sent the IRS filing unredacted that they
found out that this was funded by the Mercers.
So I just want to tell you this story to say that there's something
wrong in this environment that is supposed to bring more transparency
and democracy to all of us when you can have that kind of deception and
lack of transparency.
So how did we get here? So a lot of people think that the internet
was immaculately conceived in garages and dorm rooms with no policy
whatsoever. But, in fact, regulation was there from the beginning to
allow it to--allow these entrepreneurs, allow these brilliant engineers
to do what they did. In the Clinton administration they called a lot of
the steps to create the early framework deregulatory, but actually it
was pro-competition, pro-openness.
So what do I mean? In the Communications Act of 1996, they made
sure that the underlying telecommunications network would allow the
connection of competitive networks. There were spectrum auctions that
were contained in the Budget Reconciliation Act of 1993. AOL was
allowed to connect to the underlying network without paying the prices
that the long distance carriers had to pay. Otherwise, you would have
had to pay a permanent charge for sending an email.
There was a limited liability system that was set up for internet
platforms. It's referred to as Section 230 of the Communications
Decency Act. What it said was we want these platforms to take down
incendiary stuff, stuff that's bad for kids, stuff that violates laws,
but if you do that, platforms, you won't be considered a publisher so
go ahead and do it. And what that's meant over time is that they can't
be sued. They're not liable for the information that's on there, and
the idea of that was we want this to be a neutral platform. We don't
want a lot of friction in the system. We want everybody to be able to
speak, and that was the original intention.
On privacy, we have these global principles that came out of the
U.S. and were adopted in Canada and then the OECD socialized them--the
fair information policy practices that, again, are very based on notice
and opt in--and then we had this whole multi stakeholder system where
we said instead of having government figure out the ongoing rules we're
going to have this very dynamic system where civil society and
engineers and so on set the rules for the internet.
So that was the original idea is that we're going to have this--the
internet is going to be this competition of the underlying technology
space. It's going to be open. Everybody can be a speaker. And at the
beginning, it was incredibly exciting and it still is. I mean, people
were putting up blogs. Dissidents were able to speak. It was really
giving voice to the voiceless and power to the powerless and that was
very, very exciting.
But as time went on, as these technologies do, it became much more
centralized. So if you look at the social media networks that dominate
the internet now, there very much is an intermediary, right. There is
a[n] algorithm that mediates which of your friends' postings you see,
which of the people that you follow on Twitter you see. This algorithm
is determined in order to keep you engaged so that you're online
longer, so that more ads can be served to you. So this ad revenue-
driven system, this algorithm, really mediates your experience. It's
much less that I communicate to anybody I want and they communicate
back.
That story about Secure America Now tells you there's a lack of
transparency. You didn't know until recently, often, whether something
was an ad. Now you know it's an ad but you don't know necessarily who
it's coming from. You don't know what kind of data is being collected
from you. You don't know who's--whether you're being micro targeted.
You often don't know if you're talking to a bot or a human. There are
fake accounts or these closed groups. So there's a real lack of
transparency in this system whose primary value is, really,
transparency.
So there were efforts along the way to update a bunch of policies
but they faced obstacles. So the Federal Trade Commission wanted to get
rulemaking authority. But it was taken out of a bill. The Obama
administration proposed a Consumer Privacy Bill of Rights that didn't
get adopted. The FTC, as we all know now, negotiated consent decrees.
The one with Facebook really anticipated a lot of the problems that we
saw with Cambridge Analytica. But the FTC doesn't have the resources,
the manpower, the ability to pay people so that you get sophisticated
computer scientists to really follow up on that. The FTC doesn't have
subpoena power.
On the campaign spending side, where you could have caught a lot of
this stuff if you'd had some of the same kinds of rules as you do on
broadcasting but they'd been updated, the FEC--the Federal Election
Commission--completely deadlocked. The IRS--I don't know if you all
remember this pseudo-scandal under Obama of the IRS was supposedly
being biased in looking at charities and whether they were involved in
politics.
What that was really about was after Citizens United corporations
were allowed to play in politics, and through a series of IRS and FEC
decisions, charities, which don't have to disclose their donors, got
more and more and more involved in politics. The IRS was really trying
to check that. That failed, and so that's a dark money loophole as well
for advertising online and various activities online.
And then in the competition space, we have this focus on consumer
prices in the anti-trust space, and so that doesn't really do much for
you when you're in an environment where things are free. So the FTC
approved Facebook acquisitions of WhatsApp and Instagram. There's no
portability interoperability like there is with telecom.
When I was at the OECD, we negotiated these internet policymaking
principles. And the idea was, just as you were saying at the end, we
want to have free flow of information. We want to have human rights
principles, free expression. Let's not constrain that. But, certainly,
governments--individual national governments can set up policy
frameworks. But, unfortunately that wasn't followed up on very much.
And then we haven't had the kind of, just stepping back, the big
picture conversation that we had when broadcast came into being, even
about newspapers at the turn of the century. What is the purpose of
this new media and how is it going to further democracy as opposed to
undermining democracy?
So when it came to broadcast television we said, well, we're going
to have a different First Amendment approach to it because they're
using the public airwaves, and so we're going to have a public interest
requirement for broadcasters and we don't want them to step all over
local news so they're going to have to air local news. They're not
going to do certain things in the family hour. They're going to cover
both sides. They're going to have to say when it's an ad. They're going
to have to provide information to political opponents so they can get
the same price for ads.
We had a whole conversation about it, and then we constructed
public television to get at some of what you were talking about--what's
the positive that we want--and they funded public broadcast stations
and CPB did the content. In the case of cable, we got C-SPAN. So we've
had these kinds of big societal debates about what's the purpose of a
new media in a democracy and we haven't had that conversation.
So the problem is that we're left with the situation that Matt was
describing and I just want to highlight two things--two misconceptions
that people have about what goes on online. I think they think that the
internet reveals politics as they are. So you'll hear people say that
how are we to know that when we connected people there would be bad
people and our politics is broken. We just really have to fix our
politics.
But the internet really changes politics and I think it's really
important to pay attention to that. So how does that work? Well, it
influences smaller numbers of--it augments and makes seem louder and
bigger smaller numbers of voters in smaller states. There are very few
media sites--supposed news media sites that are responsible for much of
the fake news--that we see all over the internet.
So the Knight Foundation found that just 10 sites were responsible
for 65 percent of the fake news and conspiracy site links that appeared
on Twitter during the 2016 election. So the long tail is amplified. You
get a sense of fake consensus. You join a group--a fake group. Let's
say it's Blacktivist, which was a Russian group, or the Tennessee GOP
had a fake site that was run by the Russians. You joined the group.
It sounds like you. It's posting things that feel familiar, and
then gradually it radicalizes you--and this happens with the YouTube
recommendation algorithm as well. They call this affinity fraud. You
feel like you're in a group but you're not really in a group. So it
creates this fake consensus and, as I said, it changes politics. We
have this idea of astroturf in politics--you know, something that
pretends to be grassroots but it's really astroturf. This is astroturf
on steroids.
So there are a bunch of solutions. What I'd like to see [are]
solutions that really focus on transparency. The initial value of the
internet was really openness and transparency, and I hate to see
discussions where it's presumed that the only solutions have to do with
the government coming in and deciding what's true and what's not and
who can speak and who can't, or Facebook or Google deciding who can
speak and what's not. I think a lot more transparency. I think we're
not aware of how much centralization and lack of transparency there is.
So I'll give you a couple of examples. The Honest Ads Act was
introduced by Warner and Klobuchar and McCain. It has not passed into
law. The platforms have said that they've built these ad transparency
data bases. Jonathan Albright has just recently published showing all
kinds of problems with even what's contained in the data bases.
But I'd like to see them go even further and have almost a ``know
your customer'' kind of procedure, so that if you're Secure America Now
it's not just the front group that's listed in the ad transparency, but
they have to find out who's funding Secure America Now. If you want to
advertise online, we need to know who's funding you and that's--if you
look at the law, McCain-Feingold, that was a provision that's in there
that hasn't actually been validated by the Supreme Court. The Supreme
Court has usually found transparency to be of value in political
speech.
Another example I just want to throw out there on the news media
piece--when you pick up The New York Times, there's all kinds of
metadata, if you will, that you don't think about. They've separated
out what they claim to be news and what they claim to be opinion, and
there's a legal basis for that. You can sue them if something that they
claim is news is defamatory. That's less likely on the opinion. So you
know that you can trust what's in the front of the paper more than the
back because of this legal scheme. They have a masthead that says who
their publisher is, who their editors are. There's a byline on the
article. You know what their sourcing scheme is because there's
generally accepted practices.
When you see an article online, all of that metadata is stripped
out. You just see an article--it's from The New York Times. It's from
Breitbart. It has the same font. You don't know if they're claiming--
often a Breitbart article will sound like news but it's actually--
they've put in a couple of woulds and coulds so that if you ever tried
to sue them they would say, Oh, no, no, no, it's opinion, or it's
satire. You don't know often who's written it. You don't know what the
sourcing scheme is, who's paying for it, who's editing it, and what
procedures they're using.
So, again, in a mode of transparency, what if the platforms
whitelisted outlets that actually followed a certain procedure--
commonly accepted journalistic procedure about transparency so you
could know that these outlets actually follow the systems and you could
find that information. That would do--I think that's sort of in between
your two ideas of emotional and informational is let's give a user
interface that really helps people understand, gives watchdogs
information so they can present a narrative about what's going on.
So those are the kinds of things. I have a bunch of other ideas on
that score. But I think we just really have to move away from this idea
that the only way to fix the problem is by going right to content.
There's a lot we can do in terms of transparency and empowering users.
Mr. Toner. Nina.
Ms. Jankowicz. Thanks for having me. It's really an honor to be
here with such a distinguished panel, and that was a great segue,
Karen. So thank you for that.
So from calling the influence of malign foreign actors on our
electoral discourse and processes a ``pretty crazy idea''--that's a
quote from Mark Zuckerberg in 2016--to inviting regulation, however
begrudgingly, the social media platforms have come a long way since
2016. Facebook, Twitter, and Google have made political advertising
more transparent, creating searchable data bases of political ads, and
have tightened restrictions on who can purchase them.
In order to reduce the amount of fake news being spread by ads,
Facebook has updated its policies to block ads from pages that
repeatedly share stories marked as false by third-party fact-checking
organizations and Twitter's policies no longer allow the distribution
of hacked materials. Facebook has attempted to increase authenticity
and transparency around the governance of pages as an influence
vector--which were an influence vector of Russia's internet Research
Agency in 2016.
It claims that administrators of pages with large audiences undergo
extra verification in order to weed out fake accounts, and Facebook has
also made other adjustments to arm users with information about the
pages that they follow. All of the platforms have made adjustments to
their algorithms in order to attempt to combat the problem of
disinformation. Facebook did this by focusing on content from, quote,
``friends and family'' while Google's Project Owl changed the search
engine's algorithm to surface more, quote, ``authoritative content.''
And Twitter has reverted its newsfeed to a more chronological
timeline with less algorithmic intervention. Facebook and Twitter have
also invested more in content moderation to identify and remove content
that violates platforms' policies, including those related to false
information, fake accounts, and hate speech. This is not an exhaustive
list of the changes that they have made in the past 2 years but,
rather, an overview of the more well-known and purportedly messianic
features meant to deliver us from all means of internet evil.
But I'm here to tell you that they are not enough. Among the
features that I've just described, loopholes have been exploited,
missteps unforeseen, and pernicious disinformation allowed to flourish
to a point where there's no question in my mind that social media self-
regulation has been a failure. Just a day before the midterm election,
over a hundred Facebook and Instagram accounts likely controlled by the
IRA [Internet Research Agency] were still active and Facebook only
removed them after a tip from the FBI.
This is a more complicated problem, as my panelists--co-panelists
have alluded to, than just playing whack-a-troll or removing fake
accounts and increasing transparency on political ad buys. These
measures are, of course, first steps toward ensuring authentic healthy
online discourse. But even a cursory look through the performance
metrics of the ads released by the House Democrats after 2016 reveals
that plenty of the 2016 IRA disinformation performed very well
organically. Because the IRA had, over time, built trust and community
with their audience of sometimes hundreds of thousands of users per
page, many people saw and engaged with that content without the
purchase of a single ad.
Today, a lot of this type of content is spreading through
Facebook's groups, which the platform's algorithms prefers based on the
misguided understanding that they promote content between friends and
family, and these are not subject to the same level of content
moderation that public content is.
Yes, of course, groups are means of connecting people, but they're
also breeding grounds for disinformation due to their privacy settings.
Closed and secret groups are not searchable or transparent and the
content shared in them is only visible to members so Facebook is less
likely to moderate the content within them. What's more, the platform
still incentivizes and promotes this group activity.
Groups were a key vector in my investigation for BuzzFeed News into
fake profiles supporting an independent candidate for Senate in
Massachusetts. A number of fake persona[s] controlled not by lines of
code but, as Matt pointed out earlier, by humans and thus able to slip
by some of Facebook's detection tools for fake accounts would astroturf
groups with posts in favor of their candidate, creating the guise of
grassroots support for the campaign. And Columbia University's Jonathan
Albright has also researched how groups support the spread of
disinformation on Facebook, and he's noted that banned groups and pages
and brands such as Infowars often move their activity to closed groups
after their public pages have been banned.
Finally, the spotty and opaque enforcement of platforms' terms of
service including with brands like Infowars, which have a record of
spreading hate speech and disinformation, undermines the entire
discussion of content moderation to begin with. Legitimate voices are
often being silenced for small infractions, such as repeated uses of
profanity, while groups with more considerable public reach that are
violating much more serious clauses of the platforms' terms of service
are allowed to continue their diatribes until public outcry becomes too
great.
And to that end, transparency around take downs, as Karen
mentioned, has been lamentable. While Twitter has released the entire
archive of take downs related to state-backed activity on its service--
and that's a very good move, in my opinion--Facebook releases this
content only selectively. For instance, we don't know how many fake
individual profiles they've removed since 2016 related to, for
instance, the IRA or Iran. They only talk about pages, which is a
problem. And Google really never releases this information and this
contributes to the opacity of the problem and both the congressional
and public lack of understanding of how best to solve it.
But I do believe that, especially with the new Congress coming in,
there's an opportunity to join together in a bipartisan manner and
address this issue, and I have a few ideas about where this might
start. I agree with Karen that the Honest Ads Act needs to be passed
and I think it should be passed before 2020. There's no reason that
online political advertising, which, in 2018, saw at least a 200
percent increase in spending compared with the 2014 midterms, should be
subject to different rules than TV, radio, and print ads.
The sooner these rules are harmonized across platforms, including
smaller online advertisers--right now, the legislation focuses on kind
of the big kahunas and I think we need to talk about the smaller ones
as well--and integrated with existing FCC and FEC regulations, the
safer and more equitable our electoral processes will be. But as I
noted earlier, regulating advertising only covers a fraction of the
malicious information shared on social media. So I believe that
Congress should really push for more transparency surrounding groups,
pages, and some of the fronting organizations that purchase ads, as was
mentioned before.
And, further, I think Congress should explore the establishment of
a specialized independent regulatory or oversight mechanism that could
harmonize definitions of concepts like hate speech, abuse, and
disinformation across the internet because right now all of the terms
of service define these concepts differently and that makes it
extremely hard to implement an overarching solution related to them.
They ought to define and require that platforms obtain informed and
meaningful consent to terms of service, serving as an awareness-
building mechanism about data privacy issues and the limits of speech
on the platforms, because they are not free speech zones. They are
private platforms. They could also serve as a neutral appellate body
for users who feel their content has been unjustly removed, which right
now there is a really limited appeals process, as well as conduct
public audits of algorithms, account take downs, and data stewardship.
And then, finally, Congress also should consider the role of
education, in particular, media and digital literacy, critical thinking
skills and civics, and I totally agree with Matt's point that this
needs to be considered from an emotional point of view as well in
protecting online discourse and empowering citizens. Congress could
consider earmarks for grants for educational initiatives in this area.
I detailed a lot of this in testimony before the Senate Judiciary
Committee earlier this year, so you can look that up if you're
interested. \1\
---------------------------------------------------------------------------
\1\ https://www.c-span.org/video/?c4735072/nina-jankowicz-testimony-
senate-judiciary-committee; https://www.judiciary.senate.gov/imo/media/
doc/06-12-18%20Jankowicz%20Testimony.pdf
---------------------------------------------------------------------------
But I'd also be interested in exploring the use of taxes or fines
paid by social media companies to fund such initiatives. The U.K. is
looking into this right now. Ultimately, these are generational
investments and ones that Congress needs to begin now because no
regulatory or oversight solution can be complete without an informed
and discerning electorate, in my opinion.
Finally, I think it's just important to note the critical awareness
building and oversight role that Congress can play even without the
passage of new legislation. It was pressure from investigative
journalists and Congress that led the social media platforms to begin
to reform over the past couple of years and that should continue in the
new Congress.
Thank you very much.
Mr. Toner. Thanks very much. Those were three very good and very
comprehensive presentations. I don't know if I should run from the room
screaming with some of the--you certainly broadened my awareness and
I've covered or looked at this issue for many years. But the extent of
the sea change, however you want to frame it--seismic change--in the
way we communicate is sometimes overwhelming.
I'll start with a question. Someone said--you've addressed it,
Nina, and then others as well--but to what extent is this a question of
a lack of competition in the industry and, really, a lack--because of
that lack of motivation on some of these companies to make the kind of
changes such as greater transparency, such as sourcing political ads in
a way that show where they come from and what--and kind of buyer
beware, readers can see where they come from--to what extent is the
problem that, that you've got a few companies who dominate the space as
opposed to the internet of the 1990s where it was just big open space
that everyone thought would bring greater freedom of expression and
what not?
And my second question is maybe best to you, Matt, but others can
chime in, is how close are we with AI to having the tools to look at
content and trace it back to where it's coming from and provide that
transparency to users?
Mr. Chessen. Okay. So I'll talk about the second one first. So on
AI there [are] various aspects to this. When you're talking about the
automated detection of malicious behavior, this is something where
there is a lot of advances in this area. There's a lot of companies out
there now that are--their business model is basically around helping
companies and organizations identify this coordinated manipulative
behavior and they're using AI and machine learning techniques to do a
lot of this.
I heard a statistic from Twitter that just a couple years ago they
were only filtering about 20 percent of the accounts that were
suspended using automated tools. The rest were, basically, filtered and
reviewed by people. Now it's about 95 percent, and of that 95 percent,
75 percent of those accounts are actually suspended before they even
push out one tweet. So they're actually able to identify patterns of
behavior that are highly correlated with this malicious behavior before
people are even pushing out content. So there's a lot more use of these
types of tools.
Now, the problem is that when you're actually going and looking at
the content--so a lot of the success in this area is in a very content-
neutral type of way. They're using these tools to really identify these
patterns in behavior where different accounts that may not seemingly be
connected to each other are in fact engaging in coordinated
manipulative activity and some of these platforms are actually changing
their terms of service to specifically ban that type of activity.
They're not necessarily looking at content in the first instance.
There's a lot of challenges with that because AI systems just aren't
that good at extracting meaning yet from the content. And so this is
going to develop over time. The problem we have, really, right now is
either over fit issues or under fit issues, right.
So you're either basically suspending accounts for journalists who
are talking about ISIS as well as the actual people that are promoting
the ISIS ideology, or you have an under fit issue where you're not
getting enough of the extremist content that's out there, right, or
enough of the disinformation that's out there.
So this is when we start filtering based on content that's sort of
fraught with problems, right. What we don't want to do in this case is
do anything that censors legitimate speech, and we don't want to
undermine free expression online. But anytime you're starting to make
decisions about this content is good, this content is bad--Karen, you
mentioned sort of this whitelisting, blacklisting--I think if you do
that off of objective criteria that's one thing. If you start going
into what the content is and making content decisions, that becomes a
lot more problematic, and right now AIs are just not good at figuring
out what the meaning is in content well enough to have sort of these
automated filtering tools that are just going to basically--that's not
going to solve the problem for us.
Amb. Kornbluh. Just a minor addition to that. I think that's why
it's useful, what I was trying to say, is to focus on the outlet rather
than the content and the practices of the outlet, and even better,
transparency as a key practice of the outlet so that you're not doing
that. I think the problem with what they're doing is it doesn't get at
some of these issues where the First Amendment talks about the press
and I think online we treat the press as just another speaker, and I
think having a legitimate press as part of our information integrity is
incredibly important.
And if you have an outlet like the Daily Caller, the Koch Brothers,
apparently, fund--this just came out over the weekend--close to 100
percent of the Daily Caller Foundation, which supplies the content for
what's, in large part--not completely, not every piece of content on
there is false--but a lot of conspiracy theories. The birther stuff
came out there, the caravan stuff. We don't want to be looking at the
content and saying, this is false--this is true.
But you step back and that outlet functions in a very different way
from what we consider traditional press and we have to have some
objective fashion to get these--these platforms should welcome some
objective fashion of getting them out of the business of deciding
whether or not an outlet like that is going to be treated exactly the
same as the Washington Post or The New York Times. But they haven't
welcomed that and so now they're up in front of the Judiciary Committee
again, being accused of censoring.
Ms. Jankowicz. One quick update. On my way here, I saw that
Facebook has now updated its ad policies related to publishers, so news
organizations that are buying ads to promote content related to
political or issue organizations. So they're no longer going to be
subject to the same disclosures, which is a good development. It should
have been that way to begin with and I'm glad that they've heard the
outcry.
On the AI point that Matt was talking about, there are a couple of
examples of AI false positives related to the ad transparency that I
have seen. There was a researcher who is at the University of
Pittsburgh, I believe, who has a podcast about Russia and was talking
about how Russians view Trump, and he was not allowed to place that ad
for his podcast because it had the words Russia and Trump in it, and
even when he appealed a number of times to a human being, still, he was
not allowed to place that out without becoming a certified political
advertiser. So that's one of those wrinkles that the platforms don't
seem to be keen to figure out.
And then going back to Mark's initial question about whether these
platforms just have too much market share, I certainly think it's a
problem. In preparing for this panel, I read a little bit about the
jurisprudence related to antitrust law, and Karen touched on it. It
seems like new regulations are going to need to be developed if there's
any way forward in that regard.
Karen, maybe you want to speak about that a little bit. I don't
know. [Laughter.] It's a big thorny issue. It certainly seems to me
that Facebook should not have the market share it has not only with its
billions of users but with WhatsApp and Instagram under its wings as
well. And that's all I'll say about that.
Mr. Chessen. I just want to mention that point on lack of
competition. I mean, I think I haven't made up my mind on this one. But
I would like to point out, you know, who remembers MySpace? Who
remembers Friendster, right? So they dominated--I'm surprised. It looks
like a younger crowd tonight. [Laughter.]
So, you know, they dominated the social media space and then
Facebook came along and out-innovated them. And I don't see necessarily
major obstacles to other companies coming in and out-innovating
Facebook. I think there are some concerns around sort of things like
data portability, right, and the fact that probably one of the most
valuable things about Facebook for a lot of people is just the network
of contacts that you build and that would be very hard to replicate on
any other platform.
So I think there are things that can be done in maybe a regulatory
context or, you know, through legislation that could basically improve
the opportunities for other companies to basically have the ability to
innovate and compete. But I am also not necessarily convinced that
Facebook and Google or Twitter or anyone else is not vulnerable to the
type of disruption that Friendster and MySpace experienced.
Mr. Toner. Paul.
Mr. Massaro. All right. Great. Well, thank you all so much. I have
two sort of totally unrelated questions, so I'd like to ask the first
and then get some answers, if anyone wants to bite, and then I'll ask a
second.
And my first actually deals with an innovation topic, and that's to
say: You know, it seems to me like a lot of the issue is how these
platforms are monetized and that was through ads, right, essentially
through the use of user data in some capacity. As consumers sort of get
hip as to what's going on and as platforms come under fire for
exploitation, have we seen any trend toward subscription-based models,
other forms of alternative monetization that would sort of reduce the
problem in just kind of like a, just a standard sort of creative
destruction kind of pattern? I know WhatsApp at some point had like--
you pay a dollar a year. And obviously it's hard to do ads with an end-
to-end encrypted service, even if it's owned by Facebook.
Ms. Jankowicz. I think I would say we've seen kind of the opposite,
at least in Europe, with the GDPR [General Data Protection Regulation]
regulations, in which people can opt out of their data being used for
advertising and still use the service. In fact, it's a requirement of
the legislation that people can opt out and still use the service. It
just--how many people do that, because it affects your experience, and
how much understanding there is around that I think is a separate
question. But no, I'm not sure anyone would pay for social media, the
same way that very few people pay for news these days.
Amb. Kornbluh. I also--a lot of people talk about how the problem
is that ad-
supported--the ad-revenue-supported model, and I think there's a lot to
that. One response that you get is, well, if you want to constrain that
in any way, you're going to hurt small businesses, and there's a lot of
talk about the commercial implications. And I think one of the
reasons--I just want to take a step back for a second. One of the
reasons that the Cambridge Analytica scandal I think hit so hard was
because it's one thing for these big murky data brokers to have
information on my shoe size and my taste in shoes. I might prefer that
they advertise better options to me on shoes, instead of the same pair
of slippers that I bought 2 years ago that follow me around the
internet. But when it comes to my political beliefs and my
philosophical beliefs and they're inferring that and using that to
micro-target propaganda at me, I think people felt very, very
differently. And I think we haven't teased that out enough. And it's
driven by the ad system, but there's no reason that the platforms need
to treat those two in the same way, the commercial advertising and the
political advertising.
And Helen [sic; Elizabeth] Denham, the information commissioner in
the U.K., has suggested even a pause in micro-targeting in the
political space until we figure this all out. I don't know if that
would pass First Amendment scrutiny here, but it's really interesting.
And one of the things--Nina mentioned GDPR. One of the things that
European law does is, it treats your political and philosophical views
as sensitive information. So, theoretically, they're supposed to come
to you and say, an opt-in basis, is it okay if I infer your
philosophical views by the magazines that you subscribe to in order to
micro-target propaganda to you? I've yet to see that question be asked
of anybody, but theoretically----
Ms. Jankowicz. It should be asked exactly that way. [Laughter.]
Amb. Kornbluh. Exactly. Theoretically that should be a brake on
some of this.
But I do think it's useful--in the U.S. it's really difficult
because we don't--we affirmatively don't want to burden political
speech. But there has to be some way we can think about the whole ad-
revenue ecosystem in the political sphere, differently from are small
businesses going to be able to target their users online?
Mr. Chessen. So just a couple points on that, also. I mean, I think
money generally is something we really need to take a close look at.
You know, part of--there have been studies that showed that basically
what drives engagement on some of these platforms is anger, right? And
so in a lot of ways, that's what some of these algorithms are actually
optimizing for, because it keeps you on the platform longer, then
you're looking at more ads. There's also this phenomenon where there's
certain actors who are pushing out disinformation, not because they
necessarily care about the content or the substance of the
disinformation, but because they're trying to basically earn ad
dollars. And they're either trying to get the revenue from the clicks,
or they're actually trying to drive you off of the platform to their
own site, where they then essentially put adware slash spyware on your
computer that then can track your behavior throughout the internet and
then they can sell you information. So there's some malicious ad
networks that these directly plug in to, and there's some
disinformation actors, where there is this nexus--there is--I know one
in particular where there's a nexus between the ideology but it's also
a for-profit, money-making enterprise. Right? So I think we have to
tease that out.
We also have to look at individual choices here, because all of our
individual choices are really dragging this system. We've all got
addicted to free stuff on the internet. We all have gotten addicted to
free news. And so the market and the systems have responded, and so
basically when you have clickbait-driven ads and clickbait-driven
revenue for news sites, you know, that's sort of eviscerated local
media, it's eviscerated a lot of the editorial departments, and it's
incentivized this sort of yellow journalism and clickbait-driven news.
So we have to think about how we as individuals are actually
incentivizing this.
And I think the last thing to think about is, is I'm sort of
intrigued by this idea that people are throwing around of the public
utility for social, right? So if you think about why some of these
public utilities came around for water or electricity, it's because
those entities that were providing those services, their interests
weren't necessarily directly in line with the users, and there were
some negative incentives there. So maybe that's something we need to
consider is, do we need some sort of new institution for democracy that
basically serves this type of civic function? And it wouldn't
necessarily be social media; it could be something that basically
serves other functions, like connecting people better to government or
enabling, you know, Members of Congress to better connect to their
constituents.
Mr. Massaro. So I thought my second question was not related and
then you both brought up the EU, so the second question is: The EU
approach--you know, they've taken a very proactive stance on a lot of
these issues. What are the sort of positives and pitfalls of what's
going on there, in your view?
Ms. Jankowicz. Well, I think it's too early to say, really, what
GDPR has done for privacy- and disinformation-related issues in the EU.
But one thing that I have been looking at and I'm going to mess up the
name--I forget exactly what it's called, but the EU in September signed
with the social media companies essentially a code of conduct related
to disinformation. And actually--was that just yesterday?--two days ago
in the U.K. the nine parliamentarians or members of Parliament who
convened for the DCMS [Department for Digital, Culture, Media and
Sport] select committee also signed a related document. And I think
that's something that prior to legislation coming forward should happen
here, just setting the rules of the road, defining what we mean by
disinformation, by hate speech, by all these things, and kind of
identifying a common set of core values going forward that can govern
the decisions that the United States makes legislatively. I think
that's something to aspire to. We'll see if the social media
companies--one of the clauses in this agreement they made with the EU
is that they need to report on their activities related to take-downs,
so we'll see next September if they actually do that and to what level
of transparency they do.
Amb. Kornbluh. As I'm sure you're aware, California passed its own
privacy law that's modeled to some degree on GDPR and now there's talk
about a Federal privacy law in the U.S. And I think there's going to be
a lot of conversation then. As Nina said, it's too soon, really, to say
about GDPR, in part because the enforcement actions haven't been taken
yet, and so there's been a lot of complaints about the way some of the
companies have been implementing GDPR, that they appear to give you
choice, but the way the user interface is designed, it looks like you
don't have a choice and so you click ``agree'' because you think that's
the only way to get the service, when in fact that's not what the law
says.
So we'll see how the enforcement goes, and that should become clear
so that we'll have some data points into how all of this is working and
what makes sense and what doesn't before the Federal law is passed. And
I think it will be really interesting to see how that debate goes and
what happens between now and then. The California law isn't scheduled
to come into effect for some time, so it gives us that window.
Mr. Chessen. So I think, generally, looking at the EU approach, I
think one of the things that's been really positive is the fact that
they have taken a very broad multi-stakeholder, consultative approach
toward these issues, so they had a high-level experts group that met
for several months that produced a report. The EC then took that and
actually put out some policy guidance. One of those pieces of guidance
was for this code of practice, which they negotiated with the social
media companies. There are some concerns that this could be--lead to
regulation or legislation that could de facto regulate U.S. companies,
so I think a lot of people are watching to see where this goes.
But I think the lesson from the EU approach that we might take is I
think we need to engage in a national conversation about these issues.
And so since we're here at Congress, you know, one of the things that
I've talked about is maybe Congress needs to convene a commission on
data privacy, information security and disinformation, because I think
we need to have a national conversation in this country about our data
privacy practices. I think we need to have a conversation about a lot
of these issues, about whitelisting, blacklisting, content controls,
what's acceptable, do we want to change the business models for these
companies? Do we need to regulate them to promote more innovation? This
is not a conversation we're having in any sort of--you know, with a
methodology or any sort of systematic way. And I think it's a
conversation we need to have before we sort of rush in and regulate or
create policy around this. And that's really what I see missing from
the U.S. discussion.
Mr. Toner. Thank you so much. At this point, I think we'd like to
open up to any audience questions. So if you've got questions, I know
we have a mic available. Did I see a hand come up?
Questioner. Yes. I can try without a mic.
Mr. Toner. You can try without a mic. [Laughs.] I can hear you.
Questioner. Matt, just to your--[inaudible]. To your point about a
commission, why haven't we seen tech executives sort of brought up to
the Hill the way the cigarette company executives were in 1994 and
really account in a much more comprehensive way for their sort of
complicity in a lot of this? We've had that, obviously, piecemeal, but
I think--I mean, I wonder why. Why has Congress not done that, do you
think?
Mr. Chessen. Well, I'm not going to speak for Congress. I mean, one
thing that I can guess is because I think in the difference with the
cigarette case is there was a sort of intentional, perhaps malicious
activity involved and some intentional disinformation there. I think
what you're seeing with the tech companies--it's not that they're
malicious. I mean, there's a lot of fantastic people working in the
tech industry--they're trying to build great products, they're trying
to build products people use. There's actually a lot of great people
who are very strong proponents of U.S. national security who work for
Facebook and other companies. I think it's more that they didn't
necessarily see these problems coming, like a lot of us didn't see
these problems coming. And so I do think there needs to be a lot more
accountability for, hey, now that you know that the problem's there,
what are you doing about? Right? But I don't think it rises to the same
level of, you know, sort of maliciousness or--it wasn't necessarily as
problematic. That's just my personal take; you know, my colleagues
probably have opinions as well.
Amb. Kornbluh. Yes. I mean, just having been involved since the
early regulatory discussions, there is this idea that this technology
itself was pro-democratic and that there was no need for regulation
because it was going to further the goals of democracy. It was instead
of broadcast, which was, you know, concentrated, that it was--you know,
everyone was a publisher, everyone was a speaker. It was going to give
voice to the voiceless, power to the powerless. And as things have
changed, as it's become more centralized, as our lives, our whole lives
have moved on to the internet, I think we still have that image of the
guy in his garage, and I think maybe we need to mature a little bit
more and say, gee, if our campaigns are all taking place online, maybe
we have to update our campaign finance laws. And if we're purchasing
online, maybe we have to update our consumer protection laws. I think
as a society we haven't taken the responsibility seriously enough.
Ms. Jankowicz. Just briefly, I'll also add, I agree with everything
that's been said and I think one of the reasons that we have been slow
to kind of coagulate around this issue is because of how we came to it,
through a very political lens, and I think it's really important to
take a step back and de-politicize this because it's a question of
democratic discourse. It's not partisan.
Mr. Toner. Another--there you go, over there.
Questioner. Is this good? Okay, sorry. [Laughs.] Growing up as a
Millennial/Generation Z, however you want to go and classify that, I
was able to see firsthand the way that social media is able to impact a
generation. And one of the scary things that I have seen is that people
are moving away from reading books and reading more legitimate news
sources like The Wall Street Journal, so forth, and going more toward
free, more interesting content, I guess, on social media. So I want to
know--and to me, the way that I see it is almost trading knowledge for
information. So I want to know, are the platforms partially themselves
responsible for creating this problem by giving us too much information
than what we're able to process and creating more extremist content by
that?
Mr. Toner. Sorry, could you give us your name and affiliation?
Questioner. Sure.
Mr. Toner. I know this guy over here so I can--[laughter]--I can
get it afterward. But----
Questioner. I'm Jake Hannigan, working for Congressman Tom
O'Halleran of Arizona's First District.
Amb. Kornbluh. I mean, there's been a lot of work by folks who are
looking at--to applying different sociology, anthropologies, psychology
to what's happening on the internet, and the platforms themselves are
starting to take this really seriously, so you'll notice that you're
now getting reports about how much time you've spent online. You know,
that's still a very blunt instrument; it's not telling you what you
spent your time doing. But I think there's starting to be more
awareness of this, just how many hours people are spending. What's the
impact on kids? You know, what's our impact on delayed gratification if
everything's a click?
My kids, I had them watch one of these former Google guys who's
done this whole presentation about what notifications does to you and
how it trains your brain. And my kids are very loath to take any of my
advice but they watched this and they turned off all their
notifications. I think there's starting to be an awareness, but not
nearly enough. I think it's a really interesting question.
Mr. Chessen. I would answer that by just saying yes. I mean, I
think that the technology is an enabler of all of us getting
information overload and having short attention spans. I mean, if you
just look at a news feed, right--so whether you--take your news app.
Right? So it used to be that you would read a newspaper and the
newspaper had a set amount of content, and when you were done with the
newspaper, you put it aside and you were done. News feeds basically are
endless, right? You keep reading, you keep reading. So that encourages
behavior where you would scroll and scan and have much less detailed,
in-depth engagement with things. And that sort of trains us in how we
consume things.
I think some other interesting factors are just, you know, the fact
that it was explained to me once when someone was talking about why
their kids were using Snapchat and just exchanging images with each
other. And it was explained to them, well, that's how they're talking;
they're actually communicating using that. And so, if you don't
understand meme culture and you don't understand how much information
can be spread during memes, which is just an image sometimes with text
with it, you know, that drove a lot of sort of the expression during
the 2016 campaign. You're seeing it drives a lot of expression online.
And this is around this idea, you know--you mentioned the images about
sharia law in Paris. Right? I think that was less about people actually
believing that those things were credible and more about the posture.
Right? And so this is what you're seeing a lot more of now, and I think
this is driven by technology, is it's not so much the content, the
substance; it's actually the posture that that content or that meme or
that image pushes out that reinforces someone's belief and identity.
So you're seeing these very profound changes, you know, that are
being driven somewhat by technology or enabled by technology. And I
don't necessarily think we know exactly where this is going to go. When
you start having AI systems producing a lot more content, images, TV
shows, things like that, it's going to change this dynamic further. And
so yes, pay attention in this space because I think there's going to be
a lot of dynamic knowledge generated, but I think it's also something
we need to pay attention to vulnerabilities in this space as well.
Amb. Kornbluh. Can I just add one other thing? Because I don't want
us to be all grim. There have been a couple of studies that have
actually shown that there's an upside, too, for kids and for teens
about social media. Vicky Rideout has done some amazing work, and there
was just another study out really recently about how when kids are
lonely they can find mental health services online.
So, you know, again, I think sometimes it's not the technology,
it's how it's used, how aware we are of how we're being manipulated and
so on, and that it can be a really valuable--I know for me, I read the
news all the time now, but I'm finding all kinds of specialized sources
that I wouldn't have found otherwise. You know, the whole Me Too
movement--people have talked about it, but just from a personal point
of view, watching young women speak up has changed my perceptions of
things. So I think there's a huge amount of good in this, but I think
your generation is going to be so much more aware of its impact and
it's going to be able, I hope, to mold it so we get more of the good
and less of the bad.
Questioner. Hi. Maria Yates [ph] from the Voice of America Russian
Service.
I just wanted to touch on Russian disinformation tactics. Do you
see Russia being, indeed, less active because of the public scrutiny?
Or do you see them developing new strategies or tactics for
disinformation, developing--[inaudible]--in the future?
Thank you.
Ms. Jankowicz. As far as I'm concerned, a lot of the Russian
disinformation that likely happened during the 2018 midterms has just
not been located yet. If you think back to November, December 2016, we
didn't know the extent of what was going on then. And I think it's been
shown through the criminal complaints and indictments that Mueller's
team has released that these efforts are ongoing and that they are--
because the rules and regulations on the platforms have changed,
Russians and also other malign actors, including Iran and China, have
been able to get around those regulations and be a bit more clever
about how they are putting out their information operations.
Mr. Chessen. I think the only thing I would add to that is, you
know, Russia is not giving up the information game. I think they,
maybe, are just going to be doing it in a way that is a little bit more
savvy and isn't going to--isn't going to generate the obvious public
blowback, but they're still committed to information operations. What I
think also we need to pay attention to is that there a lot of other
actors who have learned from how Russia behaved and a lot of their
tactics, and those actors are now adopting those tactics. There's a lot
of people who are more worried about domestic actors now pushing out
disinformation than they are about some of the international actors. So
that's particularly concerning.
Mr. Toner. We do have time for just a couple more questions, if
there are out there.
Questioner. Thank you. [Inaudible.]
I had a followup to my--[inaudible]--question. If we think of
Facebook as a social network from, like--[inaudible]--and now it is
this big company moving to the news gig, do you think maybe Facebook
moved too fast or maybe wasn't supposed to go on that market? And
what's your stand on other social networks? Should they go to the news
gig, too, or no?
Mr. Toner. Sorry, just to clarify, you said the news becoming kind
of a media----
Questioner. Yes, I wanted to----
Mr. Toner. ----or like a news outlet?
Questioner. Yes. Yes, do you think that maybe it was really a
mistake of Facebook to go that fast to the news market?
Mr. Toner. Market, right.
Questioner. Yes.
Amb. Kornbluh. I'm troubled by what's happened--[laughs]--to the
press online and I'm not sure that--you know, I'm not sure that I would
blame Facebook as much as I would blame society for not thinking more
about this. But, you know, a lot of these outlets felt the only way
they could--they needed to go where the eyeballs were; the eyeballs
were online. They started to go online but they lost--as I was saying
before, they lost a lot of the information that lends to their
credibility. And so you've had this flattening where--and this due to a
lot of causes but where the mainstream news has gotten less credible,
and some of these conspiracy sites have gained by looking like normal
press outlets. They've gained credibility. And that's just really
unfortunate.
And obviously there have been a lot of causes that have been
leading to that, but if you think about in the physical world where you
had a newsstand and maybe the tabloids were separated, maybe they were
in a different place behind the checkout counter or they were in the
back of the kiosk, and now it's all muddled together. I mean, you don't
even know if it's an opinion piece from The Wall Street Journal or a
news piece when you see it online. So all of that, that framework that
we had in our brains that allowed us to trust a piece of news, we're
now realizing we have less trust, but we also are still--the remaining
trust we have we're attaching to these other outlets that are just
conspiracy sites, so it's really troubling.
And as Nina laid out, there's this effort to try to dial some of
that back by Facebook where they're emphasizing your friends and family
in your feed instead of news, but I think we have--we've created quite
a mess and it's going to be really hard to get out of it.
Ms. Jankowicz. One small anecdote that I'll relay from a recent
conversation I was involved in with Facebook and a number of European
governments: One of the advisers to a prime minister of an Eastern
European government stood up and asked the representative from Facebook
what--this is analogous to the news platforms but also, you know,
information coming from governments--he said, Since you changed your
algorithm, the content that we need to deliver to our constituents is
no longer getting the engagement that we would hope that it gets. Are
we expected to buy ads to reach our constituents? And the Facebook
representative actually suggested that this adviser to a prime minister
create a Facebook group in order to reach their constituents, because
that was prioritized in the algorithm. And I think it's really crazy
that legitimate information from governments, from news sources is
being demoted in order to just kind of feed this engagement mechanism
on Facebook. And I don't think they've really thought that through,
that, you know, people might be fed more and more content from Ariana
Grande rather than their elected officials. So it's something I agree
with you that they probably moved too fast in that scenario.
Amb. Kornbluh. But I do like what Matt was saying and I've been
thinking about that as well: What's our civic infrastructure? You know,
it's one thing for us to complain, it's another thing for us to think
about, what do we want to actually see? What are the positive
infrastructures that we want to see? And people are just starting to
put that together and think about that. And then how do we get that--
how do we make sure that that actually gets eyeballs, that's attractive
enough for people to actually go on to it--so what's the PBS or the C-
SPAN or whatever your analogy is?
Mr. Toner. It's in some ways a problem of curation, which is what
you talked about--you know, that you know what you're getting with The
New York Times, you know that it's been looked at by an editorial staff
and positioned correctly, where the bad stuff has been thrown out and
culled. And you don't have that----
Amb. Kornbluh. It's not so much that you know, gee, I know that
editor and I trust him, but you know what their standards and
procedures are.
Mr. Toner. Exactly. And their reputation's on the line. Yes.
Amb. Kornbluh. Right. So, you know, I'm sitting over here looking
at Yahoo! News and there's ProPublica. There are a bunch of very
credible outlets that have come online since we've been reading all our
press online. You don't have to be a physical newspaper, but you need
to know what procedures they use.
Mr. Toner. Agreed.
Amb. Kornbluh. Yes. Yes, yes, yes. So I think that kind of
credibility--how do we establish that kind of credibility so that we
can have a public discourse online? And that's going to--you know, we
have to think about things like--you know, government has thought of
itself as a provider of certain facts. You know, in the Constitution
you find reference of the census. The government has put out scientific
studies and facts. You know, how are we going to help people to find
that and to have this fact base I think is a real interesting
challenge.
Mr. Toner. Okay. Go ahead, Kyle.
Questioner. Thank you all. I'm Kyle Parker, Helsinki Commission.
On one level, there's an optimism I hear up here. But it's not
where I think the optimism should be. I sense and take this view that
politicians meddling in political things, don't trust them, don't trust
the government, you know, don't trust Facebook either, but probably
trust Facebook more than the government, and don't want the government
to wreck things.
A lot of these proposals for solutions sound to me like so much
manipulation to correct other manipulation. I don't trust it, and I
feel like the mess you're talking about has a real bright side and I
hope it gets even messier. I wonder if we're not simply reacting to
what a democratic or a more democratic free-for-all public square
actually looks like. And it's uncomfortable to us, particularly for
those of a generation that watched Cronkite at 7 o'clock and got their
news delivered that way.
Why are we so naive? Technology is not a good. Why this misplaced
faith in linear progress? To me it's about time that the scales fall
from our eyes. As that happens, it seems to me the real problem is
civic hygiene. Voting is serious business. And I don't trust The New
York Times' news or any of them. There's opinion involved in creating
the news.
So I'm concerned that these measures that aim to try to steer or
control or, as you mentioned, make news appealing. Yet again, more
Madison Avenue tactics to package information and get this to the
electorate in a way that they'll actually click on it or watch it--is
perhaps--creating of more problem, because the internet being what it
is--like you say, you ban something, you drive traffic to it, there's
mirror sites. You can't get rid of it.
Maybe the best solution is for people to think about civic hygiene,
improve critical thinking skills. Don't trust anything you read
anywhere. Be skeptical. Get off the internet and stop flattening your
civic engagement to this one-dimensional online medium and get out and
talk to your friends and neighbors. Be active in your local community
--this focus on hacking elections to me is an unhelpful reduction of
civic engagement to showing up for, what, a half an hour every few
years to vote? You call that civic engagement? There's a lot more to
civic engagement. To me our outrage just reeks of establishment fury.
Here we are in Washington, working for politicians. If you're a
politician, elections loom large in your mind. Well, they don't loom
nearly as large in the minds of the American people. There's a whole
lot more to this country and to real civic engagement. Yet I see these
bills come across my desk. I saw one last Congress; I think it was a
hundred million dollars to the Department of Homeland Security to
educate Americans about consuming news. This is frightening, Orwellian,
even. I don't want the government and politicians in this space. I want
more of the public square in this place, more of a free-for-all. We're
the ones who are ultimately the problem. We're the market for this
clickbait and we're the ones who click on it. So what do we expect from
Facebook and others? They do exactly what they're designed to do and
they do it well.
Amb. Kornbluh. So the one thing that I'll throw back at you is
where I come down again and again and again is the role for government
is transparency, that you can't--you can get all the education and
literacy that you want, but if you don't know if you're dealing with a
bot, if you don't know how many actual people endorsed this and it
looks like it's a lot more than it really is, if you think that
Blacktivist is an actual, you know, group of likeminded people but it's
actually the Russians, if you don't know that this travel ad is being
sponsored by somebody who has a conflict of interest, then you can't
look out for yourself.
So I don't think of that as a real public square if you don't have
the information, if you don't know what's going on. And people--
unfortunately, the internet, which we thought of as this great
transparent voice of the voiceless, there's far too much opportunity
for fraud, really. So I come down on, isn't there a role for the
government in transparency? And the Supreme Court, even Scalia, has
said that in political speech there's a role for transparency. So I
think that's a role for government. The government isn't saying what's
true, what's not true, what's good, what's bad, what should be
whitelisted or not; you know, they're just saying let's have--let's
give you more information about what you're seeing online. I think that
can help a lot.
Mr. Chessen. So, actually, a lot of what you said resonates with
me. I think the way I would conceptualize what you're talking about
is--taking it back to the Enlightenment again--is that you've got a lot
of institutions that are being run by elites. That's how we've run
things for 200 years. And now you have hyper-empowered individuals who
are using technology to push out a lot of ideas that are challenging
those elites and institutions. I think all of that is very healthy and
I think--I am a firm believer in free expression in the marketplace of
ideas, and people have a right in this country to put out
disinformation, if they want to.
Where I think it goes off the rails, and what we have to worry
about, is when it turns into coordinated, manipulative activity using
technology. Right? And so I think that's where we start to draw the
line. It basically goes to what Karen and Nina have said. You know,
it's basically when you have groups of people that are using techniques
from human cognition and human psychology and the technology tools and
then hacking the features of those technology tools to basically
manipulate people without their knowing they're being manipulated,
that's where it becomes a problem.
So I--but I totally agree that I think we are really missing out--
and Karen talked about this, a lot of the positive aspects of
technology and what it's doing for society. I think we can't lose sight
of that focus and that vibrancy and the fact that this diversity of
ideas is a good thing and we don't need to say that this is the death
of the Enlightenment, we need to create Enlightenment 2.0.
I would also caution that I think there's a difference between
promoting critical thinking and promoting people having sort of a
questioning eye, versus just saying don't believe anything you see.
Right? That's actually pushing us into this post-truth world. That's
what adversaries like Russia want, right? They want us in this post-
truth world, because in that world a fact is just whatever you can
convince people of, and then countries that may not be able to compete
with the United States diplomatically, economically, militarily, but
can compete on an information basis, they have outsized power. Right?
And so that's the real danger. And like I said, that post-truth
world, where you don't believe anything, in that world people tend to
retreat into their tribal identities; they tend to retreat into their
little window of what they believe about the world and they interpret
everything through that because they just don't know what to believe.
This is really fundamentally a trust issue. Right? And this gets
back to the transparency. People need to know what they can trust. They
need to know where the information's coming from, who's doing it, and
they need to be able trust that the information they're getting is
being provided to them in a way where they're not manipulated.
Questioner. Just to clarify: The point is not ``don't believe
anything you see'' but it's ``don't believe anything you see without
constant critical reassessment.'' So it's not a one-time thing where
[you] trusted this paper 20 years but everything that's coming in is
vetted through a notion of is--you know, a critical----
Mr. Chessen. Keep your brain on.
Questioner. Right. [Laughter.]
Mr. Toner. I think we're going to have to conclude on that very
good exchange. But I can't thank our three panelists enough for really
a very illuminating discussion. I've learned a lot today. I hope all of
you have. I think this is a really useful exchange. And I hope, as I
said, it's just the beginning.
But thank you so much for coming today. I appreciate it. Thank you
on behalf of the Helsinki Commission. [Applause.]
[Whereupon, at 12:04 p.m., the briefing ended.]
[< < <
This publication is intended to document
developments and trends in participating
States of the Organization for Security
and Cooperation in Europe [OSCE].
< < <
All Commission publications may be freely reproduced,
in any form, with appropriate credit. The Commission
encourages the widest possible dissemination of its
publications.
< < <
www.csce.gov @HelsinkiComm
The Commission's Web site provides access
to the latest press releases and reports,
as well as hearings and briefings. Using the
Commission's electronic subscription service, readers are
able to receive press releases, articles, and other
materials by topic or countries of particular interest.
Please subscribe today.