[Senate Hearing 117-888]
[From the U.S. Government Publishing Office]
S. Hrg. 117-888
PLATFORM TRANSPARENCY: UNDERSTANDING
THE IMPACT OF SOCIAL MEDIA
=======================================================================
HEARING
before the
SUBCOMMITTEE ON PRIVACY,
TECHNOLOGY, AND THE LAW
OF THE
COMMITTEE ON THE JUDICIARY
UNITED STATES SENATE
ONE HUNDRED SEVENTEENTH CONGRESS
SECOND SESSION
__________
MAY 4, 2022
__________
Serial No. J-117-60
__________
Printed for the use of the Committee on the Judiciary
[GRAPHIC NOT AVAILABLE IN TIFF FORMAT]
www.judiciary.senate.gov
www.govinfo.gov
______
U.S. GOVERNMENT PUBLISHING OFFICE
56-698 WASHINGTON : 2026
COMMITTEE ON THE JUDICIARY
RICHARD J. DURBIN, Illinois, Chair
PATRICK J. LEAHY, Vermont CHARLES E. GRASSLEY, Iowa, Ranking
DIANNE FEINSTEIN, California Member
SHELDON WHITEHOUSE, Rhode Island LINDSEY O. GRAHAM, South Carolina
AMY KLOBUCHAR, Minnesota JOHN CORNYN, Texas
CHRISTOPHER A. COONS, Delaware MICHAEL S. LEE, Utah
RICHARD BLUMENTHAL, Connecticut TED CRUZ, Texas
MAZIE K. HIRONO, Hawaii BEN SASSE, Nebraska
CORY A. BOOKER, New Jersey JOSH HAWLEY, Missouri
ALEX PADILLA, California TOM COTTON, Arkansas
JON OSSOFF, Georgia JOHN KENNEDY, Louisiana
THOM TILLIS, North Carolina
MARSHA BLACKBURN, Tennessee
Joseph Zogby, Chief Counsel and Staff Director
Kolan L. Davis, Republican Chief Counsel and Staff Director
SUBCOMMITTEE ON PRIVACY, TECHNOLOGY AND THE LAW
CHRISTOPHER A. COONS, Delaware, Chair
SHELDON WHITEHOUSE, Rhoade Island BEN SASSE, Nebraska, Ranking
AMY KLOBUCHAR, Minnesota Member
MAZIE K. HIRONO, Hawaii LINDSEY O. GRAHAM, South Carolina
JON OSSOFF, Georgia JOSH HAWLEY, Missouri
JOHN KENNEDY, Louisiana
MARSHA BLACKBURN, Tennessee
Cassie Field, Democratic Staff Counsel
Julissa Milligan Walsh, Republican Staff Counsel
C O N T E N T S
----------
OPENING STATEMENTS
Page
Coons, Hon. Christopher A........................................ 1
Sasse, Hon. Ben.................................................. 2
Klobuchar, Hon. Amy.............................................. 4
WITNESSES
Haidt, Jonathan.................................................. 12
Prepared statement........................................... 35
Harper, Jim...................................................... 14
Prepared statement........................................... 47
Keller, Daphne................................................... 10
Prepared statement........................................... 70
Persily, Nathaniel............................................... 8
Prepared statement........................................... 92
Silverman, Brandon............................................... 6
Prepared statement........................................... 107
PLATFORM TRANSPARENCY: UNDERSTANDING
THE IMPACT OF SOCIAL MEDIA
----------
WEDNESDAY, MAY 4, 2022
United States Senate,
Subcommittee on Privacy, Technology,
and the Law,
Committee on the Judiciary,
Washington, DC.
The Subcommittee met, pursuant to notice at 2:03 p.m., in
Room 226, Dirksen Senate Office Building, Hon. Christopher A.
Coons, Chair of the Subcommittee, presiding.
Present: Senators Coons [presiding], Klobuchar, Ossoff,
Sasse, Hawley, and Blackburn.
Also present: Senators Padilla and Cruz.
OPENING STATEMENT OF HON. CHRISTOPHER A. COONS,
A U.S. SENATOR FROM THE STATE OF DELAWARE
Chair Coons. This hearing will come to order. I'd like to
thank all of our witnesses for participating today. I'd also
like to thank Ranking Member Sasse and his staff for working
with mine to put this hearing together on a consensus basis.
This is a critical topic, and I'm looking forward to a
productive conversation today.
Social media companies, as we all know, play an enormously
important role in our lives and in our society. They have
helped to connect billions of people across the world and
deliver a whole range of new and innovative services in ways
that provide tremendous value to individuals, families, and
communities. At the same time, there are critical questions
about the potentially negative effects these platforms may
have--concerns about propagation of misinformation, incitement
to violence, or serious impacts on self-concept or mental
health.
A central issue that we face in confronting these
questions, whether as a consumer, as a parent, or policymakers
is what are the facts? What are the actual facts? Right now, we
don't really have a well-grounded data-driven understanding of
how social media platforms are impacting us in our society. The
reality is the vast expanses of human interaction that occur on
any given large social media platform can be studied and
analyzed really only by that platform. It's a problem, and it's
why, in my view, we need to promote greater transparency.
Transparency is important for consumers and the public who
should know how a platform they're choosing to engage with is
potentially affecting them. Transparency is important for
policymakers so we can better write whatever rules may regulate
these platforms going forward. And transparency, I would argue,
is important for the platforms themselves, because transparency
itself is a positive means to promote change.
Greater transparency can address the deeply-rooted market
failure of imperfect information. We've seen how public
disclosure of internal research or other data can create a
conversation, can alter consumer behavior, or lead to
regulatory scrutiny, and then initiate market pressure that
leads platforms to adjust their behavior in potentially
positive ways.
On top of all of that, I think pursuing greater
transparency is, and should be, nonpartisan. All sides, all
points along the political spectrum, have questions that can be
answered by greater transparency. Some of the most pressing
topics, like the effects platforms have on our children's
mental health, transcend political affiliation. This hearing
will explore how to bring about greater transparency. What are
the right pathways? What are the questions that we have and we
need to ask and answer?
Last winter--late last winter, I released a discussion
draft of the Platform Accountability and Transparency Act,
known by the catchy acronym PATA with Senators Portman and
Klobuchar. That draft bill would provide for transparency by
creating, at first, a provision to require platforms to make
certain key information available to the public on an ongoing
basis--high-level information about ads, and algorithms, and
widely viewed content. Second, a mechanism for data access by
truly independent researchers, through the National Science
Foundation and Federal Trade Commission. And third, a safe
harbor so that researchers conducting research in the public
interest need not fear legal action from platforms.
We released this bill as a discussion draft, knowing that
it raises important questions. I look forward, in this hearing,
to discussing those questions as well as many others, alongside
my Ranking Member. I think those questions include making sure
we're striking the right balance between privacy and
transparency, and appropriately also weighing what colleagues
are doing on these issues in Europe, and what legislation may
arise in the States.
Of course, PATA is not the only bill out there to address
these issues. I want to thank some of my colleagues, Senators
Blumenthal and Blackburn, for example, for their work on the
Kids Online Safety Act, also relevant to this discussion.
With Senator Sasse's cooperation, we have assembled an all-
star panel, with a diversity of views and perspectives, to
grapple with these questions today, and I look forward to it.
I'll introduce you shortly, but now I'd like to turn to my
colleague and friend, Senator Sasse.
OPENING STATEMENT OF HON. BEN SASSE,
A U.S. SENATOR FROM THE STATE OF NEBRASKA
Senator Sasse. Thank you, Chairman, and thank you to all
five of our witnesses for participating today. I want to
applaud Chris for his leadership, and his team, for focusing on
an incredibly important issue. We're here to talk about
transparency on social media, which is another way to say how
do we understand the role social media platforms play in our
lives, in our families, and in society more broadly?
There's no question that social media has changed American
society. My colleagues on both sides of the aisle have serious
concerns about how social media has changed the country,
although their concerns are often quite diverse in what they
think they're aiming at. My Democratic colleagues are often
concerned with whether social media platforms take down enough
posts that they think are harmful. And many of my colleagues on
this side of the aisle spend a lot of time arguing that social
media companies should stop censoring speech with which they
differ. And most of the time, the people making these arguments
are talking far past each other, and the sense that there's
consensus about what to do is really just consensus that
something is wrong, because we can't even agree on some of the
basic facts about what content moderation is, how it works, how
it should work. It is not surprising that we can't see eye to
eye on these facts, because frankly, we don't under the
platforms very well. Congress is often undereducated about how
these things work, and you're going to help us learn more
today, and we're grateful for that.
Social media companies have opaque policies that are
applied inconsistently, and there often seems to be one set of
standards for the wealthy and the well-connected, and another
set of standards for everyday Americans. As I've said before,
I'm pretty skeptical that turning to the Government is
necessarily going to fix many of these problems, but there are
things we should certainly start to do to understand the
problems better together.
What we do know is that a lot about social media is
complicated and has negative effects. One of the most important
things we know is that social media companies have a business
model that exists to maximize engagement. It's a pretty basic
truth, but when we get to a lot of the screaming about what
legislation might or might not fix the problem, we first need
to admit that the business model of these companies is to
maximize engagement, and social media algorithms, and company
policies, and business decisions are ultimately designed to
keep people on platforms for as long as possible.
So, it should come as no surprise to many of us that social
media is habit-forming over time. For some, it slips into
something that should rightly be called addiction. And as we've
integrated social media more deeply into our lives and our
institutions, most of us haven't stopped long enough to ask,
what habits are we forming? What loves are we developing? And
are these good and healthy habits? How do our social media
habits impact our families and our kids, on the one hand, and
our politics and our political, and social, and cultural
institutions on the other?
So, let's start with families and kids. Ask just about any
parent of a teenager whether they worry about social media, and
whether it's impacting their kids, and almost all parents, but
especially parents of teenaged girls, almost immediately begin
telling a story that's worrying to them, and they start talking
about things they've seen social media do to impact and
influence their kids.
Like a lot of parents, Melissa and I worry about our three.
We want to raise our daughters to be strong, confident women
who will love their neighbors and serve the world. We want them
to know their worth and understand their identity, their
purpose, and their value, which is infinitely more than you'll
ever get on social media's infinite scroll.
I've often joked that I'd prefer my 11-year-old son steal
the keys to the pickup and go joyriding with an open bottle of
bourbon than to wander with a smart phone, unmanaged,
unattended, on the internet. It's a joke, but it's not a joke,
because these are serious, lifelong, identity-warping forces at
play. We have also seen a lot of new neurobiological research,
which I won't try to summarize in this slot because we have
Jonathan Haidt here, who is much better qualified than almost
anybody on earth to talk about this.
Social media also poisons our political discourse. One of
the most important ways that social media platforms maximize
engagement is by sending users two powerful, unending messages:
You're right, and everyone who disagrees with you is evil.
Platforms show Americans the news, posts, and commentary that
they already agree with on the one hand, and they amplify the
most outrageous behavior by crazy partisans on the other side.
This creates echo chambers where political weirdos are mostly
just talking to political weirdos with their certainty that all
political weirdos are only on the other side. It turns out, the
weirdness is mostly driven by being addicted. It isn't mostly a
right/left problem. This is mostly an addiction problem, versus
the vast majority of people who don't want to be addicted.
Over time in these echo chambers, people tend to adopt more
and more extreme views, and then those extreme views are picked
up by the other side, which claims victory for having been
right all along, and having recognized that only the crazy
people--only people on the other side are crazy, and the
virility spiral, the polarization, continues.
As we step back and think about the role of transparency
about social media platforms, we should not lose sight of the
fact that we already do know quite a lot of basic introductory
things about social media, and how it impacts us, and how it
harms kids, families, and institutions, and how it polarizes
our politics.
Hopefully today, we can learn more, and we're grateful to
the five of you for helping us in that quest. Mr. Chairman.
Chair Coons. Thank you, Senator Sasse. Thank you for that
insightful opening comment and several powerful visual images.
I'd like to take the Chair's prerogative and invite Senator
Klobuchar to give some opening remarks. She's been a great
partner on developing the Platform Accountability and
Transparency Act, and she has a scheduling conflict that'll
prevent her from staying for testimony and questioning, but
wanted to make some opening remarks. Senator.
OPENING STATEMENT OF HON. AMY KLOBUCHAR,
A U.S. SENATOR FROM THE STATE OF MINNESOTA
Senator Klobuchar. I promise I'll look at the transcripts.
Thank you so much to both of you, and thank you, Senator Coons,
for your leadership on this bill with Senator Portman. Thank
you, Senator Sasse, for that great description.
I had one mom tell me that it's like she keeps trying to
get control of what her kids are seeing, but it's like water
that's coming out of a faucet that won't stop, and it's
overflowing, and she's just standing out there with a mop,
trying to do it, and she can't. And so, I think so much of this
has to do with the magnification, the amplification, of what
you have called this kind of polarized speech. I don't want to
get your words wrong, what you just said, that. Yes. Political
weirdos. Okay. There you go.
And I think what you see with that is that the influence of
these platforms is just simply unprecedented. Even though these
platforms are seemingly free, they're really not because you're
seeing advertising, and they're making money off of you, and
then they use the information they have, and then make more
money off of you. In fact, a number of even looking at the data
shows that they make more money off of us than other
industrialized nations, just simply because we don't have any
rules of the road in place.
For so long, we've been hearing, ``Trust us. We've got
this,'' and I think that that era of blind trust is coming to
an end. And that's why you see the work that's going on the
competition side.
I'm also on the Commerce Committee. That's why you see the
work that's going on with updating some of the kids' laws, and
that's why you're seeing the work that's going on on privacy,
and then, of course, on algorithms, and getting more
transparency.
You know, we all know that you can't yell fire in a crowded
theater. That is not considered free speech. That's not
considered okay. And the way I think about it is if you're a
theater, okay. Someone yells fire, it's not your fault, but you
better have exits so people can get out of there. But if you
had speakers, and you amplified that guy yelling fire in all of
your multiplex of theaters deliberately, that would be a
problem if you thought somehow you'd make more money. That is
kind of what we're talking about here with algorithms.
This hearing is also about how quickly misinformation and
disinformation can spread online. A recent report from the
Center for Countering Digital Hate identified a dozen specific
content producers as the original sources of an estimated 35--
65 percent of coronavirus disinformation online, where,
literally, people die when this stuff is not corrected.
We need transparency because these tools can hurt millions
of people. That's why this bill is so important, the Platform
Accountability and Transparency Act. It will make it so
companies like Facebook can't block researchers from looking
into their platforms and algorithms. I note that President
Obama touted the importance of this legislation in his recent
speech at Stanford, and I am committed, and I know Senator
Coons is, to seeing this through.
So, I want to thank all of you. I look forward to reading
your testimony, and I know that my staff will be here as well.
So thanks for coming before us today.
Chair Coons. Thank you, Senator. Today, we welcome five
witnesses to testify about the need, the value, the importance
of greater transparency in social media companies, and how that
can most reasonably be achieved.
Our first witness is Brandon Silverman. Mr. Silverman
founded CrowdTangle, a social analytics tool used to monitor
and understand how content is performing on social media
platforms in real time. CrowdTangle was purchased, acquired, by
Facebook in 2016, where it became an indispensable tool for
journalists, fact-checkers, and external organizations. Mr.
Silverman worked to implement and improve the product while at
Facebook until recently departing the company.
Next, we have Professor Nate Persily. Professor Persily
teaches at Stanford Law School and is the codirector of the
Stanford Cyber Policy Center, Stanford Program on Democracy and
the Internet, and the Stanford-MIT Healthy Elections Project.
We'll next hear from Professor Daphne Keller. Professor
Keller directs the Program on Platform at Stanford's Cyber
Policy Center, where her work focuses on platform regulation
and internet users' rights.
After that, we have Professor Jonathan Haidt, who is
testifying remotely today. Professor Haidt is the Thomas Cooley
Professor of Ethical Leadership at NYU's Stern School of
Business, where his work focuses on morality, emotion, and
politics.
Finally, we have Jim Harper, a nonresident senior fellow at
AEI, the American Enterprise Institute, where he focuses on
privacy issues and select constitutional law issues.
Let me briefly lay out the mechanics after we swear in all
of the witnesses. Each witness will have roughly 5 minutes to
provide an opening statement. We'll proceed to questioning.
Each Senator will have initially 5 minutes. Hopefully, we will
have time for a second round after that. So, if you would all
please stand to be sworn in?
[Witnesses are sworn in.]
Chair Coons. Thank you all. Mr. Silverman, you may now
proceed with your opening statement.
STATEMENT OF BRANDON SILVERMAN, FOUNDER AND FORMER CEO,
CROWDTANGLE, OAKLAND, CALIFORNIA
Mr. Silverman. Thank you for having me. My name is Brandon
Silverman, and I was the CEO and cofounder of CrowdTangle, a
social media analytics tool. We made it easy to see what was
happening on social media, and we were very good at what we
did. In 2016, we were acquired by Facebook, and over the next
few years, we partnered with thousands of outside organizations
around the world, to help them study and monitor social media,
including journalists, academics, human rights activists,
nonprofits, and more. By the end of 2020, the New York Times
called us perhaps the most effective transparency tool in the
history of social media.
However, in the spring of 2020, after an intense internal
debate about whether the transparency we provided was worth it,
Facebook paused all work on CrowdTangle and disbanded the team
running it. I left the company shortly after.
Today, the future of CrowdTangle is up in the air, and
along with it, one of the most robust and successful
transparency programs that Facebook operates.
I'm here today to share what I've learned working on
transparency for over 10 years, including what it's like to try
and advance transparency from inside one of these large
platforms, and why I believe it's time to stop sitting back and
hoping that platforms voluntarily share data, and to pass
legislation that makes it safe and responsible to share data
with the outside world.
In 2016, CrowdTangle was acquired by Facebook, and over the
next few years, we expanded our work to a wide variety of new
industries. We added new features and new data, and it became
one of the most widely used tools in the industry. Our data was
frequently cited here in congressional testimony over the
years.
There was no shortage of use cases for what we built, but
there was a limit to how far we could push transparency from
inside the company. The truth is, is that it's not easy to work
on transparency from inside a major platform. For one thing, it
can be incredibly uncomfortable when your work, and the work of
your team, are constantly fueling criticism, some fair and some
not, of the company where you work. Those normally take a toll
on your team, but they also make it harder to get resources,
they make it more difficult to launch new features and add more
data, and ultimately, they provide constant ammunition to
executives who are skeptical about doing transparency at all.
On top of that, there are also organizational challenges
that make it hard. There are conflicting regulatory and legal
requirements all around the world that are all interpreted
differently by whichever set of whatever team or lawyers you
might talk to. There's the fact that no matter how much
transparency you do, you're rarely going to get credit for it
in the public eye. There is a constant and fierce competition
for resources and the attention of leadership. And that's just
to name a few. But more than all of those factors combined, the
single biggest challenge is the plat--single biggest challenge
is that platforms can get away without doing any transparency
at all.
YouTube, TikTok, Telegram, Snapchat, these platforms
represent some of the largest and most influential platforms in
the United States, and they provide almost no functional
transparency into their systems at all. Moreover, despite a lot
of Facebook's laudatory transparency efforts, there are also
incredibly significant parts of their platform that still
remain entirely inside black boxes.
Today, CrowdTangle is still available, but it's in
maintenance mode. Facebook has stopped onboarding new partners,
no new features or major updates have been released in 2 years,
and the global partnerships team that used to run it no longer
exists. But while we were there, our team saw the power of
transparency. Almost every single day, our team saw examples of
partners using social media data to help protect elections,
prevent real-world violence, fight global pandemics, empower
independent journalism, hold platforms accountable, and more.
We saw how transparency can be a tool to make sure that social
media lives up to the promise of strengthening free and open
societies, instead of being used to undermine and weaken them.
I think what happened to CrowdTangle should be seen as a
bellwether, and that it's too hard to make progress on these
issues, at the scale and breadth we need, from inside a
company. As a result, we've seen that the industry as a whole
has simply not made enough progress equal to the
responsibilities they have. And I don't think there's any
reason we should expect that to change going forward. If
anything, I think we should expect less voluntary efforts.
That's why I believe it's time to create legislation that
makes data sharing and transparency a requirement for entire
industry, and creates mechanisms to do it in safe and
responsible ways. I think the Platform Accountability and
Transparency Act represents an important step in the right
direction. There are real challenges to getting this
legislation right, and we're going to hear about some of those
today. However, if we don't find ways to move forward, we will
continue to be in the dark about the real impact of social
media on our lives. We'll continue jumping from one anecdotal
data point to another, from one leaked document to another,
from one urban myth to another without ever establishing
baseline evidence-backed conclusions about the role it plays in
our lives.
When I think about social media and a lot of the public
discourse about the role it plays, I'm reminded of anecdote
about the policeman who sees a drunk man searching for
something under a streetlight and asks what the drunk man has
lost. He says he lost his keys, and they both look under the
streetlight together. After a few minutes, the policeman asks
if he's sure he lost them there, and the man replies, ``No.''
He lost them in the park. The policeman asks, ``Well, why are
we searching here?'' And the man says, ``This is where the
light is.''
For too long, our public discussion about social media has
been focused on the moments where we briefly found ourselves
under a streetlight. It's time to turn the lights on for good.
Thank you.
[The prepared statement of Mr. Silverman appears as a
submission for the record.]
Chair Coons. Thank you, Mr. Silverman. Professor Persily.
STATEMENT OF NATHANIEL PERSILY,
JAMES B. McCLATCHY PROFESSOR OF
LAW, STANFORD LAW SCHOOL,
STANFORD, CALIFORNIA
Professor Persily. Thank you, Chairman Coons. Thank you,
Ranking Member Sasse, Senator Blackburn, and my Senator,
Senator Padilla, for being here. I really appreciate testifying
for this Committee again. I'm going to talk a little bit about
the purposes of transparency, and then also, like Brandon, I'll
talk a little bit about my experience in working with Facebook
and other platforms in trying to get data out of them.
Let me begin by just saying sort of what the bottom line is
here, which is that we cannot live in a world where Facebook
and Google know everything about us and we know next to nothing
about them. All right? And these large platforms have basically
lost their right to secrecy. Their power over the information
ecosystem is really unrivaled in world history. Given that,
it's time for a democracy to step forward and regulate them in
the public interest. The least that we can do is to get some
transparency out of them so that we understand exactly what's
going on.
As Senator Sasse said, the questions with respect to
internet regulation are extremely complicated, very difficult,
and in many respects, we don't know what the right answer is in
many of these domains. But transparency is the first step to
finding answers in any of them. And we think about the purposes
of transparency, it's sort of a, if you'll excuse the pun, you
know, it is a meta-bill. Right? It is about enabling our
ability to regulate and to act in so many other domains.
First, let me say that these firms are information
monopolies. Usually, when we talk about it, we talk about it in
the antitrust sense, but you know, in terms of competition and
the monopoly power that they have over their economic domains.
But they're information monopolies in a different sense, which
is that they control all of the information which is now most
revealing about social problems. Right? We've never been in
that position before.
Those of us who are social scientists, when we would be
analyzing data in the pre-internet age, right, most of that
data was freely available, either through Government
statistics, survey data, or other kinds of data. Now most of
the data which is relevant to contemporary social problems is
locked up in these private companies. Right? And it's only
through legislation that we're going to be able to unlock it
and to find out exactly how big these problems are.
But I want to emphasize one thing, and I think Senator
Coons, you mentioned this before, which is yes, this is a very
difficult trade-off to get right. We need to balance
transparency with privacy. I think everybody on this panel, and
in the Senate, is aware of that. But the question is not
whether this data will be collected and analyzed, the question
is whether the only people who will be able to analyze the data
are the people inside the firms who are tied to the profit-
maximizing mission of the firms. All right? And so, the only
question that PATA and other bills like this are posing is
someone who is not tied to the profit-maximizing mission of the
firm going to have access and to do research in the public
interest.
So, let me talk a little bit about the purposes of
transparency. The first set of purposes and goals, I think, of
transparency, which is sometimes undersold, is that it will
actually change the behavior of the firm. To some extent, I get
criticism a little bit, when you emphasize transparency, that
it's seen as sort of weak legislation because it's not, you
know, breaking up the companies, or it's not going right after
content moderation. But once the platforms know that they are
being watched, it will change their behavior. All right? They
will not be able to do certain things in secret that they've
been able to do up to now.
Second, as Senator Coons mentioned, I think it will lead
them to change their products, right, because once we have a
greater appreciation for what's actually going on in these
firms, those on the outside can do research that a lot of the
insiders are not doing on their products.
Second, as you both mentioned before, this will educate
policymakers. It'll educate policymakers here in Congress,
whether the issue is child endangerment, or disinformation, or
antitrust, or privacy. It will educate our European allies who
are doing much more aggressive regulation right now. Now, as
well as in the States. Look at the bills coming out of Texas
and Florida. As you said, we are legislating in the dark.
Then finally, it will educate the public not just about
what's happening inside these firms, but also the dynamics of
the information ecosystem.
As Senator Sasse was saying, there is a fundamental debate
here as to what is going on. Is the problem, you know, runaway
hate and disinformation, and the like, on the platforms? Is the
problem platforms are overcensoring particularly conservative
voices? This is what transparency is going to answer. Right?
Those are the questions that we would be investigating.
I'll say just in conclusion that I spent probably 4 years
of my life trying to get data out of Facebook through a program
called Social Science One. The people we worked with inside the
firm were fantastic. They were with us all the way. The problem
was that ultimately, when it came down to providing the most
robust data that those inside the firm had access to, they
simply couldn't do it either because they are worried about
another Cambridge Analytica, or they simply didn't understand
whether the legal environment would be conducive to that.
And so, what I would say, sort of, in conclusion, is that
we shouldn't have to wait for whistleblowers to whistle. Right?
This is something--this type of transparency legislation is
about empowering outsiders to get a better idea of what's
happening inside these firms.
[The prepared statement of Professor Persily appears as a
submission for the record.]
Chair Coons. Thank you, Professor. Professor Keller.
STATEMENT OF DAPHNE KELLER, DIRECTOR, PROGRAM ON
PLATFORM REGULATION, CYBER POLICY CENTER, FREEMAN
SPOGLI INSTITUTE FOR INTERNATIONAL STUDIES,
STANFORD UNIVERSITY, STANFORD, CALIFORNIA
Professor Keller. Thank you for the opportunity to testify.
I work on platform regulation, which as Senator Sasse
referenced, is really complicated. It's exciting to be able to
testify about a topic where I think there's actually a viable
way forward. I'm particularly honored to speak beside Brandon
Silverman and Nate Persily, both of whom have built actual,
real-world, functioning transparency models. I'm also excited
to talk about this at a time when the EU has just moved forward
on platform transparency in a dramatic way under the Digital
Services Act, or DSA. I think that is really paving the way for
a whole new era of platform transparency.
I think today, we will hear about a real diversity of
transparency tools, and that range of tools is what we need.
Not every approach is useful for every research topic, or even
for the way that every platform functions. As someone who has
worked with multiple approaches to platform transparency, going
back over a decade, including in-house at Google, I would like
to find a way to enable a whole range of these approaches.
This includes fixing the laws that constrain so-called
scraping of data from public websites. This is kind of low-
hanging fruit and would have the side effect of helping with
some interoperability and competition goals. It also includes
building APIs, which are just channels for computers to talk to
each other, so people researching things like bias and
algorithms can submit bulk queries and look at bulk results. It
includes the kinds of mandatory disclosures contemplated in
laws like the discussion draft of PATA. All of this can vastly
improve public information, both about online harms and about
what regulatory responses will actually make things better.
To make these laws work, though, there are some pitfalls we
need to navigate. Some are practical, some are constitutional,
some are political. My written testimony on this was very long.
I'm sorry. It raised a lot of questions that I think
appropriately should go to an agency, because they are in the
weeds, and they're iterative, and they will change with
technology. But it also raised questions that I think are
serious policy decisions that are Congress' job to resolve. So,
I'm going to talk about those a little more here.
One is about privacy and surveillance. For researchers to
examine what is being said, and claimed, and propagated online,
and how platforms influence that, they will need to look at
information about people. There are unavoidable tradeoffs in
deciding how much they get to do that. Sometimes, I think the
need for public information should be paramount. Other times,
user privacy should. This is something where Congress should
provide guidance ideally through Federal privacy legislation.
If you have to resolve it here, you know, it deserves
attention.
When it comes to Government surveillance, though, so not
the researchers, but, you know, law enforcement getting access
to data, I think there is a brighter line. Nothing about these
transparency laws should change Americans' protections under
the Fourth Amendment, or laws like the Stored Communications
Act. I don't think that's anyone's intention here, but clear
drafting is essential to ensure that Government can't
effectively bypass Fourth Amendment limits by harnessing in the
unprecedented surveillance power of private platforms.
A second issue is about competition and the practical cost
and benefits of using different transparency tools for
different companies. Laws designed for a Google or a Facebook
are a bad fit for companies that are far smaller in measures of
revenue, or users, or employees. Those companies may not
actually be relevant for the problems that laws like this are
trying to solve. I don't think we need a live dashboard showing
us which hotels are most popular on Tripadvisor, for example.
Hopefully, the problems of designing rules for giant
incumbents, and then applying them to a whole competitive
ecosystem, speak for themselves.
The third issue is about CDA Sec. 230. I hate to bring it
up, but as I mentioned in my testimony, I don't think tying
transparency obligations to CDA Sec. 230 solves the problems
even that CDA Sec. 230's critics want to solve. I think,
instead, it sort of opens the door to very unpredictable
litigation.
The last thing is the First Amendment. I want transparency
mandates to be constitutional, but there are serious
challenges, and I hope that you will put really good lawyers on
that because I want this thing to work.
So, those are my concerns. In platform speak, these are the
things to solve before launch, and I'm happy to answer any
questions about the many other things I wrote about later on.
[The prepared statement of Professor Keller appears as a
submission for the record.]
Chair Coons. Thank you very much, Professor Keller. I
believe we're now going to have remote testimony from Professor
Haidt.
STATEMENT OF JONATHAN HAIDT, THOMAS COOLEY
PROFESSOR OF ETHICAL LEADERSHIP, NEW YORK
UNIVERSITY STERN SCHOOL OF BUSINESS,
NEW YORK, NEW YORK.
Professor Haidt. Hello. Greetings. I hope you can hear me
properly. It's an honor to be testifying here. Senator Coons,
you're absolutely right that this issue of transparency is
completely nonpartisan. Senator Sasse, you're absolutely right
that all of us with teen kids, myself included, are concerned
about this. We don't know what's going on, but the kids are
getting sick and we want to know, is this the cause? And
Senator Klobuchar, yes. The era of blind trust is coming to an
end.
I'm a social psychologist. I study morality. I've studied
moral development. I began to notice that something was going
really wrong with teenagers, entering college students, around
2014. By 2015, it was clear all of our mental health centers
were flooded on campus. A few years later, it became clear
there is an epidemic of depression and anxiety sweeping across
the United States for teenagers.
What I'd like to do in my remarks here is to make--I want
to give you a distillation of the facts that I have aggregated
on what's the nature of the teen mental health crisis. Then
second, what is the evidence, the empirical evidence, that this
is caused, in part, by teens suddenly moving onto social media
en masse around 2011.
All of my testimony, I submitted in the PDF file that is
submitted with my testimony, I have links to two collaborative
Google documents where I've invited other experts to critique,
to say what are we missing. We have relatively complete
listings of the studies on all sides here.
Let me begin. Part one. The specific, gigantic, sudden and
international mental health crisis. I'll just make these six
points. You can find elaboration of them with graphs and links
in the submitted testimony.
First, the crisis is specific to mood disorders. This is
not an across-the-board increase in all kinds of mental health
problems. It's specific to anxiety and depression, and the
behaviors that go along with them, which is especially suicide
and self-harm.
Second, the crisis is not the result of changes in the
willingness of people that self-diagnose. As late as 2018, some
experts were still saying, ``It's not real. It's just, you know
Gen Z, they're more comfortable talking about depression. It's
not a real thing.'' Now it's clear it is, again, because of the
very sudden increase in depression, I'm sorry--in hospital
admissions for self-harm and completed suicides. All of that
goes way up in the early 2010s. The self-harm lines are quite
dramatic. Relatively flat, and then right around 2010, boom,
they start skyrocketing upwards.
Point number three, it came on very suddenly. This was not
a gradual curve up.
Point number four, the increases are very large. We're not
talking 10 or 20 percent. Depending on what you look at, it's
anything from 50 to 150 percent. Self-harm for young teen girls
in particular is up more than 150 percent, in fact, 180 in the
last data that I saw. These are enormous, sudden increases.
Point number five, the crisis is gendered. Boys and girls
both have plummeted mental health since the early 2010s, but
the girls have fallen further on some measures. On others,
they're about the same percentage wise but the girls starting
rates of anxiety and depression are higher, so the increase is
a much larger number of girls. So, the crisis is gendered.
The sixth point, the crisis has hit many countries. It's
not just the U.S.A. The patterns in Canada and the UK are
identical to those in the U.S., and Jean Twenge, and I, a
professor at San Diego, have published a study showing that
actually, across the world, loneliness at school also went up
after 2012.
Those are the points that I want to make about the mental
health crisis. It is real. It is big. It is sudden. It is
gendered.
Now what's the evidence that social media is a cause? The
first point is that correlational studies consistently show a
link between heavy social media use and mood disorders, but the
size of the relationship is disputed.
Point two, you will often hear from experts, ``The size is
so tiny, it's no bigger than the correlation of mental health
problems with eating potatoes or wearing eyeglasses.'' But this
is not true. In the main study that found that, which I talk
about in the submitted testimony, that's the relationship for
all digital media use. This is very important. When you hear
people talk about studies, usually, it's about all digital
media use, including watching Netflix, with mental health
outcomes. Those correlations are, indeed, tiny.
But, and this is point number three, when you zoom in and
look just at the relationship between social media use and
mental health outcomes, the correlation is much bigger. There's
an emerging consensus that it's between r = 0.1 and 0.2. It's
in that range.
Point number four, those correlations are even larger for
girls. Probably closer to 0.2. r = 0.2.
Point number five, the effect size is even bigger during
puberty. A recent study showed between 11 and 13, girls are
especially likely to be harmed by being on social media. We
must try harder to protect puberty. Get this stuff out of
middle school, for God's sakes.
Point number six, correlations of 0.1 to 0.2 are not small.
This is what public health matters are mostly about, the
correlation of childhood exposure to lead with adult IQ is
0.09. There's a new realization amongst psychologists that
small effects played out over millions of people over many
years, add up to gigantic public health crises as we have with
lead and water.
Point number seven, there's experimental research as well.
This is not just correlational. The experiments mostly show
that if you randomly assign people to conditions, reducing
social media use tends to have beneficial effects.
Point number eight, the eyewitness testimony is there. We
asked the kids, as Facebook did, and what they found,
Facebook's own research found that teens blame Instagram for
increases in the rate of anxiety and depression. This reaction
was unprompted and consistent across all groups.
I'm going to conclude by saying, along with the others who
have been testifying here, we really, really need to see the
data. We must have platform accountability and transparency.
Imagine that all of our kids began eating a certain kind of
candy that had never existed before in 2011. All of them were
eating this all day long, and in 2012, they started developing
leukemia in large numbers. And some people said, ``Well, you
know, correlation is not causation. We can't be sure. We can't
be sure.'' Okay, fine. We can't be sure, but can we at least
compel the candymaker to tell us what our kids are eating?
Thank you.
[The prepared statement of Professor Haidt appears as a
submission for the record.]
Chair Coons. Thank you. I very much look forward to
continuing our conversation and appreciate the structured and
engaging way in which you delivered those remarks. Mr. Harper,
if you might, you're our final witness today.
STATEMENT OF JIM HARPER, NONRESIDENT
SENIOR FELLOW, AMERICAN ENTERPRISE INSTITUTE,
WASHINGTON, DC
Mr. Harper. Thank you, Senator Coons and Senator Sasse.
Thanks for the opportunity to testify today. This hearing, and
the Platform Accountability and Transparency Act, raise many
interesting issues.
When I finished writing my 22-page testimony, I thought I
had barely scratched the surface. I also thought mine would be
the longest, but I lost.
[Laughter.]
We're all in favor of transparency. I've done a good deal
of work on Government transparency over the years, as detailed
in my testimony, and I always hoped for more. The bulk of my
attention over the last two decades, though, has been on
privacy.
I'm sorry to play skunk at the garden party a little bit,
here, but there are very high privacy costs, I think, to the
mandated disclosure regime found in PATA. I see no limit on the
subject matter of data held by platforms that the legislation
affects. It would allow the National Science Foundation and
Federal Trade Commission to take essentially any data from
platforms, to give to approved researchers. In my opinion, this
runs contrary to an emergent property rights regime that is an
important protection for consumers and their privacy.
My belief is that platforms and communications providers
make contractual promises to protect privacy in their privacy
policies and in their terms of service documents. In doing so,
they also divide up nascent and property rights in data. In the
bundle of sticks model of property taught in law school, the
right to exclude others from personal information remains with
the consumer, subject to narrow exceptions like protection of
the platform itself, response to law enforcement, and so on.
Across the legal landscape, there are strains of
recognition that this is the case, including in the Supreme
Court. If congressional legislation took data from platforms to
give to researchers or anyone else, it would cut against the
grain of this trend. It would treat data held by platforms,
including data held as a bailment or in trust for users, adds
so much soup in a tureen waiting to be ladled out. The legal
immunities in the PATA legislation further erode confidence
that privacy will be protected in a mandated disclosure regime.
An unconstrained disclosure mandate may also be
unconstitutional. The Supreme Court appears likely to revive
the nondelegation doctrine soon, and in an unrestricted grant
of authority to mandate disclosure for the all-purpose, but
unfortunately nebulous goal, of transparency, may not survive
it.
That's not just a constitutional point, I don't think, but
it's a prudent one. We don't know what the future holds for our
politics, and we don't know what type of data a future
administration might use the NSF and FTC to order from
platforms about Americans, our habits, and our communications.
I also specifically recommend, in my written testimony,
against bringing transparency to platforms' moderation
practices and to their security efforts, both how they secure
themselves and how they provide security to their users. The
PATA legislation considers this. It provides for privacy and
security regulations pertaining to the mandated disclosure of
data, but the mandatory disclosure regime itself would increase
vulnerabilities and the attack service, as they say, in
computer science, of the systems. Moderation and security
systems are constantly probed by bad actors, who want to turn
platforms to their own ends or just ruin them. Mandated
disclosure would create new opportunities for wrongdoing.
I'll say this, since reading my colleagues' testimony and
some of the authorities they cite, I'm less enthusiastic about
my written testimony on the First Amendment issues. I am taken
by the breadth of the data seizures made by possible by PATA,
and my frame of reference is Fourth Amendment privacy. There
are areas where Government requires businesses to disclose or
publish information about themselves. This is not the case,
though, I think, where that includes the editorial choices of
institutions engaged in free speech.
The internet and social media are strange, but real,
descendants of the printing press, disembodied and given to
everyone to use as much as they want. Social media companies
aggregate and augment this mass exercise of expression. I think
it's well within precedent to regard what they do editorially
as protected by the First Amendment.
Where disclosure mandates exist, you can probably find a
close fit between means and ends. When auto dismantlers are
forced to reveal their inventories and sources, that makes them
much less likely to become fencing operations. Requiring
audited financial statements from public companies prevents
various forms of fraud. So far as I'm aware, and unfortunately,
platform transparency through disclosure mandates does not have
that kind of tight nexus with recognizable protections or
social gains.
Professor Haidt's testimony, reading it, you know, the
effects of social media on teen girls, that natively powerful
testimony, and that's important stuff. He is especially
credible to me because of his work through various
organizations to strengthen academic inquiry norms and honest
debate. I would like it if transparency could directly fix the
problems he cites, or make our democracy evidentially more
functional, but the benefits of transparency, unfortunately,
are contingent and remote, where the privacy losses are
immediate and, I think, real.
The solutions to all of these problems will be years in
coming, and they will come from a wide variety of adaptations
and sources, university research being just one contributor.
This hearing has helped emphasize the importance of
transparency. I don't think, unfortunately, that it lays much
groundwork for a mandatory disclosure regime like the one found
in PATA. It's excellent to have had a discussion draft, though,
to work from.
Thanks very much.
[The prepared statement of Mr. Harper appears as a
submission for the record.]
Chair Coons. Thank you. I'd like to thank all of our
witnesses today for your thoughtful testimony. This is exactly
the sort of broad-reaching hearing I was hoping we would have,
and I am excited that the vote has not yet been called, and so
we may have a little more time to explore it. I particularly
appreciate the perspectives offered on transparency, the
potential ways to achieve it, questions about its cost and the
cost benefit.
I'm going to start my questioning by setting the scene a
little bit and exploring various pathways toward greater
transparency, which, I believe, would be beneficial,
bipartisan, and the role that Congress should play. We may well
take a second round, depending on what happens with votes.
If I might, first, Professor Persily, I appreciate the
point you made about how transparency can transcend political
ideology and might be, I think as you said, a bipartisan first
step on the road to sounder policy. Could you say more about
the kinds--the kinds of questions independent researchers might
answer with greater access, and how answers to those questions
might, in a more immediate way, deliver on the promise of
better policymaking?
Professor Persily. Thank you for that. There is a
fundamental disagreement between conventional wisdom and what
the platforms is happening on their services. For example, we
have no sense of the prevalence and size of the problem of hate
speech, disinformation, incitement, child endangerment, and the
like. If you ask the folks at the platforms, they will say, for
example, that a lot of the interpretations that were made off
of CrowdTangle data, which looks at different engagement
metrics and the like, that that was misleading. That actually,
the average person's experience is not one which is replete
with hate speech, disinformation, and the like. That's an
empirical question. If we had access to the data, we would be
able to figure out how big a deal some of these problems are in
the average person's newsfeed.
And not just the average person, because I think Senator
Sasse was exactly right when he was talking about weirdos in
his opening comment. Part of the question is not just what the
average user sees, but how concentrated minorities of users,
what rabbit holes they may be going down.
That also is a second area of inquiry which is open for
debate. What role are the algorithms playing in sending users
down rabbit holes of polarization, conspiracy theories, and the
like? On the outside, it's conventional wisdom right now that
the algorithms are having a huge impact in doing so. If you ask
the folks of the platforms, they say no. That's not actually
what's happening. If anything, the algorithms are putting them
toward more moderate content. We can answer that question,
right, if you just give us access to some of the same data, we
can figure it out.
Finally, particularly on the issue of censorship and
whether the content moderation practices of the firms have an
ideological bias to them, which a lot of people believe they
do, that's the kind of thing that if we could look at the
content moderation enforcement in a detailed way, we would be
able to figure out whether this is sort of neutral principles
that are having a disparate impact, or whether there is
something more nefarious going on. That again, as I said in my
opening statement, this is what will change the behavior of the
platforms. Right? If they know that their content moderation
decisions are going to be transparent and viewed by others,
then they will make decisions in a different way immediately.
Chair Coons. Thank you. Mr. Silverman, if I might. I think
it's important for people to understand tangibly how
transparency of the kind that CrowdTangle as a tool created,
and how that could possibly benefit society.
Could you just give us a few concrete examples of how
greater transparency might benefit society? I think Mr. Harper
appropriately raised the question about, you know, harm to
privacy being immediate, and the benefits from greater
transparency being remote or contingent, I think was the term.
Mr. Silverman. Yes. Absolutely. I'll go through a few
examples, and I'll try to focus on the ones that are immediate,
and real-time, and you could get fairly quickly if this sort of
legislation was passed.
First there are--is one of the challenges with social media
is its scale. We've seen through a lot of research as well as
some of the leaked documents over the last year, that it is
very hard for these platforms to manage the sheer number of
communities, dialects, languages, individual nuances of places
where they exist and where they're incredibly important. If you
give the outside world the ability to also see what is
happening with public content, they can play a role in helping
with that effort.
One specific example, in the Philippines, this year's Nobel
Peace Prize Winner, Maria Ressa, and her news organization,
Rappler, for years have used CrowdTangle to help identify
coordinated inauthentic networks that were violating the
standards--the community standards of Facebook, helping
identify them, flag them for Facebook to get removed before the
platform would have themselves.
One, is you have the ability to engage not just the
platforms, but a much broader swath of society, in helping
monitor, debate, and engage with what's happening on those
platforms.
A second one is there is a lot of academic research that
can be done on public data. There are very privacy-sensitive
datasets that you have to build a lot of controls around to
deal with in a safe and responsible way. But even--there's also
a lot of publicly available data that can empower a lot of
research. We have over 1,000 academics and researchers who have
used CrowdTangle, and published hundreds of research papers,
including in nature, in science, and others looking at this
data to inform lawmakers and policymakers.
I'll give you this one last one is, I think if you talk
about simply, like, at a really high level, the idea of a
marketplace of ideas, if in any way we want these platforms,
which already house so much of our civic and political
discourse, to actually be functional marketplace of ideas,
where people can engage and debate with them, you can't do it
if people can't see what's in the marketplace. And so simply as
creating healthy, dynamic places of liberal debate and
engagement, one of the just first foundational things you need
is to make it easy to see what's happening. We saw over and
over, local news outlets, independent journalists,
investigative news, using the data that was coming through our
system to both cover the platforms themselves, but also simply
talk about issues of the day, and report on them to the public.
Chair Coons. Last question to you, if I might, Mr.
Silverman. Some would argue that to your point about
marketplace of ideas, that the market should just sort it out,
and that any regulatory effort by Congress is likely to
overreach or pose as profound risks to privacy. You saw in your
experience at Facebook, I don't want to put words in your
mouth, but that it was in the market interest of a lot of
platforms, other platforms, to simply avoid any transparency
whatsoever.
Help me understand why you think it is that large social
media platforms want to avoid scrutiny through transparency.
And why, in your view, if this is the case, we can't just rely
on voluntary disclosures, or on the market--commercial market,
to solve this problem?
Mr. Silverman. Yes. I think there are a number of things
happening. I'm not an economist, so I don't want to out too far
above my expertise, but I think there is a reality in which
there are not fully diverse, you know, free markets happening
in some of these industries. That there are dominant players
that have enormous percentage of the market share and are less
subject to the whims of, you know, a free, you know, market of
users that can pick among a wide variety of choices. One is, I
just think there is a reality of the actual nature of the
markets in which these companies operate.
Two is, at this point, we have 10-plus years of evidence
that some of these companies can do very little, and it doesn't
matter. The reality is there are a lot of challenges to doing
it inside Facebook, but one of them was certainly the question
of, why are we putting ourselves out on a limb when others
aren't? I don't see that dynamic changing at the moment, and
maybe that's a judgment call, and I could be wrong. But I think
there is just too much evidence in the industry right now of
too little efforts around transparency relative to the scale
and the need out there.
Just to add one last note on this, while also at the same
time the executives all talk about how important transparency
is, so I think they both acknowledge its importance, but I
think the level of efforts across the industry just haven't met
the need we have as a country. And so, now, it's time for
legislation.
Chair Coons. Thank you, Mr. Silverman. Senator Sasse for 8
minutes.
Senator Sasse. Eight? Wow. That's generous.
[Laughter]
Senator Sasse. That's good. Let me just first underscore
Chris' point. Five for five really useful testimony, the stuff
you submitted, even if 300 pages, but also all of your verbal
comments here have been useful.
Professor Haidt, I want to start with you because there
were a few side comments about how unrepresentative our
stereotypical sense of the messed-up conversation, public
square, digital public square, is compared to maybe the median
experience. You didn't, in your opening statement, refer to
your Atlantic piece 2 weeks ago. But given how viral that has
gone, in this case, in a good way, would you be willing to
unpack your four-part typology on who the overrepresented, very
online, very angry, loud people are?
Professor Haidt. Yes. I've spoken with Mark Zuckerberg a
couple of times. I've heard the arguments, and I've heard him
say, ``How could it be wrong to give more people more voice?''
That sounds great. If everybody, especially those who have less
voice, were all lifted up, that would be great. But in fact,
what I argue has has been happening, is one social media
developed a tool to make it really, really easy to attack
people, criticize them, complain. Once it became not about,
look at my nice photos of my kids, but can you believe that
this person said that? It was as though everybody was given a
dart gun, and everybody could shoot whoever they wanted. But
most of us don't want to shoot anyone.
The four groups who are doing most of the darting are the
extremists on the far-right, the extremists on the far-left,
trolls, who are mostly men with personality disorders. They
enjoy harassing people and showing off, and Russian
intelligence agents.
Social media, especially Twitter, but also Facebook and
others, have been an incredible gift to those four groups. The
other 80 percent of us lost voice. We're afraid to speak up. We
see this among our students in class. It's horrifying. Our
students are literally afraid to challenge something because
they're afraid someone will record it, someone will talk about
it, someone will shame them.
Social media, there was a period when we were incredibly
optimistic. A period of techno-democratic optimism when it
seemed, in the early 2000s, like this was going to be the best
thing for democracy ever. But I believe what we're seeing is
James Madison's nightmare where it just promotes factionalism,
and fear of speaking up, and it gives us a distorted
marketplace. I would love it if Twitter, if Elon Musk could
clean it up so that we weren't so afraid to speak, but right
now, people are.
Senator Sasse. Really helpful. For folks who haven't seen
it or read it, I highly recommend his Atlantic piece 2 weeks
ago. It wasn't exactly called the dumbest decade, but it's
something like that.
Professor Haidt, can we stay with you a minute on the
teenaged girl harm effects that we see? You worked hard in your
written testimony to distinguish between social medias--studies
that look at social media in particular from those who look at
screen time in general. Can you unpack why that distinction
matters?
Professor Haidt. Yes. Because, you know, look. When I was a
kid, we all watched too much television. It turns out that was
a moral panic. Televisions screens, they don't rot your brain.
It was hard to link that to bad health outcomes. Many have said
the same thing is happening now, and it turns out, when you
look at the research on, say, videogame playing, in general,
it's not particularly harmful. Now you know, if you play huge
amounts, it's different.
The point is, digital technologies, screens are not bad
intrinsically. It depends on what you do with them. If a screen
promotes social learning and engagement, in a healthy way,
between kids who can connect and play together, that's great.
That's not harmful. Watching Netflix is not harmful. Watching
videos are not harmful.
And so, much of the research has looked at all digital
activities, and they say, ``Hey, all digital activities for all
kids, the correlation is so tiny, it's not even worth worrying
about.'' But what I found over and over again is when someone
sends me a meta-analysis this happened on Twitter just this
morning. Someone said, ``Here's a meta-analysis from 2020, that
this proves your thesis.'' If you look at the meta-analysis,
yes. They find nothing when you look at everything. But
whenever you zoom in on social media for girls, you almost
always find a much bigger correlation up in the range of r =
0.1 to 0.2, which is, clinically, very serious. This is what
public health effects are.
We really have to distinguish. Don't get caught up in
digital media and screen time. Focus on social media for girls,
especially in middle school. For God's sakes, let kids go
through puberty first, before we encourage them to live their
lives, putting out pictures and asking people to tell them how
pretty they are.
Senator Sasse. Really helpful. I would like to ask some
more questions about the solution set relative to that problem
because, to me, that is one of the biggest problems we face.
Our republic is not going to survive and thrive unless we, the
people, have a set of habits that recognize our digital
consumption habits are things we're going to have to be
responsible for. That doesn't mean there are no collective
solutions to any of these problems, but fundamentally, most of
what we're talking about with digital addiction is going to
have to be responded to at the level of individual, and family,
and communication, localist, and healthy institutions that
figure out ways to put constraints on our own addictions and
consumption.
It seems to me that kids are a completely different
category. And so, if I weren't out of time, I would ask a
number of you who are advocating for this particular piece of
legislation to help me understand why we wouldn't really want
to be starting by targeting solutions that deal with the teen
problem, or with the minors problem. Given that we're going to
have vote soon, and we've got other Members waiting, I'll defer
it for later. Thanks.
Chair Coons. Thank you, Senator Sasse. Senator Ossoff.
Senator Ossoff. Thank you, Mr. Chairman. Mr. Silverman,
what are the implications of Government access to some of these
research tools? Are there Fourth Amendment implications?
Privacy implications? Surveillance implications?
Mr. Silverman. I might also let some of the other experts
on this panel answer some of that.
Senator Ossoff. Sure.
Mr. Silverman. But what I will say is any answer to
transparency shouldn't be a one-size-fits-all approach. There
needs to be tiered access, or in my opinion, there should be
tiered mechanisms with different access points, with different
audiences, designed to serve different purposes.
Each one of the different set of constraints, and just to
give you an example, one way I think about it sometimes is a
pyramid or a funnel, where at the very top, you can have the
most widely accessible forms of transparency. I think about
that as reports. Right now, a lot of platforms put out reports
on hate speech, or coordinating authentic behavior, etc. That's
available to the public. It's privacy safe. As far as I know,
there haven't been any issues raised about Fourth Amendment or
First Amendment about it.
As you go down the funnel into more sensitive datasets,
with smaller audiences, I think some of those issues get much
more prevalent. The idea of having a tiered system would
ideally give you a way to address each of those in safe and
responsible ways.
But I'll also say there are real issues, and I think ones
that also private companies shouldn't be trying to figure out.
There's an argument for why Government should be looking at
those tradeoffs and figure out a solution.
Senator Ossoff. Anyone else want to weigh in on the Fourth
Amendment or privacy aspects? Concerns that are raised by
Government access to these transparency or research tools?
Mr. Harper. I will try to just briefly. As I detailed
somewhat in my written testimony, certainly the PATA
legislation, broadly written as it is, would have Fourth
Amendment and privacy considerations. The argument I make is
that the written materials, the privacy policies and terms of
service that platforms put forth, are contracts that allocate
that personal information as data. The right to possession is
often with the platform. The right to exclude others, which is
privacy protection, is with the consumer, subject to narrow
exceptions that I think are generally appropriate.
Taking that away, ladling that out for researchers or for
any other purpose, would be taking of property from the
business, taking of property from the individual. It would be
taking the papers and effects, in Fourth Amendment terms, of
those people.
The Supreme Court, had some--there's some whisps of heading
that direction. The Riley case referred to as digital
materialism and the phone as a person's effects. That signifies
that it's something owned by the individual. I think that's
true whether it's on your phone or whether it's housed for you
by a service provider.
Senator Ossoff. If there were to be enacted transparency
requirements for these platforms, would it be sensible that
there be transparency requirements for Government access of
those transparency tools?
Mr. Harper. First, limitations. I think the legislation
should specify certain categories, perhaps, that should not be
available. Certain reasons. Again, the PATA legislation, a
discussion draft is very broad, but it gives to the NSF and the
FTC the ability to decide what data and what for, and that's
far too broad. Congress should actually be the policymaker on
those questions.
First, limitations. Then certainly, transparency if the
Government accesses that data, either through the NSF/FTC
program, or from the researchers themselves. I think, if not
entirely barred from access, there should be transparency, if
given access that way.
Senator Ossoff. Professor Haidt, I want to invite you to
elaborate on some of what you shared with Senator Sasse with
respect to the impact on public discourse and political
discourse of how these platforms are functioning. I'm not sure
if your research has included this specifically. If not, I
invite you to speculate in public about the impact on elite
opinion, the opinions of policymakers, and those who staff
them, and the activists who are most vocal in advocacy. And
then ask you what solutions you're proposing, please.
Professor Haidt. Thank you, Senator Ossoff. I think that
the effect on the elites is extraordinary and is overwhelmingly
bad. Many people will say, ``Twitter's not that important
because, you know, 80 percent of Americans are not even on it,
so who cares?'' Almost all journalists are on it. If Senators
and Congresspeople aren't on it, their staff certainly is, and
they're very, very responsive.
We want our leaders, we want our representatives, to be
responsive to all of these constituents and to the broader
country. But now that they're all on Twitter, they're
responsive to the most loud, vocal, angry people on it, so it
takes them away from their duty. I would say that
representatives who pay a lot of attention to social media are,
in a sense, violating their fiduciary duties to the country. I
understand why they're doing it. We're all doing it.
Of course, journalists, my God, think about how many
stories on the news are about something someone said on
Twitter. You know, right-wing ecosystem is much more dependent
on cable TV, but even that, they get the stories from Twitter.
So they know exactly what the most angering story is going to
be.
So, don't listen to anyone who says, ``Only a small
percentage of people are on it.'' The influence on those
people, then, goes out through many other channels to affect
the entire country. We can't have a deliberative democracy if
we're not able to deliberate.
I would say that social media has given us an environment
which we're sucked into fighting over trivia. We don't
deliberate.
Senator Ossoff. Solutions?
Professor Haidt. So the solutions, there are a number that
are crucial for making social media less toxic. The most
important thing is, how can we make it so, you know, of course
the extremes are always going to have more voice. They care
more. They're more passionate. The middle is always going to
have less. Fine. But what happened about 2012, 2014, was the
extremes got so amplified, along with the trolls, and the
middle went down.
How can we undo that? The most important thing is please
stop talking about content moderation. I am so sick of it. What
we learned from Frances Hangen, it doesn't actually even matter
all that much, whether they get a little bit more or a little
bit less. Look at the architecture. That's what changed. That's
where the solution lies.
The most important thing is verifying identities. Not that
you have to post with your real name, but just as you can't go
to a bank and give them a bag of money and say, ``Open an
account,'' banks have to know your customer laws. Systemically
significant platforms that effect the health of our country
should be like banks. Not that they're going to tell you what
you can say and can't say, but they're going to say, ``In order
to speak on this platform that has Section 230 protection. We
get Section 230 protection, we have a duty to at least verify
that you're a human being, that you're old enough to be using
the platform, we have to be getting to do some age gating and I
think also verifying that you're in a particular country,
because we'll have different rules in different countries. If
we do those three things, that would wipe out most of the bots.
It would make us less afraid to speak. It would elevate the
center and quiet the extremes.
That's one, but there are many others. Architectural
changes. Don't focus on content moderation.
Senator Ossoff. Professor Haidt, I'm way over time. I just
want to ask you if you got a 15-second response here. How do
you respond to folks who point to, for example, the role that
anonymous pamphleteering played at various points in our
history? Or the capacity of whistleblowers, or those who
possess sensitive information to disclose such information and
not putting themselves at risk. How does that conflict with
what you just suggested, if at all? And that--that'll be my
last question. Thank you.
Professor Haidt. Sure. The spread of information has always
been an issue in democracies. All the things that are going on
now, most of them have a precedent. All of these things that
used to happen, they didn't make us afraid of each other. They
didn't make us afraid to speak up in class. This is something
new. This is changing the social dynamics.
Don't just focus on information and false information.
Focus on the fear of speaking up. That is what is making our
institutions structurally stupid.
Senator Padilla [presiding]. Thank you, Senator Ossoff. In
Senator Coons' absence, I'm presiding for the time being and
will proceed with my questions by recognizing the next Member.
Let me just begin by thanking Senator Coons for holding this
hearing on platform transparency, and to all the witnesses for
exceptionally detailed testimony.
You know, transparency has been a key piece of the debate
over the responsibilities of technology companies and the
impact their services have on our democracy, and the welfare of
our loved ones and our neighbors.
As one of the few Senators with an engineering background,
I believe that policymaking is at its best when we're also
armed with research and data. To that end, I support the calls
to enhance platform transparency to better serve the public,
users of these services, the company themselves, and lawmakers.
We have an important opportunity for real productive and
effective policymaking.
Before I jump into my first question, I also want to note
that I have the honor of representing the State of California,
home to more innovators, and executives, and investors in
technology than any other State in the Nation. Also, home to
more consumers of technology and users of platforms than any
other State in the Nation. And of course, more employees in
this sector, than any State in the Nation.
And so, my first three questions are a little rapid-fire.
Mr. Silverman, what State you from?
Mr. Silverman. California.
Senator Padilla. Professor Persily, what State are you
from?
Professor Persily. California.
Senator Padilla. Professor Keller, what State are you from?
Professor Keller. California.
Senator Padilla. No disrespect to the other witnesses.
[Laughter.]
Senator Padilla. I just wanted to drive home the point
California plays a huge role in this discussion and in the
problem solving as well, from all angles.
Advocates are routinely highlighting the failure of
technology companies to moderate content in non-English
languages. As Frances Hangen testified at a previous hearing,
87 percent of all spending combatting misinformation on
Facebook is spent on English language content, but only 9
percent of Facebook users are English speakers. That's
important data. Other platforms likely have similar
disparities, but they haven't publicly disclosed similar
information, and I certainly would love to see that data.
Professor Keller, in your testimony, you shared that the
EU's Digital Service Act requires public transparency reporting
around the number of content moderators platforms employ and
what their linguistic expertise is. Do you think similar
transparency reporting should be considered in the United
States?
Professor Keller. I agree with you that the language issue
is incredibly important. I think one of the most important
disclosures from my former client, Frances Hangen, you know,
was that not only in languages that don't have as many speakers
is the content moderation not as good, but also they don't
build machine learning systems that are as good as a result. So
the problem is just compounded.
I do think that attention to additional languages is
incredibly important. I mean, unfortunately, it's part of what
makes of measures of things like prevalence so expensive
because it means that you're deploying people in many
languages, who need many cultural contexts to try to
extrapolate how much content exists across the platform. But
yes, I am in reporting that you described.
Senator Padilla. Just briefly, is there any other types of
information or data that would be helpful in addressing this or
other disparities that concern you?
Professor Keller. One of the most important things is to
see the actual content that platforms took down, or left up, or
demoted. You know, if researchers can't see that, then the
platforms are grading their own homework. The researchers can't
see if there's a pattern of bias, or if they're making
mistakes, or whatever.
This collides with the privacy issues that I identified in
my testimony for platforms like Facebook. But certainly, for
publicly-shared content, you know, where the person who posted,
it hasn't tried to take it down. Having an opportunity for
third parties to look at what's actually going on, in multiple
languages, you know, from multiple cultural contexts, is really
important.
Senator Padilla. Let alone liability. It further speaks to
the complexity of the policy here. On a separate note, when
legislating on technology-related issues, I think it's
important, and I raised this in the Full Committee previously,
important to not enable or incent politically ambitious
Government officials with legal authorities who may use our
efforts to undermine the ability of platforms to limit the
spread of hate speech and election disinformation, just as two
examples.
Question for Professor Keller again. In your testimony, you
cited the Texas attorney general's ongoing dispute with Twitter
as an example of States trying to influence online speech. As
Congress considers regulating the conduct of tech companies
with respect to transparency and competition, what should we
know about the battle over online speech regulation? And what's
taking place in the States?
Professor Keller. Yes. I think it's really important to
appreciate the rash of cases out there, litigation, and State
laws, some of which have passed, that have the goal of
compelling platforms to carry content that the platforms don't
want to, that violates the platforms' rules. This includes, you
know, white nationalists saying that they have a right to post
white nationalist content on Twitter. You know, and there have
been over 70 of these cases. The laws that were passed in Texas
and Florida both create these obligations to carry content that
violates the platforms' rules.
I'm sympathetic in a way. Like, these are the public forums
of today. I understand why people are very concerned about
being excluded if they are being excluded. But creating a
mechanism for government actors to effectively strong-arm
platforms about their policies, which I think is what is going
on with the Texas AG investigation, is quite dangerous.
I think this is something to worry about with some
competition bills pending that are otherwise, I think, really
good ideas. I'm sorry that Senator Klobuchar isn't here to hear
me talk about her bill, but you know, the 95 percent are about
saying platforms shouldn't self-preference and promote their
own properties over their competitors. And then, both her bill
and Senator Blumenthal's bill, and he's not here either, both
have just a couple of sentences that permit a different
category of lawsuit. A lawsuit that's Breitbart saying they
should be treated like the Wall Street Journal, for example. I
think it's a very serious change in the law, if that comes into
effect.
Senator Padilla. Thank you. Speaking of Texas, Senator
Cruz.
Senator Cruz. Thank you, Mr. Chairman. The single biggest
threat to free speech in this country, in my judgment, is the
power of Big Tech. A handful of Silicon Valley billionaires who
have arrogated to themselves complete monopoly power over the
public discourse. Professor Keller just referred to these
social media sites as the public square, and that is very
accurate. It is how we speak with each other. Big Tech has
gotten more and more brazen in its abuse of that power.
This is a hearing on transparency for Big Tech. I would be
in support of almost anything imaginable to increase
transparency for Big Tech. The bill being discussed here is a
fairly modest step that gives access to some academic
researchers. I suppose that would be fine. It's not clear to me
why a professor at Harvard or Stanford should have some special
access that Joe Q Citizen should not. But when it comes to
transparency, the people have a right to know. But to the
extent academic research marginally increases the ability of
the people to know what's going on, I imagine that's a positive
step.
The lack of transparency is not an accident. It is a
deliberate feature of how Big Tech has set up its systems. To
all of the witnesses here today, the 2016 elections. Does
anyone know how many posts from Republican candidates for
office were blocked? Does anyone know how many posts from
Democrat candidates for office were blocked? How about the 2018
election? 2020 election? Does anyone know the average--average
ad rate charged by Google or by Facebook to Democratic
candidates for office? Does anyone know the average ad rate
charged by Google or Facebook for Republican candidates for
office?
Nobody knows. I don't know. The Chairman doesn't know. I'll
tell you, Mark Zuckerberg has sat at that table. I have asked
him that question. I've asked the CEO of Google those
questions. I've asked them those questions in writing, and they
hire teams of lawyers to write letters back that say, in every
way possible, pound sand. We refuse to tell you. By the way,
trust us. We're not censoring. We're just not going to tell
you.
During the Trump administration, I begged the Department of
Justice, if they did nothing else on Big Tech censorship, to
use the subpoena authority of DOJ to get answers to basic
questions on transparency. I think there were multiple people
in the administration who wanted to do that, but they did not
get that accomplished.
Few things illustrate the abuse of power of Big Tech over
free speech better than the reaction, in the last 2 weeks, to
Elon Musk announcing that he's buying Twitter. I find it quite
remarkable. I think Elon Musk's buying Twitter is, without
exaggeration, the most significant development in favor of free
speech in decades.
I also find it astonishing the reaction of much of the
corporate media in the left to Elon Musk buying Twitter and, oh
my God, suddenly conservatives being allowed to speak, and it
is truly Armageddon. It's cats and dogs living together. It is
the worst imaginable. Watching the public histrionics of the
left, if their opponents are not silenced is amazing.
By the way, Elon Musk, the last I've checked, is not some
right-wing character. He's a lifelong Democrat who voted for
Barack Obama twice. This is the scary specter because he's
dared stand up and say, ``You know, we'll allow free speech.
We'll allow speech I disagree with.''
Look. I'll give you one data point. I asked how many were
blocked. Nobody knew. I could ask how many were shadow banned.
No one would know because they don't tell you. But I'll give
you one data point from my own Twitter page. I am active on
Twitter. I spend a lot of time on social media. Twitter
accepted Elon Musk's offer to purchase on April 25th. All
right. On April 22, my Twitter account gained 1,488 new
followers. On April 23, it gained 1,526 new followers. On April
24th, it gained on 1,486 new followers. On April 25th, it
gained 1,214 followers. Going back, I was pretty consistent
gaining 1,000 to 2,000 a day.
Twitter accepts Elon Musk's offer to purchase them. The
next day, April 26th. Mr. Chairman, I would ask you how many
new followers do you think I gained the next day?
Chair Coons [presiding]. More than 1,000.
[Laughter.]
Senator Cruz. You would be correct. The next day, I gained
51,405. The next day, April 27th, I gained 61,261. The next
day, April 28th, I gained 70,584. In the week and a half since
Elon Musk purchased Twitter, my Twitter followers went from 4.8
million to 5.1 million. Conservatives all across the country
have reported numbers like that, have put up numbers like that,
and it is obvious. Someone flipped a switch. The Governors they
had on that said ``silence conservatives'' were flipped off.
That is the only rational explanation from going from 1,000 to
70,000 the day after he bought it. I'll just point out, he
hasn't even taken it over yet. This is just the in terrorem
effect of some engineers who, I imagine, are running the
document shredders like crazy going, ``Crap. They're going to
find out what we're doing. Turn the stuff off.''
That activity illustrates the need for transparency
profoundly, and I hope Congress does something about it.
Chair Coons. Thank you to my colleague from Texas. You, I
think, highlight the many ways in which, whether it's from the
right or from the left, there are lots of questions about how
these platforms operate, how they censor, how they reinforce,
how they amplify. I am tempted to ask whether any of you have
an alternative suggestion for exactly why the Senator's Twitter
followers may dwarf mine in number and scope. A question,
perhaps, of interest really only to me, or maybe my colleague
from Texas. Mr. Harper, if you'd like to offer----
Senator Cruz. I follow you.
[Laughter.]
Chair Coons. He does follow me, which I appreciate. One of
the very few. My Twitter followers are infinitesimal.
Mr. Harper. If it's not too late, I wanted to ask Senator
Cruz to retweet me. I'm @ Jim_Harper.
[Laughter.]
Chair Coons. Mr. Harper, if you'd like to speak to that,
then I have a second round of questions. My understanding is
Senator Blackburn is on her way. Are there other Members on
their way? Because relatively soon, I am going to have to close
the hearing out and go vote the second time. Mr. Harper, if you
would, to the intriguing characterization the Senator from
Texas made about the unavoidable conclusion one must reach
about what is happening at Twitter.
Mr. Harper. I think we know less than that about how things
work in those companies and what occurred in the days that the
Musk takeover was reported. Let me say, though, thematically,
both his question/testimony and a lot of what we've discussed
here, for me, are speech topics. We're talking about what we
want our speech in our country to go like, how we want it to
go.
This, by the brusque language of the First Amendment, is
not the place where that stuff is supposed to be decided.
Again, I don't think my written testimony was the strongest on
First Amendment issues, and there's a lot there to consider.
But I'm sorry to say it, this body is largely disqualified from
regulating speech in the country, and that's for the good for
different reasons.
Chair Coons. No. I don't think you should hesitate to say
that. I think that is something we can all agree on. This was
simply a follow-on to Senator Cruz's questioning.
I'll now yield to Senator Blackburn. She, I think, will be
the last of our first round of questioners. Then Senator Sasse
and I may have a second round. Senator Blackburn.
Senator Blackburn. Thank you, Mr. Chairman. I want to thank
each of you for being here. Over at the Commerce Committee,
Senator Blumenthal and I have held five hearings this year on
Big Tech and the effects that it has on the public writ large,
and also on children.
Professor Haidt, I want to come to you with a question, if
I may. Let's see, I'm hoping that he is still on----
Professor Haidt. Yes. I'm here.
Senator Blackburn. Okay. Great. As long as you're still
there online. Thank you for your opening statement. I have a
question for you pertaining to the Kids Online Safety Act that
Senator Blumenthal and I have filed. We have a provision in
there that would allow independent researchers access to the
data that--to these two Big Tech platforms, into the data, how
they are holding that data, how they're utilizing, crunching
that data, when it comes to harms to minors. What I'd like for
you to do, for just a moment, is talk about in your research
and the work that you are doing with research, and parents, and
teachers, physicians, policymakers, what are you finding, when
it comes to kids 18 years of age and under as to the impacts
and the experience that they have on these social media
platforms?
Professor Haidt. Thank you for that question. The
fundamental reality, the thing we have to keep our eye on, is
that using social media is not like consuming sugar. A lot of
the research has looked at, you know, if you consume a lot, do
you get more sick? If you consume a little, do you get a little
sick? But if we look at it from the kids' point of view, what
we have to see is that children have always wanted to play with
each other, and talk to each other, and do things with each
other. It's when, around 2011, plus or minus, when they all got
online, they spent so many hours online performing for each
other, they don't have much time to actually connect and play.
Now when they're actually online, sometimes they say it's fun.
Sometimes they say it's not. That data is mixed. The overall
impact on them is clearly negative, and this has been
overwhelmingly consistent.
I gave a talk at my old high school, Scarsdale High School,
just before the pandemic, and all of the teachers are like,
``We can't get through to the kids. They're not paying
attention. This is a disaster.'' By the time they come in
already, a lot of them are depressed and fragile.
I gave a talk at my middle school in Scarsdale, and I spoke
to the principal there, and the teachers there. They said the
same thing. In fact, they even said by the time they come to us
in sixth grade, they're already addicted to their devices, many
are fragile, many of them are depressed and anxious.
What we're hearing consistently from all who work with kids
is this is messing them up, and we didn't give----
Senator Blackburn. Okay.
Professor Haidt. We don't want this to be happening. I
think it's been terrible from all reports. I don't know anybody
who's happy that our kids are now spending, you know, 4 to 10
hours online on social media every day.
Senator Blackburn. Yes. We are hearing some of the same
things, and it's why we have put the provision in there, in the
bill, that would allow this research.
Ms. Keller, it's good to see you again. I want to talk with
you about privacy because, as you know, the EU has moved
forward with a Digital Services Act and a Digital Marketing
Act. One is more closely akin to our 230, the services. But
their Marketing Act does deal with privacy.
We have yet to address this issue in the U.S., even though
some of us are hard at work on it. I would like for you to
touch on why we need to make certain that we have a consumer
privacy bill that exercises Federal preemption.
Professor Keller. I'm an ex-privacy lawyer. It's good to
see you again, too.
Senator Blackburn. Yes.
Professor Keller. I'm an ex-privacy lawyer, and most of
what I've worked on is the GDPR. The General Data Protection
Regulation in the EU, and it gives them a baseline, to start
with, of how privacy is supposed to work. Then you can build
transparency on top of it and have some starting point rules,
and some mechanics for resolving the new questions that arise,
you know, when it comes to platform transparency. The Digital
Services Act, as I mentioned in my testimony, it has multiple
transparency provisions. It has almost all of the provisions
that are in PATA, although it does not have the scraping
provision, which is the sort of like democratic scrappy kind of
research. That one's not in the DSA.
In particular, the researcher access, so that analogue of
the first four or sections of PATA, which is in Article 31 of
the DSA. Because the GDPR is there, and because they're able to
convene experts, and this is happening right now. There should
be a code out by the end of the month, trying to say how this
is going to work with privacy.
You guys are in a much more difficult position because you
don't have that baseline. And so, the lack of Federal privacy
legislation to start from leaves you having to answer questions
that shouldn't be your job to answer it.
Senator Blackburn. Thank you. I appreciate that. We need to
get to that starting point. Thank you, Mr. Chairman.
Chair Coons. Thank you, Senator Blackburn. Senator Sasse,
any closing question you'd like to ask before you have to go?
Senator Sasse. No. Thank you, again, to all of the
witnesses, and to not just you, but your team for organizing
this. It's been instructive, and I'm sure we'll have follow-
ups. Thanks.
Chair Coons. If I might, I'm going to ask a couple of more
questions until I get told I have to go back and vote again.
Professor Keller, to the point that Senator Blackburn was
raising, help me better pull apart the tensions, competitive
impacts, privacy concerns, and how you think progress, that's
how I view it, in your--the foundation of GDPR and the
potential. We don't yet have the full text for Digital Services
Act, but where do you think Europe is going? What should we
learn from European regulatory legislative efforts in terms of
the balance we might strike, both around individual privacy,
and privacy rights in the digital domain, but also competitive
consequents, competitive concerns? Do you think PATA strikes
the right balance? Do you think there are other ways in which
we might make sure that we are protecting privacy and
companies' interests in competitive concerns?
Professor Keller. Yes. There was a lot in there. Please.
Chair Coons. Yes. Take your time, unpack, and then I'm
going to invite anybody else who wants to respond to that, as
I'm sure Mr. Harper will.
Professor Keller. Yes. There are things I really like about
the DSA as a regulatory approach. I actually had an op-ed in
The Hill a couple of years ago saying, ``Congress can learn
things from how Europe approached the equivalent of 230
issues.''
Chair Coons. Your editorial saying, ``We hope that Congress
can learn things.''
Professor Keller. I was more polite than that. You know,
there are great things about it. I do think that it gets the
balance of competition and other values a little wrong. If you
want to get really in the weeds, I have a blog post on that on
something The Verfassungsblog, which is a German constitutional
law blog. It's awesome. You know, it goes into how effectively
I think the DSA is sacrificing competition goals in the name of
content regulation goals, or you know, speech control goals by
putting lots and lots of mechanical burdens even on very small
platforms, you know, that I think are just, you know,
disproportionate. It'll make it harder for them to grow up and
compete with the bigger platforms.
Chair Coons. One of the concerns you've raised in our
conversations about PATA is making sure that we're not writing
rules that are unduly burdensome for smaller platforms. If
you'd make a brief comment on that, I'd appreciate it.
Professor Keller. Yes. I think that's right. I mean, so
PATA has 25 million monthly active users and up as its size
range. I put in an exhibit that I can't swear it's right
because nobody really knows these numbers, but it at least
tries to get at a question of which platforms is this talking
about? You know, and a lot of them, I think, are not the kinds
of platforms that people have in mind when they're asking for
this transparency.
I should say, also in transparency, I do some consulting
for Pinterest, and Pinterest is somewhere on that list. But you
know, think of a platform like GlassDoor, which is people
talking about their employers. You know, that doesn't have the
democratic discussion consequences of a Twitter or a Facebook.
And so the justification for wanting transparency is lower, and
their capability of carrying that burden is also lower.
Chair Coons. Small-D democratic.
Professor Keller. Small-D democratic.
Chair Coons. If I may, I'm going to jump to Mr. Silverman.
I'm going to keep going until I get a text saying, ``You got to
leave now, Senator.'' PATA, the Platform Accountability and
Transparency Act, creates a requirement platforms provide
ongoing disclosure of information of the type that has fewer
privacy implications, public-facing content, typically viral
public-facing content, platform advertising practices. How
would expanding disclosure of that kind of information, that I
would argue has a lower privacy risk, actually help the public
better understand how platforms are impacting online discourse?
Mr. Silverman. Yes. Thank you for that question. I'll start
off by saying that I agree with your assessment, that there are
types of content on these platforms that have greater privacy
risks, and there are also types of content that have less. We
had a very stringent set of lawyers and policy folks at
Facebook, but the work we were able to do was able to be
blessed from their perspective, in terms of whether it was
privacy-safe or not.
There is a lot of content on these platforms that has very
minimal privacy risk. One version of that that I think PATA
tries to capture, and I think is a really promising place to
move, is they give out content that comes from particularly
public, or in some cases what some of us may call reasonably
public accounts. If you think about the President of the United
States or, you know, a major media outlet that uses an account
in a very public way, has millions of followers, and uses it
for their official business, can you, in some ways, treat that
as a type of content where there are less privacy implications,
and you could make publicly available to a wide swath of the
public.
When you do that, there is a lot of discourse that is
shaped by influential large accounts on these platforms. Even
just providing a real-time window into those accounts and that
content is enormously valuable to being able to see an
important part of what's happening on these platforms.
Chair Coons. Mr. Harper, you've raised some concerns about
privacy in particular. If I understood your opening testimony,
it was essentially that the balance of interests, there's not
enough public good to be accomplished here to outweigh the
risks to privacy. Could you see some reasonableness to Mr.
Silverman's point, that there may be some accounts that are so
clearly geared toward large-scale public communication where
the privacy risks of offering, sort of, an under-the-hood look
at those accounts are less, and arguably worth, that tradeoff?
Mr. Harper. Yes. I think there are a lots of ways probably
to slice and dice what information is made available to
researchers or made available publicly. Obviously, information
that is not personally identifiable doesn't have privacy risks.
It has risks that will be reidentified, and that's important to
consider.
I think that's an interesting idea, the accounts that are
so public that we don't treat their behavior as private. It
might be appropriate to modify the terms of service so that if
you are a public figure, for example, you don't get the same
privacy that a private citizen does. I, you know, can't think
through here on my feet exactly all of the manifest of issues.
Chair Coons. I understand.
Mr. Harper. But certainly, there are ways to limit the
privacy consequences of this hearing. My point, which I think
again, is that the wholesale form that PATA allows for
threatens privacy because it could be lots of communications. I
appreciate Professor Keller joining me over here on the skunk
side of the garden party to highlight those kinds of concerns.
If I may just briefly, in the negative 2 minutes you have
left, I see things--so we line up in interesting ways, of
course. But I sort of see things, a phase shift, differently on
this question of GDPR, for example. It's an outgrowth of the
Fair Information Practices, which arguably arise from the
Privacy Study Committee that the Health, Education, and Welfare
Department put together in 1971. I'm saying 1971, as I am
unsure of what year it was. It was back around then.
Ever since the FIPs came into existence, they have been
sort of the intellectual way we figure out what privacy
protections should be. They were obviously more firmly adopted
in Europe. They're here in the Privacy Act, for example, of
1974. But they were in adopted in Europe because of the civil
law tradition in Europe, which is an intellectual tradition.
The smart, thoughtful people get together and figure out what
the rules should be.
Common law and civil law aren't entirely separate from one
another. The categories aren't clean. But the common law
tradition of England and the United States is a little more
loosey-goosey.
Really, in the privacy area, you see distinctions
culturally between Europe and the United States, where in
Europe, privacy is really more of a dignity value so that
everybody gets the same treatment as the princes, and kings,
and queens have gotten. In the U.S., it's more of a liberty
value, that is keeping King George out of my house. We have
common law privacy protections. A contract is common law.
Property is common law, but also the privacy torts here in the
United States.
I think when it comes to getting data out of these
companies, GDPR is a significant disincentive. Mr. Silverman
talked about having to go through the lawyers because of the
extreme pressure they're under on the privacy side. If your
rule is do no harm, you've got a little more leeway to act than
if your rule is we've got this legislation, and we've got this
regulator in the FTC who's over our shoulder.
In my written testimony, I talk more about the common law,
and I think it's a better way to get more innovation and to
release a little bit of pressure so that there can be a more
give and take that allows researchers access to data.
Chair Coons. Thank you, Mr. Harper. Professor Persily, if I
might, because I'm going to ask you one last question,
Professor Haidt one last question, and then go off to the joy
that is legislating with my colleagues.
Platform Accountability and Transparency Act, sorry to keep
coming back to it, but you know, part of my goal here today was
to take both critical and complimentary input, and then further
refine it. It requires that platforms turn over datasets to
researchers upon review and approval by the NSF. Some of this
data really could be sensitive in ways that I'm concerned
about, and I'd be interested in your explanation, your
insights, into how the FTC could best work to ensure that user
privacy isn't compromised in datasets disclosed to researchers,
and best could be effective as an overseer or policing function
of how that data is used going forward, in ways to be most
protective of individual privacy.
Professor Persily. Thank you for that. I think PATA does
the kind of tiered disclosure that Brandon was talking about,
which is that for the sections that look at algorithms, the
content, and advertising, right, that is going to be more
publicly available. We'll have aggregated reports.
The same may be true for some of this, what we're going to
call, sensitive data here that sometimes aggregated information
is going to be the way that it would be given to researchers.
No personally identifiable information should be given to
researchers. I think the FTC would make that clear.
Researchers are not interested in going and looking at
individual accounts. Right? We want to know sort of groups of
people and what they are going to be sharing and engaging with.
And so, I think whether it's through technology as a
differential privacy, other kinds of anonymization techniques,
this is something that we do all the time, and this is
something that the FTC, which also, as we know, has fined
Facebook $5 billion for privacy violations, right, is the right
institution to oversee this.
We do need to have privacy baked into this, both in the
legislation and the regulations that would come out of it, but
this is a balance that can be struck. We need to be clear, you
know, and I think whether it's heaping more privacy protections
into the bill itself or just making clear through the iterative
process of the Agency that no, there should not be any
individual whose privacy is compromised by this, but we need to
make sure that, you know, someone other than the data
scientists at the firm are able to make the kind of inferences
that those on the inside are able to do.
Chair Coons. Thank you. I think it was framed really well,
I think it was by you, that we need to be able to have the rest
of the world have insights as comparable to those who are
actually designing, and operating, and running, and making
profit from these firms rather than our current situation,
where they know virtually everything and we know virtually
nothing.
Professor Haidt, if I might invite you to make a closing
comment if there's concerns, or issues, or questions that have
been raised by the testimony so far, I'd welcome. My question
to you is why is this important? Why act now? Why not simply
wait and let the market work itself out? You're muted.
Professor Haidt. Okay. Thank you. Thank you, Senator Coons.
First, I want to apologize to Senator Ossoff. I misunderstood
his question. He was asking when I recommended identity
authentication, is this going to put whistleblowers at risk.
The answer is no, because identity authentication doesn't mean
you post under your real name. You just prove that you are
somebody who you say you are to a third party, ideally a
nonprofit or something that would be protected. Even if you
were afraid of that, you could still just put up an anonymous
blog anywhere you want on the internet, and just tell a
journalist, ``Hey, here's the whistleblowing stuff on this
blog.'' Identity authentication would reduce a lot of the
garbage and the nastiness. Still, there's plenty ways to be
able to explore complete security.
With regards to your question about any sort of closing
statement, and especially why now, in the graphs in my
submitted testimony, I show that there are hockey stick-type
graphs. That is a long, straight line, and then it goes up, and
up, and up. I deliberately focused my analysis up to 2019,
before COVID because I didn't want to be confused with what
COVID has done. Everything I have said is so much worse now
that COVID has basically taken kids and given them less free
play outside and more time on their devices.
The problem keeps getting more and more serious. It's now
to the point where something on the order of a quarter of our
girls seem to have depressive disorder or severe anxiety. And
it keeps going up, and up, and up. We should have acted--we
should have acted 10 years ago. Well the data was only really
clear about 6 years ago. We should have acted then. We didn't,
and I think we have to act now.
Even if you're not convinced that I'm right about
causation, we don't have the data. We're trying--as my NYU
colleague, Josh Tucker, put it, researchers are working with
one hand tied behind our back. We have to do secondary studies
and try to look at shadows, whereas they've got the data. They
know exactly what the kids are consuming, and they know how
happy the kid is. They can code their content. Almost all of
the data is there. We can't get to it, so we're guessing. We're
flying blind. This really needs to change, and I hope that
Congress can help us here.
Chair Coons. I'd like to express my thanks to all of the
witnesses who have appeared today. I'm also grateful to the
Members who attended and asked thoughtful questions. I'm
particularly appreciative to Ranking Member Sasse for being a
great partner on this Subcommittee and allowing us to hold this
hearing.
It's reinforced my view that there is much more that
platforms can and must do to be more transparent about how
their products and services actually affect each of us, our
families, our community, our society, our democracy. I believe
that both the public and policymakers need better information
if we're going to understand and act on the impact of social
media and find better solutions that will allow us to take
advantage of all that social media has to offer, while limiting
the harms that it creates or even exacerbates.
It's clear we've got a lot more work to do, and I look
forward to working with my colleagues and with each of you. I,
really, again, deeply appreciate your constructive comment and
input today.
Members of this Committee can submit questions for the
record for the witnesses. They are due by 5 p.m., 1 week from
today, thus on May 11th. I want to thank our witnesses again
for participating in this compelling and engaging hearing.
With that, today's hearing is adjourned.
[Whereupon, at 3:49 p.m., the hearing was adjourned.]
[Additional material submitted for the record follows.]
[GRAPHIC(S) NOT AVAILABLE IN TIFF FORMAT]
[all]