[Senate Hearing 116-622]
[From the U.S. Government Publishing Office]
S. Hrg. 116-622
THE PACT ACT AND SECTION 230: THE IMPACT
OF THE LAW THAT HELPED CREATE THE
INTERNET AND AN EXAMINATION OF PROPOSED REFORMS FOR TODAY'S ONLINE
WORLD
=======================================================================
HEARING
before the
SUBCOMMITTEE ON COMMUNICATIONS, TECHNOLOGY, INNOVATION AND THE INTERNET
of the
COMMITTEE ON COMMERCE,
SCIENCE, AND TRANSPORTATION
UNITED STATES SENATE
ONE HUNDRED SIXTEENTH CONGRESS
SECOND SESSION
__________
JULY 28, 2020
__________
Printed for the use of the Committee on Commerce, Science, and
Transportation
[GRAPHIC(S) NOT AVAILABLE IN TIFF FORMAT]
Available online: http://www.govinfo.gov
______
U.S. GOVERNMENT PUBLISHING OFFICE
52-821 PDF WASHINGTON : 2023
ONE HUNDRED SIXTEENTH CONGRESS
SECOND SESSION
ROGER WICKER, Mississippi, Chairman
JOHN THUNE, South Dakota MARIA CANTWELL, Washington,
ROY BLUNT, Missouri Ranking
TED CRUZ, Texas AMY KLOBUCHAR, Minnesota
DEB FISCHER, Nebraska RICHARD BLUMENTHAL, Connecticut
JERRY MORAN, Kansas BRIAN SCHATZ, Hawaii
DAN SULLIVAN, Alaska EDWARD MARKEY, Massachusetts
CORY GARDNER, Colorado TOM UDALL, New Mexico
MARSHA BLACKBURN, Tennessee GARY PETERS, Michigan
SHELLEY MOORE CAPITO, West Virginia TAMMY BALDWIN, Wisconsin
MIKE LEE, Utah TAMMY DUCKWORTH, Illinois
RON JOHNSON, Wisconsin JON TESTER, Montana
TODD YOUNG, Indiana KYRSTEN SINEMA, Arizona
RICK SCOTT, Florida JACKY ROSEN, Nevada
John Keast, Staff Director
Crystal Tully, Deputy Staff Director
Steven Wall, General Counsel
Kim Lipsky, Democratic Staff Director
Chris Day, Democratic Deputy Staff Director
Renae Black, Senior Counsel
------
SUBCOMMITTEE ON COMMUNICATIONS, TECHNOLOGY, INNOVATION AND THE INTERNET
JOHN THUNE, South Dakota, Chairman
ROY BLUNT, Missouri BRIAN SCHATZ, Hawaii, Ranking
TED CRUZ, Texas AMY KLOBUCHAR, Minnesota
DEB FISCHER, Nebraska RICHARD BLUMENTHAL, Connecticut
JERRY MORAN, Kansas EDWARD MARKEY, Massachusetts
DAN SULLIVAN, Alaska TOM UDALL, New Mexico
CORY GARDNER, Colorado GARY PETERS, Michigan
MARSHA BLACKBURN, Tennessee TAMMY BALDWIN, Wisconsin
SHELLEY MOORE CAPITO, West Virginia TAMMY DUCKWORTH, Illinois
MIKE LEE, Utah JON TESTER, Montana
RON JOHNSON, Wisconsin KYRSTEN SINEMA, Arizona
TODD YOUNG, Indiana JACKY ROSEN, Nevada
RICK SCOTT, Florida
C O N T E N T S
----------
Page
Hearing held on July 28, 2020.................................... 1
Statement of Senator Thune....................................... 1
Statement of Senator Schatz...................................... 4
Statement of Senator Wicker...................................... 5
Statement of Senator Klobuchar................................... 59
Statement of Senator Fischer..................................... 61
Statement of Senator Blumenthal.................................. 63
Prepared statement from Nicole, Mother of a Child Whose
Sexually Abused Images Were Circulated Online.............. 64
Letter dated July 27, 2020 to Hon. John Thune and Hon. Brian
Schatz from Computer & Communications Industry Association,
Consumer Technology Association, Engine, and Internet
Infrastructure Coalition................................... 91
Statement of Senator Moran....................................... 67
Statement of Senator Udall....................................... 69
Statement of Senator Gardner..................................... 71
Statement of Senator Peters...................................... 74
Statement of Senator Cruz........................................ 76
Statement of Senator Lee......................................... 78
Statement of Senator Baldwin..................................... 80
Statement of Senator Blackburn................................... 82
Statement of Senator Tester...................................... 83
Statement of Senator Rosen....................................... 85
Witnesses
Hon. Christopher Cox, Counsel, Morgan, Lewis & Bockius, LLP;
Director, NetChoice............................................ 7
Prepared statement........................................... 9
Jeff Kosseff, Assistant Professor, Cyber Science Department,
United States Naval Academy.................................... 29
Prepared statement........................................... 30
Elizabeth Banker, Deputy General Counsel, Internet Association... 39
Prepared statement........................................... 40
Olivier Sylvain, Professor, Fordham Law School................... 49
Prepared statement........................................... 51
Appendix
Response to written question submitted to Jeff Kosseff by:
Hon. Shelley Moore Capito.................................... 95
Hon. Rick Scott.............................................. 95
THE PACT ACT AND SECTION 230:
THE IMPACT OF THE LAW THAT HELPED
CREATE THE INTERNET AND AN
EXAMINATION OF PROPOSED REFORMS FOR TODAY'S ONLINE WORLD
----------
TUESDAY, JULY 28, 2020
U.S. Senate,
Subcommittee on Communications, Technology,
Innovation and the Internet,
Committee on Commerce, Science, and Transportation,
Washington, DC.
The Subcommittee met, pursuant to notice, at 10 a.m., in
room SD-106, Dirksen Senate Office Building, Hon. John Thune,
Chairman of the Subcommittee, presiding.
Present: Senators Thune [presiding], Cruz, Fischer, Moran,
Gardner, Blackburn, Lee, Schatz, Klobuchar, Blumenthal, Udall,
Peters, Baldwin, Tester, and Rosen.
Also present: Senators Wicker, Ex Officio, and Cantwell, Ex
Officio.
OPENING STATEMENT OF HON. JOHN THUNE,
U.S. SENATOR FROM SOUTH DAKOTA
Senator Thune. Good morning. We'll get this hearing
underway.
I want to thank everybody for being here today, both
virtually and in person. Our panelists today are all appearing
virtually, so we look forward very much to hearing from all of
you.
We are here to examine the legacy of Section 230 of the
Communications Decency Act, which was enacted into law 24 years
ago, and to discuss a proposal Senator Schatz and I have
introduced to reform Section 230 known as the Platform
Accountability and Consumer Transparency Act, or the PACT Act.
Section 230 was written to protect Internet platforms--both
large and small--from being held liable for user-generated
content while also enabling these platforms to take an active
role in moderating such content.
The sweeping nature of these protections, coupled with
expansive readings by the courts, has come to mean that, with
few exceptions, Internet platforms are not liable for the
comments, pictures, and videos that their users and subscribers
post, no matter how harmful.
As one of our witnesses here today has written in what he
calls his biography of Section 230, the law's proposal and
passage flew under the radar back in 1996, receiving virtually
no opposition or media coverage. Today, however, Section 230 is
the subject of intense debate and media scrutiny, to the extent
that both the President of the United States and his likely
competitor in this fall's election have called for the complete
repeal of Section 230.
One of many variables that has sparked the intense debate
about Section 230 is that Internet platforms have actively
cultivated the notion that they are merely providing the
technology for people to communicate and share their thoughts
and ideas. Therefore, until only relatively recently, the
platforms largely concealed, or at the very least failed to
disclose, their moderation and curation systems to sustain this
fiction of being a neutral platform for all ideas.
Content moderation has, and largely continues to be, a
black box, which has led to deep suspicion by many users about
bias and discrimination. The reality is that platforms have a
strong incentive to exercise control over the content each of
us sees, because if they can present us with content that will
keep us engaged on the platform, we will stay on the platform
longer. Moderation is an important function that platforms must
provide in order to deliver a valuable experience to their
users. Unfortunately, it's hard for users to get good
information about how content is moderated.
The Internet has evolved significantly since Section 230
was enacted. Long gone are the days of the online bulletin
boards. Today, Internet platforms have sophisticated content
moderation tools, algorithms, and recommended engines to
promote content and connect users, all optimized toward keeping
every user engaged on the platform.
The platforms have monetized these systems through targeted
advertising and related businesses, and have consequently
become some of the largest companies in the world. Moreover,
these platforms have become essential to our daily lives, as
many Americans live, work, and communicate increasingly online.
That is why it is important to recognize that the benefits of
Section 230 for companies have come with tradeoffs for
consumers.
As the Department of Justice has noted in its
recommendations to reform Section 230, broad Section 230
immunity can pose challenges for Federal agencies in civil
enforcement matters. It is questionable whether Section 230 was
intended to allow companies to invoke immunity against the
Federal Government acting to protect American consumers in the
civil enforcement context. This has contributed to the creation
of a different set of rules for enforcing consumer protections
against online companies compared to those in the offline
world.
In addition, disparate complaint intake and transparency
reporting practices between Internet companies have led to a
limited ability for consumers to address and correct harms that
occur online. And, as Americans conduct more and more of their
activities online, the net outcome is an increasingly less
protected and more vulnerable consumer.
The Internet of 1996 is a far cry from the Internet of
2020. And, as Americans exist increasingly online, a trend now
being accelerated by the COVID-19 pandemic, as illustrated by
the fact that each of our witnesses is attending virtually
today, reevaluating Section 230 within today's context will
ensure its protections continue to balance the interests of
both consumers and companies.
Against this backdrop, the bill Senator Schatz and I have
introduced would update Section 230 to enable greater
transparency and accountability for users without damaging its
foundational economic, innovative, and entrepreneurial benefit
that helped allow the Internet to flourish in the first place.
The PACT Act would require companies that moderate content
to provide a clear and easily accessible user policy that
explains how, when, and why user-generated content might be
removed. It would also require these online platforms to create
a defined complaint system that processes reports and notifies
users of moderation decisions within 14 days.
Our legislation would require large technology companies to
have a toll-free customer-service phone line with live customer
support to take customer complaints. This requirement is geared
toward consumers who are less familiar with technology and
those in marginalized communities who may not have readily
available access to technology, but who want or need to talk to
a real person about a complaint about content on a service or
platform.
The PACT Act would also hold platforms accountable for
their content moderation practices by requiring them to submit
quarterly reports to the Federal Trade Commission outlining
material they've removed from their sites or chosen to de-
emphasize.
In addition, the PACT Act would make it clear that the
immunity provided by Section 230 does not apply to civil
enforcement actions by the Federal Government. The PACT Act
would also make clear that Section 230 does not apply where a
platform is provided with a court order finding that content is
unlawful.
Both of these provisions are also recommendations that the
Department of Justice recently put forward in its recent review
of Section 230. At its core, Section 230 reform is about
balancing the consumers' need for transparency and
accountability against Internet companies' need for flexibility
and autonomy. I believe the PACT Act strikes the right balance,
and I'm committed to achieving a meaningful bipartisan approach
to Section 230 reform that can be enacted into law sooner
rather than later.
However, I recognize that the Internet is complex and any
meaningful regulation must consider various perspectives from
diverse groups in academia, civil society, and industry.
Consequently, we have brought together today a very
distinguished panel, and I'm confident the conversation will
help ensure that we are reforming Section 230 in the right way.
Each of our witnesses has deep expertise in both the original
intent of Section 230 and how it has been interpreted by the
courts over the years.
Today we're joined by former Representative Chris Cox, the
coauthor of Section 230 of the Communications Decency Act; Jeff
Kosseff, Assistant Professor of Cyber Science at the United
States Naval Academy; Elizabeth Banker, Deputy General Counsel
of the Internet Association; and Olivier Sylvain, Professor of
Law at Fordham School of Law. Thanks to each of you for
participating on this important topic.
And I'm now going to recognize Senator Schatz, who will be
joining us remotely for his opening statement.
Senator Schatz.
STATEMENT OF HON. BRIAN SCHATZ,
U.S. SENATOR FROM HAWAII
Senator Schatz. Thank you, Mr. Chairman, for holding this
hearing today to discuss Section 230 of the Platform
Accountability and Consumer Transparency Act, the PACT Act.
Before we go any further, I just want to offer some thanks
to the Chairman of the Subcommittee, Chairman Thune. Our
process has been serious, it has been bipartisan. It is the way
the Senate should work. It is the way the Commerce Committee
should work. And I'm proud to partner with him on this
legislation.
Unfortunately, a lot of the discussion around Section 230
has been focused on provocative but sometimes reactionary and,
in some cases, unconstitutional ideas based on the perceived
political slights of the day. And there may be some who try to
use this hearing as an opportunity to create a clip for social
media or to make a few partisan headlines. But, that's not what
this hearing is for. We are here to legislate. The work we are
doing here today is a serious effort to review Section 230
objectively and on a bipartisan basis, to evaluate how this law
should be amended to benefit the American people. This work is
already difficult, and it is made much more difficult by
grandstanding.
Section 230 was, by all accounts, a prescient and novel
idea back in 1996, when it became law. It prevented online
platforms from being treated as the publisher or speaker of
third-party content, and that avoided liability for their
users' content, and, in so doing this, allowed innovators in
the United States to create and build products using third-
party content without the threat of litigation. This unique
idea is one reason why the largest tech companies began in the
United States. And, because they started here, many of these
platforms became a vehicle for the spread of free speech and
democratic ideals across the planet. But, today's Internet is
different from when Section 230 is adopted.
And this brings me to a fundamental point. It is OK to
update a law. It doesn't mean you think the law was badly
written or is deeply, deeply flawed. It just means that, as the
Telecommunications Act is periodically amended, as the National
Defense Authorization Act is periodically amended, that this
law needs to be updated so that it continues to work well.
Now that the Internet has evolved, it is important to ask
how Section 230 should evolve along with it. Last month, I
introduced the PACT Act with Senator Thune. The bill amends
Section 230 and imposes new responsibilities, but not
unreasonable ones, on online platforms. It focuses on three
concepts: accountability, transparency, and protections for
consumers online. To make platforms more accountable to their
consumers, the bill requires platforms to respond to consumer
complaints about content that is against their own acceptable-
use policies or that a court has already determined to be
illegal; to improve transparency by requiring platforms to
publish reports so that people know what a platform is doing to
moderate, based on its own rules; and it increases online
consumer protections by fixing the current legal disparities
between online and offline commerce and communications.
Some view the debate about Section 230 reform as an
opportunity to work the refs or claim bias or make people
fearful of the enforcement of content moderation policies. Our
approach has been different. This bill is not targeted at a
specific type of content, business model, or company, and its
purpose is not to censor or control, or even influence, free
speech. Diverse viewpoints make us stronger as a Nation, it's
better for the Internet, and I believe that we should preserve
robust protections that enable discourse in our country online
and offline.
But, I'm proud to be working on such a measured approach to
Section 230 reform, and I appreciate the partnership with
Senator Thune and his excellent staff. This has been truly a
bipartisan effort, and I thank him for the hard work.
Section 230 proponents say that Congress can't possibly
change this law without disrupting all of the great innovation
that it has enabled. And I just disagree with that. The
legislative process is about making sure that our laws are in
the public interest, and the PACT Act offers some commonsense
changes to the way that the statute functions so consumers have
protections, so platforms have accountability and transparency,
and so that the statute works today for the Internet.
I want to thank the witnesses for joining us and sharing
their expertise on this issue. As part of its jurisdictional
oversight on these issues, the Subcommittee looks forward to
hearing from you about how the PACT Act's provisions might be a
realistic step toward modernizing Section 230. And I look
forward to hearing their testimony.
Senator Thune. Thank you, Senator Schatz. And it has been
great partnering with you and your team, and producing
something that I think really does represent a good, balanced,
sound approach to an issue that has vexed those from all
persuasions and perspectives for some time, and I hope that we
can make some progress and move this legislation forward.
We are joined today by the distinguished Chairman of the
Full Committee, Senator Wicker, and I'm going to recognize him
to make some opening remarks.
Senator Wicker.
STATEMENT OF HON. ROGER WICKER,
U.S. SENATOR FROM MISSISSIPPI
The Chairman. Thank you very much, Mr. Chairman. And I do
want to congratulate you and Senator Schatz for working
together to solve this matter in a bipartisan fashion.
This is a very important hearing to examine Section 230 of
the Communications Decency Act. And I want to extend our thanks
and appreciation and welcome to the witnesses for appearing
with us remotely.
Section 230 of the Communications Decency Act was enacted
in 1996 as part of broader reforms to the Communications Act.
Section 230 protects interactive computer services, such as
social media platforms, from being held liable for the content
posted by their users. Section 230 also specifically allows an
interactive computer service, acting in good faith, to restrict
the availability of content that it considers to be obscene,
lewd, lascivious, filthy, excessively violent, harassing, or
otherwise objectionable. And I emphasize ``otherwise
objectionable.'' The intent of the law, as codified in the
statute, is to preserve a vibrant and competitive online
marketplace for the benefit of all Americans. Indeed, a portion
of the title chosen by the Subcommittee Chair for this hearing
is ``The Impact of the Law that Helped Create the Internet.''
True words.
At the time of its enactment, the Internet was in its
infancy. No one could have imagined the success of the digital
economy we enjoy today. Section 230 has underpinned much of the
Internet's growth and development. It has enabled social media
platforms, app developers, websites, bloggers, and others, to
host a variety of content and support the free flow of
information and ideas without being held legally responsible
for the materials generated by users. It has also empowered
interactive computer services to remove content that may
diminish the safety and security of the Internet.
Despite the vast economic and social benefits of the law,
however, I have been deeply troubled by recent reports that
suggest some online platforms are disproportionately censoring
conservative voices or posing an unfair bias through their
policies and terms of service. The administration's executive
order on preventing online censorship calls attention to these
issues.
As the Committee with jurisdiction over Section 230 of the
Communications Decency Act, it is our responsibility to ensure
that the law is applied consistently, fairly, and objectively.
To ensure greater accountability to the law, this may
necessitate a review of these statutes' legal shield for social
media companies and others to remove content that they, in
their sole discretion, deem to be, quote, ``otherwise
objectionable.'' In particular, I question whether this term is
too broad and improperly shields online platforms from
liability when they remove content that they simply disagree
with, dislike, or find distasteful. Such a term may require
further defining to reduce ambiguity, increase accountability,
and prevent misapplication of the law.
This morning, I hope witnesses will discuss the types of
content that fall under the category of ``otherwise
objectionable.'' I hope witnesses will also discuss the process
by which interactive computer services objectively determine
what constitutes ``otherwise objectionable content,'' and how
that process is communicated to users in order to preserve a
true diversity of political discourse online, as intended by
the law. This will help inform the Committee's efforts to
maintain a free and open Internet that promotes competition and
innovation, and protects multiple viewpoints.
Again, I thank Chairman Thune and Ranking Member Schatz for
convening this important hearing.
Thank you.
Senator Thune. Thank you, Chairman Wicker.
We will now turn to our panel. And, as I mentioned, we have
with us a former House member, Chris Cox, a former colleague of
both Senator Wicker and I, and delighted to have him back here.
We look forward to hearing from you, Chris. He is now speaking
here on behalf of NetChoice. As I said, Mr. Jeff Kosseff, Ms.
Elizabeth Banker, and Mr. Olivier Sylvain. And my apologies for
getting your name wrong the first time.
We'll start with Chris Cox.
Chris, welcome. And it's great to have you here. We look
forward to hearing from you. Please proceed.
And I would say, to all of our panelists, to the degree--I
know it's hard for you there, probably; you don't have a
clock--but, if you could contain or confine your oral remarks
to about 5 minutes, it'll maximize the amount of time we have
to ask questions and get your responses. And all your comments
will be made a part of the permanent hearing record.
So, Mr. Cox, you are recognized.
STATEMENT OF HON. CHRISTOPHER COX, COUNSEL, MORGAN, LEWIS &
BOCKIUS, LLP; DIRECTOR, NETCHOICE
Mr. Cox. Well, thank you very much, Chairman Thune and
Ranking Member Schatz and members of the Subcommittee. Thank
you for your invitation to join you in exploring these issues
this morning.
I want to apologize for this voice. Since you last heard
from me, I've had some surgery that, as an unfortunate side
effect, left me with one of my two vocal cords paralyzed. But,
I promise to give you, this morning, my full 50 percent.
I should also state at the outset that the views I express
this morning are my own and not necessarily those of Morgan,
Lewis & Bockius or of NetChoice, where I am a Director.
Those of you who were here in 1995 and 1996 will remember
the debate over pornography on the Internet that gave birth to
the CDA and, indirectly, Section 230, a quarter century ago. At
the time, wayward court decisions really threatened the future
of the Internet. A web portal that had done the good deed of
screening some user-generated content was being held
responsible, therefore, for screening all of it. And, under
that unfortunate rule, the good deed of at least trying to keep
the Internet free from objectionable material would have been
punished.
So, the bill that I wrote to eliminate this perverse
incentive, cosponsored by our then House colleague, Ron Wyden,
is what eventually became what we now know as Section 230.
Looking across the intervening decades of judicial
interpretation of Section 230, we can see that the law has
contributed directly to the success of the Internet by
providing a legal foundation for user-generated content today
shared not just among millions, but billions, of people.
We think about the remarkable accomplishment of Wikipedia,
something that many of us use almost daily and take for
granted, it's long since outstripped the information that was
contained in the once unparalleled Encyclopedia Britannica.
It's just one marvel of the 21st century that we take for
granted. Wikipedia relies entirely upon user-generated content.
It's operated by the Wikimedia Foundation, which is, itself, a
small organization funded by voluntary contributions. If it
were subject to lawsuits for the contributions and comments of
its volunteers and users, it couldn't sustain itself, and it
would cease to exist as an invaluable free resource for every
American.
The fundamental objective of Section 230 has always been to
protect the innocent and punish the guilty. The law achieves
this objective by protecting websites that host user-created
content when, in good faith, they become involved in content
creation for the purpose of keeping objectionable material off
of their sites, or editing content created by others, or taking
it down altogether in order to remove offensive material. To
this extent, the law says they will not be treated as
publishers.
At the same time, Section 230, as written and as
interpreted today, makes clear that becoming involved in
content creation for any other purpose eliminates any
protection from suit. And that's true even if the involvement
in content creation is only partial. And it's true even if the
Internet platform doesn't, itself, create the content, but only
develops it. And when the platform is only partly responsible
for the mere development, not necessarily the creation, it
still loses its Section 230 protection. If a website is in any
way complicit in the creation or development of illegal
content, it has no Section 230 immunity. The inclusion of this
clear language in the statute was absolutely deliberate. It was
intended to ensure that both criminal and civil laws would
continue to be vigorously enforced. And that's why Section 230
expressly states that Federal criminal law is entirely
unaffected by its provisions, and neither is there any effect
on the enforcement of State law, whether civil or criminal,
provided that the State laws are enforced consistently with the
uniform national policy expressed in Section 230.
That uniform national policy applies equally to all civil
and criminal offenses. It's important that there be a uniform
national policy, because the Internet is the quintessential
vehicle of interstate commerce, and its packet-switched
architecture makes it uniquely susceptible to multiple sources
of conflicting State and local regulation. Even an e-mail from
this hearing room to someone in the Capitol across the street
can be broken up into pieces and routed through servers in
different states. If every state were free to adopt its own
policy governing when an Internet platform will be liable, not
only would compliance become oppressive, but the Federal policy
itself would quickly become undone.
Section 230 changed none of the legal responsibilities of
any individual or business or nonprofit. The same legal rules
continue to apply on the Internet, just as in the offline
world. What Section 230 added to the general body of law was
the principle that an individual or entity operating a website
could not, in addition to its own legal responsibilities, be
required to monitor all of the content created by third parties
in order to avoid becoming derivatively liable for the illegal
acts of other people.
Congress recognized that to require otherwise would deprive
all of us of the essential benefit of the Internet: the
opportunity for realtime communication among millions of people
around the world. Section 230 succeeded in safeguarding this
quintessential aspect of the Internet. Today, there are over
370 million active websites hosted in America, and over 875
million websites accessible to American users.
But, despite the tremendous variety this represents, most
of the legislation now being drafted in the House and in the
Senate to regulate these websites seems focused on a very
different paradigm and a much smaller group of companies. The
paradigm of what needs to be regulated seems to be an enormous
rapacious company interested only in manipulating its customers
or strangling democracy in America. That doesn't come close to
describing even the largest e-commerce sites, which are mostly
the traditional brick-and-mortar companies, including Kohl's,
Target, Costco, and Walmart, and the Big Tech paradigm
certainly doesn't describe the hundreds of thousands of other
websites, of all sizes, that bring us user-created content
every day. Yet all of these websites, and, more importantly,
all of us who are their users, rely on the protection of
Section 230 to have access to the many services they provide.
So, as you consider whether--and, if so, how--to legislate
in this area, it's important to remember just how much human
activity is encompassed within this vast category we so
casually refer to as ``the Internet.'' To the extent that any
new legislation imposes too much compliance burden or too much
liability exposure that's connected to a website's hosting of
user-created content, the risk is that too many websites will
be forced to respond by getting rid of user-generated content
altogether or else scaling it way back. Either way, millions of
Internet users in the United States would feel the loss
immediately.
I feel confident in saying that if writing Section 230 a
quarter century ago was a daunting undertaking, amending it
today presents a far greater challenge. Any changes you make
will affect every business, every nonprofit, and every
individual in America in some ways that you can't even predict
and that will inevitably disappoint you.
For that reason, I commend all of you on this committee for
taking the thoughtful approach that you are and making every
effort to inform yourselves about the endless real-world
consequences before legislating in this area.
I look forward to your questions.
[The prepared statement of Mr. Cox follows:]
Prepared Statement of Chris Cox, Former U.S. Representative, Author and
Co-Sponsor with Senator Ron Wyden, Section 230
Chairman Thune, Ranking Member Schatz, and Members of the
Subcommittee, thank you for the invitation to testify on the history of
Section 230 and its application by the courts over the last quarter
century. This experiential base is an important starting point as you
consider ways to ensure that platforms are accountable for their
content moderation practices, and what legislative measures, from
transparency to accountability tools, can empower consumers online.
My abiding interest in this subject dates, of course, to some
twenty-four years ago when I joined then-Rep. Ron Wyden in writing what
today is known as Section 230. In the intervening quarter century I
have followed the developments in the case law, sometimes with awe and
occasionally with disappointment. The views I express today are my own,
and not necessarily those of NetChoice, on whose board I serve, or of
Morgan, Lewis & Bockius.
Introduction
As we consider the issues surrounding free expression and content
moderation on the internet, it is worth asking: what would our world be
like without Section 230?
This is an important question because, to a degree most of us fail
to recognize, we take its many benefits for granted. An endless variety
of useful and important content on the Internet is supplied not by
websites or social media platforms, but by their millions of users who
create the content themselves and freely share it. Without Section 230,
millions of American websites--facing unlimited legal liability for
what their users create--would not be able to host user-generated
content at all.
In this way, moreover, Section 230 facilitates every individual's
ability to publish their own content on the internet. The wide variety
of online forums for posting user-created content is the direct result
of the protection from liability for the host sites that Section 230
affords.
At the same time that Section 230 has enabled an endless diversity
of voices to reach far greater audiences than was ever possible before,
this same law has helped websites to maintain civility and fair play
through the application of bespoke standards of content moderation. In
contrast to other nations, in the United States the government does not
dictate what can be published on the Internet and who can publish it.
The proliferation of websites, each free to adopt their own rules of
the road, has simultaneously provided unparalleled opportunities for
any individual to reach millions of people around the world--and the
means by which offensive online conduct including bullying and
obscenity, as well as outright criminal activity, can be restricted
without fear of legal liability.
Before the enactment of Section 230, Internet platforms faced a
terrible dilemma. If they sought to enforce even minimal rules of the
road in order to maintain civility and keep their sites free from
obscenity and obnoxious behavior, they became unlimitedly liable for
all of the user-created content on their site.\1\ On the other hand, if
the website followed an ``anything goes'' business model, with no
content moderation whatsoever, then it could completely avoid that
liability.\2\ From the perspective of any Internet platform that
attempted to maintain a family-friendly site, it was a classic case of
``no good deed goes unpunished.''
---------------------------------------------------------------------------
\1\ Stratton Oakmont v. Prodigy Servs. Co., 1995 WL 323710
(N.Y.Sup.Ct. May 24, 1995).
\2\ Cubby, Inc. v. CompuServe, Inc., 776 F. Supp. 135, 140
(S.D.N.Y. 1991).
---------------------------------------------------------------------------
Section 230 eliminated the perverse incentive for ``anything
goes.'' By imposing liability on criminals and tortfeasors for their
own wrongful communications and conduct, rather than shifting that
liability to a website that did not in any way participate in the
wrongdoing, it freed each website to clean up its corner of the
internet. No longer would being a ``Good Samaritan'' buy trouble.
In an imagined future world without Section 230, where websites and
Internet platforms again face enormous potential liability for hosting
content created by others, there would again be a powerful incentive to
limit that exposure. Online platforms could accomplish this in one of
two ways. They could strictly limit user-generated content, or even
eliminate it altogether; or they could adopt the ``anything goes''
model that was the way to escape liability before Section 230 existed.
We would all be very much worse off were this to happen. Without
Section 230's clear limitation on liability it is difficult to imagine
that most of the online services on which we rely every day would even
exist in anything like their current form.
As Congress considers whether to amend Section 230, therefore, it
is important to keep in mind the many aspects of the modern Internet we
take for granted, and that are dependent upon Section 230's
protections. Compromising those protections risks a wide array of
unintended consequences. Among these are loss of much of the rich
content provided every day by millions of individual content creators,
loss of the ability to use social media for real time communication
with friends and family, loss of opportunities for diverse voices to
reach broad audiences throughout the Nation and across the planet, and
damage to e-commerce and the continued technological development of the
internet.
In the 21st century, Section 230's protection of website operators
from liability for content created by their users operates as an
essential buttress of free expression. It is the key to millions of
Americans' ability to share news and views and gain instant access to a
wide range of informational and educational resources. It is the
foundation supporting e-commerce sites such as Yelp, eBay, Facebook,
Wikipedia, Amazon, Twitter, and the entire Web 2.0 revolution whereby
thousands of innovative platforms offer a range of useful services
powered by user-generated content. From user-created reviews of
products and services, to educational videos, to online resources that
help locate loved ones after natural disasters, to the many online
services that have come to the rescue of millions of Americans
quarantined or in self-isolation during the Covid pandemic, all of the
rich variety now available at our fingertips is precisely what Section
230 was designed to facilitate.
But while Section 230 has been a boon to users of the Internet and
to the continued technological and commercial development of the
Internet itself, it is not a panacea. We are all familiar with the many
pathologies that continue to fester in corners of the dark web, and
that too often leach onto the mainstream Internet websites and
platforms we rely on every day. Continued challenges to a free and open
Internet include the threat of hidden platform ``censorship'' and
undisclosed viewpoint discrimination; ``fake news'' and content
manipulation by bad actors; defamation, cyberstalking, and revenge
porn; fraud on consumers; internet-facilitated criminal gang activity;
cross-border terrorism; child sexual abuse and sex trafficking; and
widespread censorship and social control by dictatorships and
authoritarian governments including not only Russia and China, but
scores of other nations besides.
Section 230 has not prevented these affronts, but neither is it the
cause of them. In many cases, it has helped mitigate their
consequences. Preserving the law's benefits for Internet users,
society, and the Nation's economy should remain an overarching
objective of any legislation to address the many looming concerns
across the rapidly-evolving landscape of the internet.
In that respect, we will be well advised to recognize the danger of
unintended consequences that would accompany efforts to reopen Section
230 to further amendment. There are many competing interests at stake,
both commercially and politically, in the constellation of issues
affected by Section 230. As each of you is fully aware, the various
criticisms of Section 230 come from disparate quarters, and are based
on radically different rationales. For example, while some critics
demand more robust content moderation, their political opposites demand
less interference with user-created content. The process of turning a
bill into law in these circumstances will require potentially trenchant
compromises.
The multiplicity of stakeholders, including competing business
interests, affected by any new legislation governing activity on the
internet--not to mention the many different committees that will be
involved in both the House and Senate, and the inevitable need to
compromise among them in order for a bill to make it through the entire
process--means that you may not recognize your legislative handiwork in
the final product. So even though it is possible to imagine that a
``perfect'' bill might emerge from the Commerce Committee that would
clarify and improve Section 230 while preserving all of its benefits,
the legislative process that will inevitably follow is likely to
adulterate that ``perfection'' and potentially threaten the essential
elements of Section 230 that make it work. This very real risk to the
many societal benefits that a majority of Congress still believes flow
from Section 230 is worth considering before opening what could be a
Pandora's box.
Background and Legislative History of Section 230
Section 230 was signed into law 24 years ago, in 1996.\3\ When my
colleague Ron Wyden (D-OR) and I conceptualized the law in 1995,
roughly 20 million American adults had access to the internet, compared
to 7.5 billion today.
---------------------------------------------------------------------------
\3\ 104 P.L. 104, 110 Stat. 56.
---------------------------------------------------------------------------
Those who were early to take advantage of the opportunity to ``surf
the web,'' including many in Congress, quickly confronted this
essential aspect of online activity: on each website, many users
converge through one portal. The difference between newspapers and
magazines, on the one hand, and the World Wide Web (as it was then
called), on the other hand, was striking. In the print world, a single
staff of human beings reviewed and cataloged editorial content that was
then distributed to a large number of passive recipients. The same was
true of television and radio. On the web, in contrast, millions of
users themselves created content which became accessible to the entire
planet immediately. While the volume of users was only in the millions,
not the billions as today, it was even then evident to almost every
user of the web that no group of human beings would ever be able to
keep pace with the growth of content on the internet.
At the time, however, not all in Congress were users of the web who
appreciated these fundamentals. The Communications Decency Act
(``CDA''), introduced in the Senate by James Exon (D-NE), was premised
on the notion that the FBI could filter the web, screening out
offensive content. This was a faulty premise based on a fundamental
misunderstanding of the scale and the functioning of the internet.
Nonetheless, in large part because the stated target of the CDA was
pornography, the Senate voted overwhelmingly (the vote was 84-16) in
favor of it.\4\
---------------------------------------------------------------------------
\4\ Id.
---------------------------------------------------------------------------
Section 230 was not part of the CDA. Instead, it was a freestanding
bill introduced in the House as H.R. 1978, the Internet Freedom and
Family Empowerment Act, in June 1995. It was intended as an alternative
to the CDA. When it was offered as a standalone floor Cox-Wyden
amendment during consideration of the Telecommunications Act in August
1995, it was roundly endorsed on both sides of the aisle during debate.
At the same time, both Democratic and Republican lawmakers sharply
criticized the CDA. They then voted nearly unanimously in favor of the
Cox-Wyden amendment, while excluding the CDA from the House version of
the Telecommunications Act.
In the conference on what became the Telecommunications Act of 1996
that followed, as is so often the case in legislative compromises
between House and Senate, the conferees on agreed to include both
diametrically opposed bills in the Conference Report. Subsequently, the
U.S. Supreme Court gutted the CDA's indecency provisions, which it
found violated the First Amendment, giving Rep. Wyden and me the
victory we did not at first achieve in conference.\5\
---------------------------------------------------------------------------
\5\ Reno v. American Civil Liberties Union, 521 U.S. 844 (1997).
---------------------------------------------------------------------------
The fundamental flaw of the CDA was its misunderstanding of the
Internet as a medium. We can now easily see that it would have been
impossible for the bulletin boards, chat rooms, forums, and e-mail that
were then budding on the web to be screened in any meaningful way by
the FBI, or by the operators of individual websites themselves, even at
the far lower volumes of traffic that existed then. Worse, if the law
were to demand such screening, the fundamental strength of the new
medium--facilitating the free exchange of information among millions of
users--would be lost.
The Prodigy and CompuServe cases
The impetus for the Internet Freedom and Family Empowerment Act,
today's Section 230, was a New York Superior Court case that I first
saw reported in the Wall Street Journal in May 1995.\6\ It involved one
of the leading Internet portals of the day. The case concerned an
allegedly defamatory bulletin board post on the Prodigy web service by
an unknown user. The post claimed that an investment bank and its
founder, Jordan Belfort, had committed securities fraud. (The post was
not in fact defamatory: Belfort was later convicted of securities
fraud, but not before Prodigy had settled the case for a substantial
figure. Belfort would achieve further infamy when he became the model
for Leonardo DiCaprio's character in ``The Wolf of Wall Street.'')
---------------------------------------------------------------------------
\6\ Milo Geyelin, New York judge rules Prodigy responsible for on-
line content, Wall St. Jo., May 26, 1995.
---------------------------------------------------------------------------
By holding Prodigy liable for the allegedly illegal content posted
by its user, the New York court established a new precedent with far-
reaching consequences.\7\ Up until then, the courts had not permitted
such claims for third-party liability. In 1991, a Federal district
court in New York held that CompuServe, another web service similar to
Prodigy that hosted a variety of user-created content, was not liable
in circumstances very similar to those in the Prodigy case. The court
reasoned that CompuServe ``had no opportunity to review the contents of
the publication at issue before it was uploaded into CompuServe's
computer banks,'' and therefore was not subject to publisher liability
for the third party content.\8\
---------------------------------------------------------------------------
\7\ Stratton Oakmont v. Prodigy Servs. Co., 1995 WL 323710
(N.Y.Sup.Ct. May 24, 1995)
\8\ Cubby, Inc. v. CompuServe, Inc., 776 F. Supp. 135, 140
(S.D.N.Y. 1991) (emphasis added).
---------------------------------------------------------------------------
But in the 1995 New York Superior Court case, the court
distinguished the CompuServe precedent. The reason the court offered
was that unlike CompuServe, Prodigy sought to impose general rules of
civility on its message boards and in its forums. While Prodigy had
even more users than CompuServe and thus even less ability to screen
material on its system, the fact it had announced rules of the road and
occasionally enforced them was the judge's basis for subjecting it to
liability that CompuServe didn't face.
The perverse incentive this case established was clear: any
provider of interactive computer services should avoid even modest
efforts to moderate the content on its site. The inevitable
consequences for the future of the Internet were equally clear: every
website would be incentivized to follow CompuServe's model of
``anything goes.'' Unless corrective action were taken the internet,
already beginning to show some erosion in standards of public discourse
that must inevitably arise when thousands and then millions of people
engage in uninhibited public expression on any topic, would quickly
become nothing but a sewer. When I read about this decision, I
immediately set to work on drafting a bill to head off its predictable
bad consequences.
Creating Section 230 and its goals
The first person I turned to as a legislative partner on my
proposed bill was then-Rep. Ron Wyden (D-OR). We had previously agreed
to seek out opportunities for bipartisan legislation. As this was a
novel question of policy that had not hardened into partisan
disagreement (as was too often the case with so many other issues), we
knew we could count on a fair consideration of the issues from our
colleagues on both sides of the aisle.
For the better part of a year, we conducted outreach and education
on the challenging issues involved. In the process, we built not only
overwhelming support, but also a much deeper understanding of the
unique aspects of the Internet that require clear legal rules for it to
function.
The rule established in the Internet Freedom and Family Empowerment
Act,\9\ which we introduced in June 1995, was crystal clear: the
government would impose liability on criminals and tortfeasors for
wrongful conduct. It would not shift that liability to third parties,
because doing so would directly interfere with the essential
functioning of the internet.
---------------------------------------------------------------------------
\9\ Internet Freedom and Family Empowerment Act, H.R. 1978, 104
Cong. (1995).
---------------------------------------------------------------------------
Rep. Wyden and I were well aware that whether a person is involved
in criminal or tortious conduct is in every case a question of fact.
Simply because one operates a website, for example, does not mean that
he or she cannot be involved in lawbreaking. To the contrary, as the
last two decades of experience have amply illustrated, the internet--
like all other means of telecommunication and transportation--can be
and often is used to facilitate illegal activity.
Section 230 was written, therefore, with a clear fact-based test:
Did the person create the content? If so, that person is
liable for any illegality.
Did someone else create the content? Then that someone else
is liable.
Did the person do anything to develop the content created by
another, even if only in part? If so, the person is liable
along with the content creator.
The plain language of the statute directly covers the situation in
which someone (or some company) is only partly involved in creating the
content. Likewise, it covers the situation in which they did not create
the content but were, at least in part, responsible for developing it.
In both cases, Section 230 comes down hard on the side of law
enforcement. A website operator involved only in part in content
creation, or only in part in the development of content created by
another, is nonetheless treated the same as the content creator.
Here is the precise language of section 230 in this respect:
The term ``information content provider'' means any person or
entity that is responsible, in whole or in part, for the
creation or development of information provided through the
Internet. . . .\10\
---------------------------------------------------------------------------
\10\ 47 USC Sec. 230(f) (emphasis added).
These words in Section 230--``in part'' and ``development of''--are
the most important part of the statute. That is because in enacting
Section 230, it was not our intent to create immunity for criminal and
tortious activity on the internet. To the contrary, our purpose (and
that of every legislator who voted for the bill) was to ensure that
innocent third parties will not be made liable for unlawful acts
committed wholly by others.
If an interactive computer service becomes complicit, in whole or
in part, in the creation of illicit content--even if only by partly
``developing'' the content--then it is entitled to no Section 230
protection.
Rep. Wyden and I knew that, in light of the volume of content that
even in 1995 was crossing most Internet platforms, it would be
unreasonable for the law to presume that the platform will screen all
material. We also well understood the corollary of this principle: if
in a specific case a platform actually did review material and edit it,
then there would be no basis for assuming otherwise. As a result, the
plain language of Section 230 deprives such a platform of immunity.
We then created an exception to this deprivation of immunity, for
what we called a ``Good Samaritan.'' \11\ If the purpose of one's
reviewing content or editing it is to restrict obscene or otherwise
objectionable content, then a platform will be protected. Obviously,
this exception would not be needed if Section 230 provided immunity to
those who only ``in part'' create or develop content.
---------------------------------------------------------------------------
\11\ 47 U.S.C. Sec. 230 (c)(2)(A).
---------------------------------------------------------------------------
The importance of Section 230 for user-generated content
In simplest terms, Section 230 protects website operators that are
not involved in content creation from liability for content created by
third party users. Without it, websites would be exposed to lawsuits
for everything from users' product reviews to book reviews. Yelp would
be exposed to lawsuits for its users' negative comments about
restaurants, and Tripadvisor could be sued for a user's disparaging
review of a hotel. Any service that connects buyers and sellers,
workers and employers, content creators and a platform, victims and
victims' rights groups--or provides any other interactive engagement
opportunity we can imagine--would face open-ended liability if it
continued to display user-created content.
How important is user-created content? Without it, it is hard to
imagine how any of us would have made it this far through the Covid
quarantines and self-isolation of 2020. Many contending with this
year's devastating tornadoes--this is already the deadliest tornado
season in the United States since 2011--could not have found their
loved ones. This year more than ever, millions of Americans are relying
on ``how to'' and educational videos for everything from healthcare to
home maintenance. During the Covid crisis, online access to user-
created pre-K, primary, and secondary education and lifelong learning
resources has proven a godsend for families across the country.
Over 85 percent of businesses rely on user-created content on their
websites.\12\ The vast majority of Americans feel more comfortable
buying a product after researching user generated reviews,\13\ and over
90 percent of consumers find user-generated content helpful in making
their purchasing decisions.\14\ User generated content is vital to law
enforcement and social services. Following the recent rioting in
several U.S. cities, social workers have been able to match people with
supplies and services to victims who needed life-saving help, directing
them with real-time maps.
---------------------------------------------------------------------------
\12\ https://www.semrush.com/blog/50-stats-about-9-emerging-
content-marketing-trends-for-2016/
\13\ Wu, Y. (2015). What Are Some Interesting Statistics About
Online Consumer Reviews? Dr4ward.com. Available at: http://
www.dr4ward.com/dr4ward/2013/03/what-are-some-interesting-statistics-
about-online-consumer-reviews-infographic.html
\14\ Kimberly Morrison, ``Why Consumers Share User-Generated
Content,'' Adweek, May 17, 2016.
---------------------------------------------------------------------------
Protecting the innocent and punishing the guilty
Throughout the history of the internet, Congress has sought to
strike the right balance between opportunity and responsibility.
Section 230 is such a balance--holding content creators liable for
illegal activity while protecting Internet platforms from liability for
content created entirely by others. At the same time, Section 230 does
not protect platforms liable when they are complicit--even if only in
part ‒ in the creation or development of illegal content.
The plain language of Section 230 makes clear its deference to
criminal law. The entirety of Federal criminal law enforcement is
unaffected by Section 230. So is all of state law that is consistent
with the policy of Section 230.\15\
---------------------------------------------------------------------------
\15\ 47 USC Sec. 230(e)(3).
---------------------------------------------------------------------------
Still, state law that is inconsistent with the aims of Section 230
is preempted. Why did Congress choose this course? First, and most
fundamentally, it is because the essential purpose of Section 230 is to
establish a uniform Federal policy, applicable across the internet,
that avoids results such as the state court decision in Prodigy.\16\
The Internet is the quintessential vehicle of interstate, and indeed
international, commerce. Its packet-switched architecture makes it
uniquely susceptible to multiple sources of conflicting state and local
regulation, since even a message from one cubicle to its neighbor
inside the same office can be broken up into pieces and routed via
servers in different states.
---------------------------------------------------------------------------
\16\ Stratton Oakmont v. Prodigy Servs. Co., 1995 WL 323710
(N.Y.Sup.Ct. May 24, 1995).
---------------------------------------------------------------------------
Were every state free to adopt its own policy concerning when an
Internet platform will be liable for the criminal or tortious conduct
of another, not only would compliance become oppressive, but the
Federal policy itself could quickly be undone. All a state would have
to do to defeat the Federal policy would be to place platform liability
laws in its criminal code. Section 230 would then become a nullity.
Congress thus intended Section 230 to establish a uniform Federal
policy, but one that is entirely consistent with robust enforcement of
state criminal and civil law.
Despite the necessary preemption of inconsistent state laws, every
state and every Federal prosecutor can successfully target online
criminal activity by properly pleading that the defendant was at least
partially involved in the creation of illegal content, or at least the
later development of it. In all such cases, Section 230 immunity does
not apply.
How Section 230 actually works
The importance to millions of Americans of so many topics that
Section 230 touches upon either directly or indirectly--for example,
the responsibility of social media platforms to their users and the
public; the ability of citizens to exercise their First Amendment
rights; the ability of law enforcement to track down criminals; the
protection of the privacy of every user of the internet--means that
almost everyone has an opinion about Section 230 itself. But
notwithstanding that Section 230 has become a household name, a
complete understanding of how the law functions in practice, and what
it actually does, is harder to come by. There are several
misconceptions abroad that merit clarification.
Some mistakenly claim that Section 230 prevents action against
websites that knowingly engage in, solicit, or support illegal
activity. This is simply wrong. But since this claim is often a
principal basis for urging amendment of Section 230, it bears repeating
that Section 230 provides no protection for any website, user, or other
person or business involved even in part in the creation or development
of content that is tortious or criminal.
In the two and a half decades that Section 230 has been on the
books, there have been hundreds of court decisions interpreting and
applying it. It is now firmly established in the case law that Section
230 cannot act as a shield whenever a website is in any way complicit
in the creation or development of illegal content. In the landmark en
banc decision of the Ninth Circuit Court of Appeals in Fair Housing
Council of San Fernando Valley v. Roommate.com,\17\ which has since
been widely cited and applied across the United States, it was held
that not only do websites lose their immunity when they merely
``develop'' content created by others, but participation in others'
content creation can be established by the wholly automated features of
a website that are coded into its architecture.
---------------------------------------------------------------------------
\17\ 521 F.3d 1157, 1168 (9th Cir. 2008).
---------------------------------------------------------------------------
There are many examples of courts faithfully applying the plain
language of Section 230 to hold websites liable for complicity in the
creation or development of illegal third-party content. In its 2016
decision in Federal Trade Comm'n v. Leadclick Media, LLC,\18\ the
Second Circuit Court of Appeals rejected a claim of Section 230
immunity by an Internet marketer even though it did not create the
illegal content at issue, and the content did not appear on its
website. The court noted while this was so, the Internet marketer gave
advice to the content creators. This made it complicit in the
development of the illegal content, and so ineligible for Section 230
immunity.
---------------------------------------------------------------------------
\18\ 838 F.3d 158 (2d Cir. 2016).
---------------------------------------------------------------------------
In FTC v. Accusearch,\19\ the Tenth Circuit Court of Appeals held
that a website's mere posting of content that it had no role whatsoever
in creating--telephone records of private individuals--constituted
``development'' of that information, and so deprived it of Section 230
immunity. Even though the content was wholly created by others, the
website knowingly transformed what had previously been private
information into a publicly available commodity. Such complicity in
illegality was deemed to constitute ``development'' of the illegal
content, as distinguished from its creation.
---------------------------------------------------------------------------
\19\ 570 F.3d1187, 1197 (10th Cir. 2009).
---------------------------------------------------------------------------
Other notable examples of this now well-established feature of
Section 230 are Enigma Software Group v. Bleeping Computer,\20\ in
which a website was denied immunity despite the fact it did not create
the unlawful content at issue, because of an implied agency
relationship with an unpaid volunteer who did create it; and Alvi
Armani Medical, Inc. v. Hennessey,\21\ in which the court deemed a
website to be complicit in content creation because of its alleged
knowledge that postings were being made under false identities.
---------------------------------------------------------------------------
\20\ 194 F.Supp.3d 263 (2016).
\21\ 629 F. Supp. 2d 1302 (S.D. Fla. 2008).
---------------------------------------------------------------------------
In its 2016 decision in Jane Doe v. Backpage.com,\22\ however, the
First Circuit Court of Appeals cast itself as an outlier, rejecting the
holding in Roommate.com and its progeny. Instead, it held that ``claims
that a website facilitates illegal conduct through its posting rules
necessarily treat the website as a publisher or speaker of content
provided by third parties and, thus, are precluded by section
230(c)(1).'' \23\ This holding completely ignored the definition in
subsection (f)(3) of Section 230, which provides that anyone--including
a website--can be an ``information content provider'' if they are
``responsible, in whole or in part, for the creation or development''
of online content. If a website's posting rules facilitate the
development of illegal content, then the website becomes a content
provider in its own right, and should be deprived of Section 230
immunity.
---------------------------------------------------------------------------
\22\ Jane Doe No. 1, et al., v. Backpage.com LLC, et al., No. 15-
1724 (1st Cir. 2016).
\23\ Id. (emphasis added).
---------------------------------------------------------------------------
Despite the fact that the First Circuit was an outlier in this
respect, the notoriety of its decision in the Backpage case has given
rise to the notion that Section 230 routinely operates as a shield
against actual wrongdoing by websites. The opposite is the case. Courts
since 2016 have consistently followed the Roommate precedent, and
increasingly have expanded the circumstances in which they are willing
to find websites complicit in the creation or development of illegal
content provided by their users.
Ironically, the actual facts in the Backpage case were a
Technicolor display of complicity in the development of illegal
content. Backpage knowingly concealed evidence of criminality by
systematically editing its adult ads; it coached its users on how to
post ``clean'' ads for illegal transactions; it deliberately edited ads
in order to facilitate prostitution; it prescribed the language used in
ads for prostitution; and it moderated content on the site, not for the
purpose of removing ads for prostitution, but to camouflage them. It is
difficult to imagine a clearer case of complicity ``in part, for the
creation or development'' of illegal content.
Happily, even within the First Circuit, this mistake has now been
rectified. In the 2018 decision in Doe v. Backpage.com,\24\ a re-
pleading of the original claims by three new Jane Doe plaintiffs, the
court held that allegations that Backpage changed the wording of third-
party advertisements on its site were sufficient to deem it an
information content provider, and thus ineligible for Section 230
immunity. Much heartache could have been avoided had these allegations
concerning Backpage's complicity been sufficiently pleaded in the
original case,\25\ and had the court reached this sensible and clearly
correct decision on the law in the first place.
---------------------------------------------------------------------------
\24\ Doe No. 1 v. Backpage, 2018 WL 1542056 (D. Mass. March 29,
2018).
\25\ Although the plaintiffs disputed this, in the original case
the First Circuit pointedly noted that the record before it expressly
did not allege that Backpage contributed to the development of the sex
trafficking content, even ``in part.'' Instead, the argument that
Backpage was an ``information content provider'' under Section 230 was
``forsworn'' in the district court and on appeal.
---------------------------------------------------------------------------
Another misguided notion is that Section 230 was never meant to
apply to e-commerce. To the contrary, removing the threat to e-commerce
represented by the Prodigy decision was an essential purpose in the
development and enactment of Section 230.
When Section 230 became law in 1996, user-generated content was
already ubiquitous on the internet. The creativity being demonstrated
by websites and users alike made it clear that online shopping was an
enormously consumer-friendly use of the new technology. Features such
as CompuServe's ``electronic mall'' and Prodigy's mail-order stores
were instantly popular. So too were messaging and e-mail, which in
Prodigy's case came with per-message transaction fees. Web businesses
such as CheckFree demonstrated as far back as 1996 that online bill
payment was not only feasible but convenient. Prodigy, America Online,
and the fledgling Microsoft Network included features we know today as
content delivery, each with a different payment system.
Both Rep. Wyden and I had all of these iterations of Internet
commerce in mind when we drafted our legislation. We made this plain
during floor debate.\26\
---------------------------------------------------------------------------
\26\ See 141 Cong. Rec. H8468-72, H8478-79 (August 4, 1995).
---------------------------------------------------------------------------
Yet another misconception about the coverage of Section 230, often
heard, is that it created one rule for online activity and a different
rule for the same activity conducted offline. To the contrary, Section
230 operates to ensure that like activities are always treated alike
under the law.
When Section 230 was written, just as now, each of the commercial
applications flourishing online had an analog in the offline world,
where each had its own attendant legal responsibilities. Newspapers
could be liable for defamation. Banks and brokers could be held
responsible for failing to know their customers. Advertisers were
responsible under the Federal Trade Commission Act and state consumer
laws for ensuring their content was not deceptive and unfair.
Merchandisers could be held liable for neg-ligence and breach of
warranty, and in some cases even subjected to strict liability for
defective products.
In writing Section 230, Rep. Wyden and I, and ultimately the entire
Congress, decided that these legal rules should continue to apply on
the Internet just as in the offline world. Every business, whether
operating through its online facility or through a brick-and-mortar
facility, would continue to be responsible for all of its legal
obligations. What Section 230 added to the general body of law was the
principle that an individual or entity operating a website should not,
in addition to its own legal responsibilities, be required to monitor
all of the content created by third parties and thereby become
derivatively liable for the illegal acts of others. Congress recognized
that to require otherwise would jeopardize the quintessential function
of the internet: permitting millions of people around the world to
communicate simultaneously and instantaneously. Congress wished to
``embrace'' and ``welcome'' this not only for its commercial potential
but also for ``the opportunity for education and political discourse
that it offers for all of us.'' \27\
---------------------------------------------------------------------------
\27\ Id. at H8470.
---------------------------------------------------------------------------
The result is that websites are protected from liability for user-
created content, but only if they are wholly uninvolved in the creation
or development of that content. Today, virtually every brick-and-mortar
business of any kind, from newspapers to retailers to manufacturers to
service providers, has an Internet presence through which it conducts
e-commerce. Whether in the offline world or the internet, the same
legal rules and responsibilities apply across the board to all.
It is worth debunking three other ``creation myths'' about Section
230.
The first is that Section 230 was conceived as a way to protect an
infant industry. According to this narrative, in the early days of the
internet, Congress decided that small startups needed protection. Now
that the Internet has matured, it is argued, the need for such
protection no longer exists; Section 230 is no longer necessary.
As co-author of the legislation, I can verify that this is an
entirely fictitious narrative. Far from wishing to offer protection to
an infant industry, our legislative aim was to recognize the sheer
implausibility of requiring each website to monitor all of the user-
created content that crossed its portal each day. In the 1990s, when
Internet traffic was measured in the tens of millions, this problem was
already apparent. Today, in the second decade of the 21st century, the
enormous growth in the volume of traffic on websites has made the
potential consequences of publisher liability far graver. Section 230
is needed for this purpose now, more than ever.
The second ``creation myth'' is that Section 230 was adopted as a
special favor to the tech industry, which lobbied for it on Capitol
Hill and managed to wheedle it out of Congress by working the system.
The reality is far different. In the mid-1990s, Internet commerce had
very little presence in Washington. When I was moved to draft
legislation to remedy the Prodigy decision, it was based on my reading
news reports of the decision. No company or lobbyist contacted me.
Throughout the process, Rep. Wyden and I heard barely at all from the
leading Internet services of the day. This included both Prodigy and
CompuServe, whose lawsuits inspired the legislation. As a result, our
discussions of the proposed legislation with our colleagues in the
House and Senate were unburdened by importunities from businesses
seeking to gain a regulatory advantage over their competitors.
I willingly concede that this was, therefore, a unique experience
in my lawmaking career. It is also the opposite of what Congress should
expect if it undertakes to amend Section 230, given that today millions
of websites and more millions of Internet users have an identifiable
stake in the outcome.
The final creation myth is that Section 230 was part of a grand
bargain with Senator James Exon (D-NE), in which his Communications
Decency Act aimed at pornography was paired with the Cox-Wyden bill,
the Internet Freedom and Family Empowerment Act, aimed at greenlighting
websites to enforce content moderation policies without fear of
liability. The claim now being made is that the two bills were actually
like legislative epoxy, with one part requiring the other. And since
the Exon legislation was subsequently invalidated as unconstitutional
by the U.S. Supreme Court, so the argument goes, Section 230 should not
be allowed to stand on its own.
In fact, the revisionists contend, the primary congressional
purpose back in 1996 was not to give Internet platforms limited
immunity from liability as Section 230 does. Rather, the most important
part of the imagined ``package'' was Senator Exon's radical idea of
imposing stringent liability on websites for the illegal acts of
others--an idea that Exon himself backed away from before his amendment
was actually passed. Now, a quarter-century after the Supreme Court
threw out the Exon bathwater, the neo-speech regulators are urging us
to throw out the Section 230 baby along with it.
The reality is far different than this revisionist history would
have it. In fact, the Cox-Wyden bill was deliberately crafted as a
rebuke of the Exon approach. When it came to the House floor for
consideration, speaker after speaker rose to speak in support, and at
the same time criticized the Exon approach. Rep. Zoe Lofgren (D-CA),
the mother of 10- and 13-year-old children, shared her concerns with
Internet pornography and noted that she had sponsored legislation
mandating a life sentence for the creators of child pornography. But,
she emphasized, ``Senator Exon's approach is not the right way. . . .
It will not work.'' It was, she said, ``a misunderstanding of the
technology.''
Rep. Bob Goodlatte, a Virginia Republican, emphasized the potential
the Internet offered and the threat to that potential from Exon-style
regulation. ``We have the opportunity for every household in America,
every family in America, soon to be able to have access to places like
the Library of Congress, to have access to other major libraries of the
world, universities, major publishers of information, news sources.
There is no way,'' he said, ``that any of those entities, like Prodigy,
can take the responsibility to edit out information that is going to be
coming in to them from all manner of sources.''
In the end, not a single representative spoke against the bill. The
final roll call on the Cox-Wyden amendment was 420 yeas to 4 nays. It
was a resounding rebuke to the Exon approach in his Communications
Decency Act. The House then proceeded to pass its version of the
Telecommunications Act--with the Cox-Wyden amendment, and without Exon.
When the House and Senate met in conference on the
Telecommunications Act, the House conferees sought to include Cox-Wyden
and strike Exon. But political realities as well as policy details had
to be dealt with. There was the sticky problem of 84 senators having
already voted in favor of the Exon amendment. Once on record with a
vote one way--particularly a highly visible vote on the politically
charged issue of pornography--it would be very difficult for a
politician to explain walking it back. The Senate negotiators, anxious
to protect their colleagues from being accused of taking both sides of
the question, stood firm. They were willing to accept Cox-Wyden, but
Exon would have to be included, too. The House negotiators, all
politicians themselves, understood. This was a Senate-only issue, which
could be easily resolved by including both amendments in the final
product. It was logrolling at its best.
Perhaps part of the enduring confusion about the relationship of
Section 230 to Senator Exon's legislation has arisen from the fact that
when legislative staff prepared the House-Senate conference report on
the final Telecommunications Act, they grouped both Exon's
Communications Decency Act and the Internet Freedom and Family
Empowerment Act into the same legislative title. So the Cox-Wyden
amendment became Section 230 of the Communications Decency Act--the
very piece of legislation it was designed to counter. Ironically, now
that the original CDA has been invalidated, it is Ron's and my
legislative handiwork that forever bears Senator Exon's label.
Measuring the PACT Act and Other Pending Federal Legislation Against
the Goals Section 230 Is Meant to Achieve
When Congress enacted what we know today as Section 230 by near-
unanimous votes in the House and the Senate, there was broad agreement
on several basic principles. Some of these are set forth in the law's
preamble; others are set forth in the operational portion of the
statute. These basic tenets are as follows:
The wide array of interactive educational and informational
services available to individual Americans via the Internet
represents an extraordinary resource worth preserving.
The ideal way to control the flow of information on the
internet, and to screen wanted from unwanted information, is
not for government to regulate that flow, but rather for each
individual user to have the greatest possible control over what
they receive.
The fact that the Internet is not centrally controlled and
regulated, but largely comprised of content created by millions
of individual users, makes it a global forum for a true
diversity of political discourse, unique opportunities for
cultural development, and myriad avenues for intellectual
activity.
The Internet has flourished, to the benefit of all Americans
who rely upon it for a variety of political, educational,
cultural, and entertainment services, with a minimum of
government regulation.
Content moderation aimed at keeping websites free from
obscenity, stalking, harassment, terrorism, criminal activity,
and other objectionable content and behavior should not be
penalized by exposing those websites who undertake it to
increased liability for their efforts.
Twenty-four years later, while the Internet itself has changed in
many ways, these fundamental principles remain sound. The task for 21st
century lawmakers is to determine whether these goals are being
achieved, and to explore ways to address any shortcomings. Remedial
legislation, if it is found warranted, should seek to preserve and
extend the benefits that Congress overwhelmingly agreed can and should
be forthcoming from the internet. Accordingly, any new bill with the
goal of updating Section 230 or related areas of Federal law should be
measured against this template.
In the current Congress, a number of bills have been introduced in
both chambers dealing directly or indirectly with content moderation.
These include S. 3398, the EARN IT Act; S. 1914, the Ending Support for
Internet Censorship Act; H. R. 4027, the Stop the Censorship Act; S.
3983, the Limiting Section 230 Immunity to Good Samaritans Act; and
S.4062, Stopping Big Tech's Censorship Act. In this committee, you are
considering the Platform Accountability and Consumer Transparency Act
(PACT Act), which is aimed at increasing transparency of content
moderation policies and ensuring that knowing participation in criminal
activity is punishable to the full extent of the law. These are worthy
objectives and I commend the committee for prioritizing them.
PACT Act
Considering the PACT Act in light of the original purposes of
Section 230, I offer the following observations.
First, the PACT Act itself embraces the important policy objectives
set out in the original Section 230. It repeats Section 230's intention
to preserve and encourage the continued technological advancement of
the internet, in recognition of the substantial benefits the Internet
provides both to consumers and to the overall economy. The bill also
highlights the fact that people throughout the United States rely on
the Internet for a wide variety of things, including communicating with
friends and loved ones as well as the wider world; gathering
information from others and broadcasting their own creations; and
conducting commercial transactions of endless variety.
Plainly, the purpose of these declarations in the PACT Act is to
set out an overarching objective of ensuring that these benefits aren't
comprised. This is an aspiration I wholeheartedly endorse. It is also a
useful standard against which to measure the operational portions of
the bill.
Finally, the preamble to the PACT Act declares that free expression
is an essential feature of the Internet that should be protected. The
bill recognizes that the Internet is a uniquely successful facilitator
of communications now essential to economic, political, social, and
cultural life in America and around the world. This essential
characteristic of the internet, which arises from its decentralized
architecture that permits millions (indeed billions) of users to
interact in real time, was of great importance to me and to the other
members of Congress in the mid-1990s when we enacted Section 230. In
this respect, the PACT Act and Section 230, at least insofar as their
ultimate aims, are aligned.
The bill is divided into three main parts, dealing with
transparency, liability, and enforcement. I will address each one in
order. Before doing so, I should note several things the bill doesn't
do. In each case, in my judgment, the decision of the PACT Act authors
to avoid going down these paths reflects the better part of wisdom.
Encryption: The bill eschews the approach of the original version
of the EARN IT bill, which had the potential to compromise existing
consumer privacy protections by raising the possibility that encryption
designed to be secure against everyone except the user who holds the
key might expose platforms to new liability. It is a noble legislative
aim to incentivize creation of a technically feasible means of ``lawful
access'' that only the government could exploit, but cybersecurity is a
constant game of cat-and-mouse in which bad actors are constantly
outwitting the latest protections. Despite best efforts, the U.S.
government has been hacked many times, and millions of people have lost
sensitive information as a result, including not only their Social
Security numbers but also detailed private information about their law
enforcement, medical, financial, and employment records containing such
highly protected data as fingerprints and mental health diagnoses, as
well as equally personal information on children and other family
members. The Pentagon, the SEC, HHS, the Executive Office of the
President, and several member departments and agencies within the
intelligence community have been penetrated.
In many cases these successful exploits of U.S. government security
have been accomplished by sophisticated foreign actors with state
sponsorship.
Congress most certainly should be examining how law enforcement
aims can be achieved in tandem with rigorous protection of individual
Americans' privacy. But leaping into that morass with mandates or
penalties that require the creation of ``backdoors,'' before the
technology exists to guarantee that the backdoors will not themselves
become the means of illegal exploitation, is premature.
Political Neutrality: As distinct from S. 1914 and S. 4062, the
PACT Act does not condition Section 230 protections for websites
hosting user-created content on their being ``politically neutral.''
Ensuring that the Internet remains ``a global forum for a true
diversity of political discourse'' requires that government allow a
thousand flowers to bloom--not that a single website has to represent
every conceivable point of view. Section 230 does not require political
neutrality, and was never intended to do so. Were it otherwise, to use
an obvious example, neither the Democratic National Committee nor the
Republican National Committee websites would pass a political
neutrality test. Government-compelled speech is not the way to ensure
diverse viewpoints. Permitting websites to choose their own viewpoints
is.
Websites that choose to be politically neutral, and hold themselves
out as such, can be held to this standard. When an Internet platform
promises its customers--through its advertising, published community
standards, and terms of service--that its content moderation policy is
politically neutral, then that promise can be enforced both by the
government and civil litigants under existing Federal and state laws.
This is far different than a mandate of political neutrality, with the
judgment of what is and is not ``neutral'' placed in the hands of
political appointees in Washington. The PACT Act wisely shuns this
approach.
Subjective Standards: Several commentators have urged grafting onto
Section 230 a requirement, derived from negligence law, upon which
existing protections for content moderation would be conditioned.
Typically taking the form of a ``duty of care'' or a ``reasonableness''
standard, the proposals would effectively make every complaint that a
website has failed to meet the standard into a question of fact. Since
such fact disputes can only be resolved after evidentiary discovery
(depositions of witnesses, written interrogatories, subpoenas of
documents, and so forth), no longer could a website prove itself
eligible for dismissal of a case at an early stage. An essential
feature of Section 230 is its objective standard: was the allegedly
illegal material created or developed--in whole or in part--by the
website? If the complaint adequately alleges this, then the website can
be treated as a publisher and held liable for the material; otherwise
not.
Without an objective standard to determine whether lawsuits can
proceed, a website would constantly be exposed to open-ended, multi-
year litigation over any or all of the user-created content it hosts.
The defining characteristic of the internet--the convergence of many
(frequently millions and occasionally billions) of users on a single
platform--means that a website would have no way to protect itself from
a multiplicity of such lawsuits, short of scaling back or eliminating
user-created content. Currently, civil suits in the Federal system that
proceed beyond a motion to dismiss on the pleadings last an average of
three years through trial; appeals can consume years more. For this
reason, over 90 percent of cases settle without a judge or jury
actually applying the law to the facts in their case. The mere filing
of a lawsuit in such circumstances can create significant settlement
value for a plaintiff. The fact that a typical website could easily
face hundreds or even thousands of such suits illustrates the severity
of the threat to the functioning of the Internet itself.
The PACT Act does not seek to graft subjective negligence-type
concepts such as a duty of care onto the currently objective criteria
in Section 230. Because ensuring that Section 230 can be applied by
courts at the motion to dismiss stage is essential to achieving its
purposes, this is an important conceptual pitfall for any remedial
legislation to avoid.
Monitoring User-Created Content: Essential to the functioning of
the internet, and to reaping the benefits of its characteristic feature
of real-time communication among unlimited numbers of users, is that
websites hosting content do not have to monitor every piece of content.
The sheer volume of communications arising from a planetary base of
potential users makes this an unreasonable requirement. Even if a
website could somehow staff up to meet this near-impossible burden,
doing so would ensure that Internet communications via that platform
could not proceed in real time. Nonetheless, several legislative
proposals would impose potential legal liability on websites that could
only be avoided by constant monitoring of all user-created content.
This is a situation that Section 230 was intended to prevent. The PACT
Act wisely avoids the imposition of a monitoring requirement, and
indeed contains language in section 5 stating that monitoring or
``affirmative fact-seeking'' is not required in connection with
complaints received. (A similar disclaimer should be added to the bill
to clarify that such an obligation does not exist in any case, whether
in connection with a complaint or not.)
Takedown Based on Private Accusations: Several commentators have
recommended that U.S. law be amended to require, following the model of
the Digital Millennium Copyright Act, the mandatory takedown of content
once a website has been notified that it is defamatory or otherwise
violative of law. Such a requirement would empower anyone willing to
allege defamation to require the immediate removal of speech with which
they disagree. The PACT Act avoids this pitfall. Instead, its
requirement of mandatory takedown of illegal content and conduct
applies only when that content or conduct has been determined by a
court to be violative of law. While there are other issues created by
the language in the bill as drafted, the legislative choice not to
create opportunities for the exercise of a ``heckler's veto'' is the
correct one.
Internet Infrastructure Services: Section 230 defines the term
``interactive computer service'' broadly, because it was intended that
the law's protections extend broadly to ensure that content moderation
and free expression would be protected. If Congress decides to use
Section 230 as a vehicle for placing new burdens and liabilities on web
platforms, care should be taken to distinguish between them and the
Internet infrastructure providers that are swept within the broad
definition of ``interactive computer service.'' For example, DNS
registries do not operate content publishing platforms and indeed have
no direct relationships with end users of the internet. As
infrastructure providers, they are very different from social media
platforms and search engines. The PACT Act does not attempt to regulate
Internet infrastructure providers, and indeed the bill includes
language forswearing this with respect to web hosting, domain
registration, content delivery networks, caching, back-end data
storage, and cloud management. This distinction between websites and
Internet infrastructure providers is an important one to make.
Turning now to the PACT Act's three main sections, and taking them
in order, I offer the following comments and suggestions.
Transparency
Transparency--meaning disclosure to consumers, regulators,
stakeholders, and the public generally of how a platform moderates
content--is a sound objective. The PACT Act's prioritization of
transparency is unquestionably constructive and consistent with Section
230 and its ultimate aims.
The mechanisms through which the bill would promote transparency
include statutory standards for each website's content moderation
policy; mandatory complaint systems for each website that include toll-
free call-in services and web-based mechanisms, to be used when
websites fail to meet the content moderation standards; required notice
and hearing, including a right to appeal, for each complaint received;
and mandatory recordkeeping and reporting of content moderation
decisions and disposition of complaints. In addition, the Federal Trade
Commission is given authority to enforce the statutory standards and
the content moderation policies of every website.
While overall these provisions could be made to be workable, as
drafted they will run afoul of the objectives of Section 230 and
threaten the smooth functioning of the Internet and the currently
robust environment for user-created content.
`Potentially policy-violating content': Specifically, section 5 of
the bill includes in its mandates for an ``acceptable use policy'' the
requirement that websites provide due process notices, hearings, and
appeals in response to every complaint that third-party content
``potentially'' violates the website's community standards. There are
three problems with this approach.
First, the website's own standards may or may not be admirable from
a public policy perspective. Given that--so long as the statutory
requirements concerning illegal content and activity are met--websites
are free to adopt whatever content policies they wish, it is reasonable
to assume that some websites will welcome content that, while legal,
the government would not wish to promote. Any government-mandated
complaint system should therefore be focused not on the purely
voluntary and idiosyncratic aspects of each website's content policies,
but rather on illegal content and illegal activity. This would amply
cover not only criminal conduct and content involving sex trafficking,
child sexual abuse material, terrorism, illegal drug dealing, stalking,
and so forth, but also the wide range of Federal and state civil
offenses including defamation and invasion of privacy.
Second, the bill's extension of its due process mandate to cover
not only actual violations of each website's policy, but also potential
violations, introduces a subjective concept that will be easily abused.
Currently, Section 230 permits a court in most cases to judge whether
or not the law applies at an early stage, based on the pleadings. This
ensures that the mere lodging of a complaint does not trigger elaborate
expense for the website--particularly important given the volume of
user-created content often handled by even the smallest websites. By
reducing what must be alleged in a telephone or e-mail complaint to the
mere possibility that content or activity could potentially violate the
website's policy, the PACT Act as written would make it trivially easy
for anyone to trigger the notice-and-hearing requirements contained in
section 5.
Third, the imposition of such a broad notice-and-hearing
requirement, which would apply in almost every case given the lax and
subjective standard for triggering it, will expose websites to
significant expense. (Combined with the high volume of hearings and
appeals the bill's subjective standard will generate, its requirement
that every complaint be initially researched, analyzed, and disposed of
within 14 days will make compliance still more expensive.) Websites
will naturally seek to avoid or at least minimize this greater expense.
If almost every complaint requires a hearing and triggers notice
requirements and guarantees an appeal, then the only way to minimize
the associated expense will be to reduce the grounds for complaints to
be filed. Since every website will have control over the specifics of
its content moderation policy, the incentive will be to minimize the
number of moderation decisions required, through the adoption of less-
robust moderation policies. Alternatively, websites could reduce or
eliminate user-created content. Section 230, on the other hand, is
intended to protect and encourage content moderation, and to facilitate
users' ability to publish their content on the internet. In these ways,
the inclusion of allegedly ``potentially policy-violating content'' as
a trigger for mandatory hearings and appeals is at odds with the stated
goals of Section 230 and the PACT Act itself.
To better align section 5 with the PACT Act's own stated
objectives, therefore, it should be amended to eliminate ``potentially
policy-violating content'' wherever it appears. In addition to
remedying the problems noted, this would also conform section 5 with
the intermediary liability provisions in section 6, which are focused
on illegal content and activity, and not on ``potentially policy-
violating content.''
Data collection and reporting: The specific requirements for data
collection and quarterly public reporting based thereon, as set forth
in section 5 of the bill, include the following:
1. The number of user complaints about specific content
2. The number of employee flags about specific content
3. The number of contractor flags about specific content
4. The number of internal automated flags about specific content
5. The number of government flags about specific content
6. The number of flags about specific content from other service
providers
7. The number of flags from outside personnel employed or
contracted by other service providers
8. The country of each provider of content that is subject to a
complaint or flag
9. The number of times each specific rule within the website's
content policy was violated
10. The number of times the website took one of the following
actions with respect to content:
a) content removal
b) content demonetization
c) content deprioritization
d) appending content with an assessment
e) account suspension
f) account removal
11. The number of appeals of decisions on complaints about specific
content
12. The number of appeals that resulted in restoration of content
previously removed
13. Each mechanism used to enforce the website's content policy,
including:
a) Software and hardware tools
b) General practices
c) Specific actions
d) Proprietary techniques \28\
---------------------------------------------------------------------------
\28\ There is an additional requirement that websites report their
actions with respect to questionable content, categorized by
``coordinated campaign, if applicable.'' See section 5(d)(2)(B)(iv) of
the bill. It is not at all clear what this means.
This ongoing data collection burden would be placed on every
website in America with an average daily number of visitors of more
than 33,333 and $12 million in annual revenue, thereby sweeping in
thousands of small businesses that would have to comply.\29\ As onerous
as the data collection and reporting could be for such websites, the
burden would grow exponentially with the size of the platform. The
largest social media platforms, Facebook, Twitter, and Yahoo, remove
about three billion posts and accounts every 90 days. The number of
``deprioritization'' decisions, given the daily and even moment-to-
moment automated adjustments that would be encompassed within that
rubric, would be far higher. The requirement to maintain detailed
recordkeeping for all of this for every individual piece of content,
which would then become the basis for public reports that would have to
be scrubbed for accuracy before publication, would impose a daunting
logistical and economic tax on all but the smallest websites.
---------------------------------------------------------------------------
\29\ This is a very low threshold. By comparison, the Small
Business Association defines a small business as one with less than $35
million in annual revenue. See 13 CFR Sec. 121.201. The PACT Act's
implicit definition of a ``large'' business would sweep in websites
one-third the size of what the SBA considers to be a small business.
---------------------------------------------------------------------------
The disincentives to do content monitoring at all that would
accompany these costly impositions would pose a genuine threat to the
goals that both Section 230, and ostensibly the PACT Act itself, are
aimed at achieving.
Beyond the sheer burden of compliance with this extensive mandate,
the language in the bill poses interpretive challenges. None of the
terms used in the long list of categories to be tracked is defined.
While ``content demonetization'' has some meaning in common parlance as
it relates to Google, for the 875 million other websites in America
that is likely not the case. The same can be said for ``content
deprioritization.'' Depending upon the website's particular business
model, the term might have no application at all; alternatively, each
website might be left to define the term for themselves, with endless
different variations on the theme. The lack of rigor in drafting this
section of the bill would make compliance, already destined to be
expensive and burdensome, needlessly more so.
Liability
The PACT Act would amend Section 230 to deny the law's protection
to any website that fails to ``remove . . . illegal content or stop
illegal activity'' within 24 hours of ``acquiring . . . knowledge'' of
it.
It is clear what is intended here. Conduct and content that are in
and of themselves illegal should be kept off of all websites subject to
the jurisdiction of the United States. That is an unassailable
objective. It is also perfectly consistent with the congressional
purposes in enacting Section 230 in the first place. Section 230 was
never intended to provide a shield for illegal activity.
Notwithstanding the authors' clear purpose, the actual language in
section 6 of the bill creates needless ambiguity that will frustrate
achievement of that purpose. Happily, sturdier language in the same
section of the bill can be used to clarify some of this unintended
ambiguity.
The first drafting problem inheres in the bill's reliance on
``knowledge'' as the trigger for the 24-hour takedown deadline.
``Knowledge'' is a subjective standard that requires an assessment of
state of mind. ``Notice'' is an objective standard, which if
substituted for ``knowledge'' in this context would eliminate any
subjectivity and at the same time fully achieve the authors' objective.
The bill attempts to undo its own use of the subjective term by
defining ``knowledge'' to mean ``notice.'' This creates needless
interpretive risk. Since section 6 of the bill already contains a
detailed definition of ``notice'' that amply serves the purpose, all
that is needed is to change the proposed amendment to Section 230 to
require that the website ``has notice of the illegal content or illegal
activity, as provided in subparagraph (B).''
The second drafting problem concerns the loose description of what
the notice must contain by way of identifying specific illegal content.
The bill states only that the notice must contain ``information
reasonably sufficient'' to locate the content. Failing to include
specific, clear minimum requirements that will in each case guarantee
that the website will be able to locate the offending material
virtually guarantees that disputes will arise. Clarity in this respect
is particularly important given the very short 24-hour deadline for
compliance. (Indeed, as millions of websites are not staffed 24/7 or on
weekends, that deadline will in many cases be unrealistic.)
The third drafting problem is the definition of ``illegal.'' The
bill defines ``illegal'' content and activity to be that which a court
has ``determined to violate Federal law or state defamation law.''
While tightly circumscribing mandatory takedowns to court-adjudicated
cases is a wise legislative choice, more clarity is required here to
specify what constitutes a court determination. Must it be a final
judgment? Must it await the expiration of appeals? And whichever
definition is adopted, what is the rationale? These are questions the
bill's authors must directly confront and resolve. From the standpoint
of websites that will have to comply with this short-fuse takedown
requirement, clarity is more important than the particular answer
Congress might settle upon.
From the standpoint of policy makers in Congress, however, which
answer you choose is indeed important. Consider that many individuals
hostile to others' speech are litigious. The automatic operation of
this provision of the PACT Act--mandatory takedown after 24 hours'
notice--means that it will be a sure-fire way to suppress speech on the
internet. In the case of speech involving important public policy
issues, by way of example, should a lower court victory be enough? And
what of default judgments, where by definition the arguments on the
other side have not been fully considered? What of the deliberate
falsification of court orders? (The bill contains no sanction for such
activity.) Careful weighing of the tradeoffs here will be necessary to
ensure that the objectives of protecting free expression and
eliminating illegality from the Internet are simultaneously vindicated.
Enforcement
Section 230 was drafted with the intention of protecting the
innocent from being held liable for wrongs committed by others. It was
equally intended to ensure that those who actually commit wrongs will
be subject to prosecution by both civil and criminal law enforcement.
One need not rely on the legislative history or the words of the
authors for this proposition. The language of the statute is plain
enough. If a website, or anyone who provides what the law describes as
interactive computer services, is complicit in the creation of unlawful
content then it may not claim protection under Section 230. The PACT
Act would undo this arrangement. Instead, Section 230 would be waived
entirely whenever the Federal government or a state attorney general is
the litigant. In every such case, websites would lose the protection
offered by Section 230.
The only conceivable justification for depriving every website of
their existing protection under Federal law in this way is that state
attorney generals and Federal prosecutors are presumed always to be
right, and websites in such cases are presumed always to be wrong. If
so, one wonders why a trial would ever be necessary. In my experience
as head of a Federal agency charged with civil law enforcement, the
agency was--in the judgment of the courts--more often right than wrong,
but hardly infallible. A number of Federal departments and agencies in
recent years, including the Department of Justice, have been chastised
by courts for violating ethical norms in the cases they bring and in
the way they have prosecuted them. State attorneys general are all
elected political figures involved in political fundraising that
frequently presents conflicts of interest. A blanket presumption that
the government is always right is too slender a reed on which to rest
an across-the-board statutory repeal of Section 230's essential
provisions.
There is no reason that Federal and state prosecutors cannot
enforce all of their laws without need of such a wholesale waiver of
Section 230. Indeed, Section 230 itself states that ``Nothing in this
section shall be construed to prevent any State from enforcing any
State law that is consistent with this section.'' So unless flat-out
rejection of the very purpose of Section 230 is the objective, the PACT
Act should not follow this course. Rather than the blunderbuss approach
of simply waiving away the entirety of Section 230 for government
litigants, it would be far wiser to more fully accommodate state law
enforcement interests through an express statutory authorization to
state attorneys general to enforce not only state laws consistent with
Section 230, but Federal laws as well. This would multiply the
potential for enforcement actions to keep illegal content off of the
internet.
Such an authorization could be modeled on the existing provision in
28 U.S.C. Sec. 543 empowering the Department of Justice to appoint
participating state Attorneys General as ``Special Attorneys.'' This
authority of the Attorney General to appoint ``Special Attorneys''
dates to 1966. (The statutory authority was most recently amended in
2010.) The internal Department of Justice authority appears in the
United States Attorneys Manual (USAM) at USAM Sec. 3-2-200. The
authority is very broad, and the terms of the appointment are entirely
negotiable. In this way, every state Attorney General who wishes to do
so could exercise the full authority not only of his or her state law,
but also Federal law. As Section 230 has no application to Federal
criminal law, any theoretical arguments about its application to a
given state prosecution will immediately evaporate.
S. 3398, EARN IT
The most recent version of the EARN IT bill was reported from the
Senate Judiciary Committee on July 20. As amended in committee, the
bill would make several changes to Federal law affecting Section 230
and content moderation. The amended bill, like its predecessor,
continues to present several serious issues, including constitutional
infirmities that could create opportunities for child abusers to escape
justice by demanding that the most damning evidence be excluded from
their trials.
The bill would mandate the establishment of Federal standards,
referred to in the bill as ``best practices,'' that would cover, among
other things, the following specific ways that websites and Internet
infrastructure providers should be involved in content moderation.
While the bill's focus is content related to child sexual exploitation,
the ``best practices'' would necessarily extend to all content prior to
its screening and identification as child sexual exploitation material.
The Federal standards to be promulgated would include requirements for
websites and Internet infrastructure providers to:
1. Preserve on their servers specified user-created content
2. Take down specified user-created content
3. Report to law enforcement and others specified user-created
content
4. Record and preserve location data for users
5. Record and preserve other personal identifiable information
concerning users
6. Develop and maintain an online service for accepting reports
from the public concerning specified user-created content
7. Develop and maintain an internal system for sorting,
prioritizing, and allocating resources to complaints and
reports received through the online public reporting system
8. Implement a ``standard rating and categorization system'' to
identify specified types of user-created content
9. Train content moderators according to the Federal standards to
be promulgated
10. Provide certain specified levels of support to content
moderators devoted to searching for online child sexual
exploitation material
11. Produce reports to the government covering:
a) the entity's policies and procedures for ``identifying,
categorizing, and reporting'' online child sexual
exploitation
b) the entity's efforts ``to prevent and disrupt'' online child
sexual exploitation
12. Coordinate with ``voluntary initiatives'' related to
identifying, categorizing, and reporting specified user-created
content
13. Implement ``age rating'' and ``age gating'' systems covering all
online content
14. Develop ``parental control products'' to limit the types of
websites, social media platforms, and Internet content that can
be accessed
15. Amend contracts with third parties, contractors, and affiliates
to require their compliance with the Federal standards
16. Develop internal operational practices operational practices to
``ensure'' that third parties, contractors, and affiliates
comply with the Federal standards
This is an elaborate list of both wide-ranging and granular
requirements. Yet despite its breadth and granularity, the broad
discretion to elaborate upon these themes--which is entirely given over
to an ad hoc commission created by the bill--would authorize the
promulgation of different or additional requirements that neither
Congress nor the regulated community can predict. The specifics of such
requirements as the mandatory takedown of user-created content are of
enormous importance; yet they are nowhere defined in the bill, and the
process for determining them would be wholly within the control of an
unaccountable group of political appointees.
The several instances of requiring ``searching for'' specified
user-created content, the requirement to store and preserve it, and the
requirement to undertaking affirmative efforts to ``prevent and
disrupt'' users' activity, together amount to a wide-ranging duty to
monitor all incoming user-created content. It would otherwise not be
possible to find what the websites are instructed look for; necessarily
the entire haystack must be searched to find the needle. As protecting
websites from having to monitor all user-created content is a
fundamental purpose of Section 230, the EARN IT bill fails in this
essential respect.
One would hope that, given the deeply intrusive nature of the EARN
IT bill's proposed regulation of the businesses of millions of U.S.-
based websites, as well as the extension of that regulation beyond
websites and consumer-facing Internet platforms to a wide variety of
Internet infrastructure providers, the Congress would be more
solicitous of information concerning how its intended new standards
would actually operate in the real world. While charging the commission
to consider issues of cost and feasibility, there is no check on what
the commission can actually prescribe.
Worse, there is no requirement for public input. Ordinarily, when
Federal agencies promulgate rules, they are first subjected to public
notice and opportunity to comment under the Administrative Procedure
Act. When commissions are created to advise the executive branch, they
are typically subjected to the requirements of the Federal Advisory
Committee Act, which similarly ensures public transparency and input.
But the EARN IT bill freezes out the public from any right of
participation in the process of developing the new Federal standards.
Instead, a commission comprised of politically-appointed individuals
will have free rein to determine what Federal ``best practices'' are,
without need of complying with either the APA or FACA. Among other
things, this makes it far more likely that whatever standards are
promulgated will be uninformed by considerations of how they will, or
will not, function in practice.
Were I still a member of Congress, I would insist that, before this
legislation proceeds further, it be amended to require the standard
public notice and input that is expected for all Federal rulemakings.
Beyond the direct impact on websites from the significant
compliance burdens that would attend compliance with these elaborate
new Federal standards, the consequences for every American who uses the
Internet would be more severe. Whereas today we take for granted the
fact that our posts and communications via the Internet will be
communicated instantaneously, compliance with the new requirements will
mean that many user posts will have to be held in suspense, pending
review by the website's legal team. Moreover, any user post that create
risks to the platform is not likely to survive scrutiny, so that some
messages will never be communicated at all. These unintended
consequences will mark an unwelcome curtailing of the ease and speed
with which Americans share their news and views online today.
Other aspects of the EARN It bill specifically touching upon
Section 230 raise different issues.
The amended EARN IT Act carves out a wholesale exception to the law
that extends to any claim made in a civil suit under any state law,
provided it can be related to child sexual abuse material. The broad
scope of the exception--it waives Section 230 state preemption
completely--will make it an attractive exploitative opportunity for
artful pleading. At a minimum, tightening up the language describing
which claims are covered by the exception is required. The language in
the PACT Act authorizing enforcement of Federal civil laws by state
attorneys general is far preferable in this respect. It requires that
the underlying claim must also allege a violation of Federal law.
An even more serious problem with this across-the-board waiver of
Section 230 for all suits based on state laws is that the statutes of
several states lack an actual knowledge standard. Instead, they
predicate liability on recklessness. As a result, every website would
be exposed to new lawsuits alleging that it was reckless in failing to
actively monitor all user-created content. It is not difficult to
imagine that such lawsuits could be successful. This would effectively
impose a nationwide requirement of a duty to monitor--a result that
Congress should wish to avoid, and that Section 230 was intended to
prevent..
Since not only the new Federal standards but also the state-law
litigation waived in by the EARN IT bill will strongly encourage
monitoring and reporting, there will be new risks of constitutional
challenges to criminal prosecutions using evidence reported in this
way. Whereas under current law, companies are required only to report
known instances of child sexual abuse material, EARN IT constitutes
government inducement to actively search for it, and then turn it over
for use by the government in prosecutions. This raises the prospect the
what are now private searches would be deemed state action, subject to
Fourth Amendment scrutiny.
With the exception of the 10th Circuit Court of Appeals (in an
opinion written by then-Judge Neil Gorsuch),\30\ most courts have held
that the mandatory reporting arrangement under current law does not
amount to state action, because the actual search that precedes the
discovery of the evidence is done voluntarily.\31\ But under applicable
Supreme Court precedent, private searches are subject to the Fourth
Amendment not only when the government requires a search, but when it
merely encourages searches. And under the Exclusionary Rule, evidence
collected in violation of the Fourth Amendment is generally
inadmissible in court.
---------------------------------------------------------------------------
\30\ United States v. Ackerman, 831 F.3d 1292, 1302 (10th Cir.
2016). see also, e.g., United States v. Coyne, 387 F.Supp.3d.
\31\ See, e.g. United States v. Coyne, 387 F.Supp.3d.387 (2018).
---------------------------------------------------------------------------
The risk posed by the EARN IT bill, therefore, is that evidence
otherwise available to convict child abusers could now be
suppressed.\32\
---------------------------------------------------------------------------
\32\ See Chris Marchese, The EARN IT Act's Collision Course With
The Fourth Amendment (2020), https://netchoice.org/wp-content/uploads/
2020/06/EARN-It-4A-Report-FINAL.pdf.
---------------------------------------------------------------------------
A further issue is that the amended EARN IT bill still threatens
the privacy protections that websites can extend to their users. While
the original version of the EARN IT bill posed a more direct threat to
encryption, the amended version continues to give broad authority to
its ad hoc commission to promulgate Federal standards that would give
the government a ``back door''--for example, by requiring websites to
scan all data before and after encryption \33\ or specifying that
device manufacturers create custom operating systems allowing
government access. (This is not idle speculation: the FBI attempted to
convince Apple to do this four years ago.)
---------------------------------------------------------------------------
\33\ As one observer has noted, the popular euphemism for this--
``client-side scanning''--is what we would otherwise call ``reading
your messages on your device.'' Carl Szabo, ``The EARN IT Act threatens
encryption,'' Orange County Register (July 14, 2020).
---------------------------------------------------------------------------
Finally, beyond these significant problems, the EARN IT bill's
carveout for child sexual abuse material presents the same overall
conceptual issue that was present during consideration of FOSTA/SESTA.
The sexual exploitation of minors is a serious crime punishable under
both Federal and state law. But it is one of approximately 4,000
Federal crimes and thousands more state law crimes that include
terrorism, extortion, mass murder, airline hijacking, rape, hate
crimes, hostage taking, sexual battery, torture, and treason. Any one
of these crimes can be facilitated using the internet. As with the
telephone and the telegraph before it, the Internet is frequently a
tool of criminals. Section 230, which is designed to apply a uniform
Federal standard in all civil and criminal cases brought in either
state or Federal forums, is wholly consistent with the prosecution of
criminal and civil claims based on the entire range of illegal activity
of which humankind is capable.
It is difficult to argue that, as horrible as the promotion of
child pornography is, it is categorically worse than mass murder,
terrorism, and a long list of other equally egregious crimes. Nor are
these other crimes any less worthy of congressional attention. As
Chairman of the House Committee on Homeland Security, I saw firsthand
how terrorists use the Internet to direct violent extremist acts.
Neither in America nor anywhere in the world should terrorists find a
``safe space'' to operate and disseminate their murderous propaganda of
mass destruction. When violent extremists further their plots and grow
their ranks by use of the internet, it stands to reason that a nation
of laws would not wish to permit laws enacted for another purpose to be
used as a shield for such acts. Likewise, when criminal gangs kidnap
innocent tourists for exorbitant ransom, using threats of torture and
murder, no law should provide them any form of immunity. When assassins
target our president, lawmakers, or Supreme Court, no one would want to
grant the murderers a legal advantage because they happened to use the
Internet in the commission of their crimes.
Yet the EARN IT bill would treat these problems categorically
differently for legal purposes, providing one set of rules for child
sexual abuse material and another, presumably more lenient, set of
rules for terrorism.
This represents a fundamental misunderstanding of how Section 230
is intended to operate. It was designed to protect the innocent from
being held liable for wrongs committed entirely by others--a principle
that should not be waived in any circumstances. It was equally intended
to ensure that those who actually commit wrongs will be subject to
prosecution by both civil and criminal law enforcement. One need not
rely on the legislative history or the words of the authors for this
proposition. The language of the statute is plain enough. If a website,
or anyone who provides what the law describes as interactive computer
services, is complicit in the creation of unlawful content then it may
not claim protection under Section 230.
Section 230, as written and as interpreted by the courts, is
thoroughly consistent with the aggressive prosecution of child sexual
exploitation. Equally importantly, it is thoroughly consistent with the
aggressive prosecution of all other crimes. It makes little sense to
countenance an interpretation of Section 230 that communicates to
judges looking at prior decisional law that henceforth, a less
stringent rule will be applied in all but the narrow categories carved
out of Section 230 by Congress. Each carveout for differential
treatment will create significant new legal ambiguities and
inexplicable horizontal disparities in both Federal and civil
litigation. Judges faced with a new Section 230 standard for sex
trafficking and child sex abuse cases will be hard pressed not to infer
that cases involving other crimes must be decided using a different
rule.
It is notable that the most of the Nation's attorneys general have
written to Congress endorsing a different approach--one that will
encompass not only child sexual abuse but all criminal enforcement
actions. Such an approach would ensure that courts do not decide to
make some Internet crimes easier, and some crimes harder, to prosecute.
While it would be a mistake to do this by scrapping the uniform Federal
policy with respect to liability for Internet platforms, it is
unquestionably correct that uniformity in the application of the
Federal policy to all crimes is necessary to prevent unintended
consequences such as the creation of loopholes that benefit criminals.
Conclusion
I applaud the efforts of Senators on this subcommittee and on the
full committee to undertake a thoughtful and dispassionate analysis of
the several competing interests involved in keeping the Internet free
from illegal content and conduct, while at the same time promoting and
protecting a vibrant Internet ecosystem with the maximum level of free
expression. As the co-author of Section 230, which has proven to be a
foundational legal underpinning for the Internet as it has developed
over the last quarter century, I am proud of the role that this law has
played in empowering the millions of content creators on the internet,
and for the protections it has effectively provided for the freedom of
speech of millions of people.
Our reconsideration of the scope of Section 230's protections comes
at a time in world history when digital authoritarianism is spreading
rapidly around the globe. As Freedom House has noted in its most recent
annual report on the state of global Internet freedom entitled Freedom
on the Net, ``repressive regimes, elected incumbents with authoritarian
ambitions, and unscrupulous partisan operatives have exploited the
unregulated spaces of social media platforms, converting them into
instruments for political distortion and societal control.'' \34\ They
note that while social media in other nations have at times served as a
level playing field for civic discussion, they now more often expose
citizens to unprecedented invasions of their fundamental freedoms, as
governments deploy advanced tools to identify and monitor users on a
vast scale. This abuse of social media is occurring not just in well-
known cases such as the People's Republic of China, Russia, Iran, and
Saudi Arabia, but also in 38 of the 65 countries covered in their
latest report.
---------------------------------------------------------------------------
\34\ Freedom on the Net 2019: The Crisis of Social Media, available
at https://freedomhouse.org/report/freedom-net/2019/crisis-social-media
---------------------------------------------------------------------------
America's approach to the regulation of social media, and of speech
on the Internet more generally, has to date followed a very different
model, abjuring government control in favor of private ordering. This
has led some critics to argue that private control of the vast amounts
of information generated by users of the Internet represents a threat
to liberty and privacy equal to or greater than would be the result of
government control. But two factors militate against this conclusion.
Importantly, the private websites and platforms with access to user
data are many, and compete with one another. And they lack the powers
of a sovereign to aggregate all available data and then to regulate the
citizenry through its exploitation. In the hands of government, social
media surveillance tools employing artificial intelligence can easily
become powerful weapons with which to silence undesirable expression
and exert social control.
Before taking even the first baby steps away from the policy
Congress and the president endorsed in Section 230 ``to preserve the
vibrant and competitive free market that presently exists for the
Internet and other interactive computer services, unfettered by Federal
or State regulation,'' legislators should be fully aware of where this
road could lead.
The landscape of the Internet continues to change rapidly, and
therefore demands continued vigilant oversight and critical scrutiny by
lawmakers. Section 230 is the creation of Congress, and subject to its
plenary authority to make and revise laws. It is not written in stone
and far from sacrosanct. But it has also provided us with the benefit
of a quarter century of practical experience, through continually
changing and often challenging circumstances. In the main, it has
performed well. To the extent that courts applying it have sometimes
given us unwanted results, we can take comfort in the fact that as of
2020 the interpretive kinks that in the past have sometimes let wrongs
go without remedy have been for the most part worked out.
Were I still in Congress, though I would be tempted to embellish my
original work (like the artist who continues to add a daub here and a
brushstroke there, with the result that the painting is never
finished), in the current environment I would hesitate to do so. My far
greater concern would be the risk, which I have so often seen
materialize in the completion of legislation with which I have been
involved, that the process of moving the bill through numerous
committees, markups, and perhaps an ultimate conference between House
and Senate would ultimately run away with my best intentions.
Unlike the placid policymaking environment in which Section 230 was
conceived and midwifed into law in 1995-96, today the cacophony that is
the debate over social media, content moderation, free speech, and
criminality on the Internet guarantees not only near-irreconcilable
conflicts but also legislative attempts to somehow square the circle.
Such deep compromises ranging from the smallest details to high-level
issues, which will be necessary if a Republican Senate and Democratic
House are to reach any agreement on a bill that achieves their very
disparate aims, will likely produce legislation far different from the
careful balancing of competing interests that this committee's
thoughtful and dispassionate analysis is admittedly capable of
producing in the first instance.
In my judgment, the chance that in the end the most important
benefits of Section 230 could be undermined, or lost entirely, is a
gamble with the future of the Internet not worth taking. Recognizing
that it is your own judgments on these questions that matter, and that
those judgments await your completion of your ongoing analysis of the
many issues involved, I stand ready to assist you in any way that I
can.
Senator Thune. Thank you, Mr. Cox.
Next up is Mr. Jeff Kosseff.
Mr. Kosseff, please proceed.
STATEMENT OF JEFF KOSSEFF,
ASSISTANT PROFESSOR, CYBER SCIENCE DEPARTMENT,
UNITED STATES NAVAL ACADEMY
Mr. Kosseff. Chairman Thune, Ranking Member Schatz, and
members of the Subcommittee, thank you for providing me with
the opportunity to testify about the history and purpose of
Section 230.
I'm an assistant professor in the Cyber Science Department
in the U.S. Naval Academy. My testimony today reflects only my
personal views and does not represent the Naval Academy,
Department of Navy, Department of Defense, or any other party.
This hearing is of the utmost importance, as Section 230 is
responsible, more than any other law, for the open Internet
that Americans know, love, and hate. I am not here today to
advocate for or against any particular legislation. Rather, my
goal is to help expand the public understanding of Section 230.
As explained in detail in my written testimony, under the
First Amendment and common law, distributors cannot be held
liable for content created by others unless the distributors
knew, or had reason to know, of the illegal content. A New York
trial judge in 1995 ruled that Prodigy did not receive this
distributor protection and, instead, was deemed a publisher
that's liable, regardless of its state of mind. The judge's
reasoning was that Prodigy has implemented detail user conduct
rules and employed content moderating.
Members of Congress passed Section 230 in 1996 in an effort
to override this decision and encourage platforms to moderate.
Some critics argue that platforms have not adequately moderated
harmful content. Other critics argue that some existing
moderation practices result in blocking certain political
viewpoints. And both criticisms have driven a number of
proposals to change Section 230.
Today, I hope to set forth some principles to guide your
evaluations of Section 230s future:
First, not all problems on the Internet are Section 230
problems. For instance, the First Amendment, and not Section
230, protects hate speech. Additionally, many defamation claims
that courts dismiss on Section 230 grounds would also, if fully
litigated, not survive common law and First Amendment
protections.
Second, we do not know with certainty how platforms would
react to a repeal or a significant contraction of Section 230,
because the modern Internet has always existed with Section 230
in place. One possibility is, the platforms might avoid
moderation, fearing that, once they encounter potentially
actionable content, they would become liable for it. There's
also the chance that there would be fewer venues for user-
generated content.
Third, Section 230 is designed to encourage, and not
discourage, moderation of user content. In the debate over
neutrality of platforms, I see a few different questions:
First, does Section 230 currently require neutrality? As I
explained in my written testimony, the answer to this question
is no.
Second, should Section 230 require neutrality? The answer
to this question is up to you, as Congress is free to amend
Section 230 as it sees fit; of course, within the confines of
the First Amendment.
If Section 230 were to attempt to impose a neutrality
requirement, I would ask what such a requirement would look
like, and how it would be implemented. Of course, Congress can,
and should, determine whether the market-based system under
Section 230 continues to meet users' expectations in 2020.
Fourth, the Section 230 debate needs far more transparency.
Last October, I suggested the creation of a congressionally
chartered commission to gather facts and recommend a path
forward. The Cyberspace Solarium Commission provides an
excellent model for this.
I commend the Chairman and Ranking Member for the
thoughtful solutions that you've proposed in the PACT Act. The
legislation addresses the need for more transparency and
content moderation policies, and begins the process of
identifying the most tailored and reasonable rules for
providing that transparency. The bill also provides people with
a mechanism to take down material that has been adjudicated to
be defamatory or illegal under Federal criminal or civil law.
We must ensure that a takedown provision is not abused, for
example, via the falsification of court orders.
As I routinely remind technology companies, Section 230 is
not set in stone and can be repealed or significantly amended
as easily as it was passed. Congress may determine that it is
in the public interest to curtail some or all of Section 230s
protections. I urge you to make any such decisions with great
care.
It is difficult to imagine how some of the largest
companies in the United States could have emerged, at least in
their current forms, without Section 230. The challenge for all
of us is to determine how we want the Internet to look over the
next 25 years, and what it takes to get it.
I look forward to taking your questions.
[The prepared statement of Mr. Kosseff follows:]
Prepared Statement of Jeff Kosseff, Assistant Professor, Cyber Science
Department, United States Naval Academy
Chairman Thune, Ranking Member Schatz, and Members of the
Subcommittee, thank you for providing me with the opportunity to
testify about the history and purpose of Section 230 of the
Communications Decency Act of 1996.
I am an assistant professor in the Cyber Science Department of the
United States Naval Academy. My testimony today reflects only my
personal views, and does not represent the Naval Academy, Department of
Navy, Department of Defense, or any other party.
It is difficult to overstate the importance of the subject of this
hearing. Last year, I published a history of Section 230, titled The
Twenty-Six Words That Created the Internet. The Internet's protocols
and technology were developed long before 1996. But Section 230 is
responsible, more than any other law, for the open Internet that
Americans know, love, and hate. By shielding online platforms from
liability for a great deal of third-party content, Section 230 has
paved the way for Yelp, Wikipedia, Facebook, Twitter, YouTube, and so
many other online services. These have primarily based their business
models on content created by individuals rather than corporations.
Section 230 also has protected a wide range of companies of all sizes
that operate websites with user comments.
When I began writing a book about Section 230 in 2016, few people
outside of technology law and policy circles knew much about what the
law does and why Congress passed it in 1996. Much has changed in those
four years, as large platforms are under unprecedented scrutiny for
their handling of user-generated content. Suddenly, Section 230 has
moved from obscure legal discussions to the headlines of major media
organizations. Many are calling for you to repeal or amend Section 230.
Indeed, there are many legislative proposals, including a thoughtful
one from the Chairman and Ranking Member of this subcommittee.
I am not here today to advocate for or against any particular
legislation. Rather, my goal is to help expand the public understanding
of Section 230, first by providing an overview of its history and
purpose, and then by suggesting principles that could guide Congress as
it considers Section 230s future.
I. The History of Section 230
To understand why we have Section 230 and what it does, we need to
look at how platform liability worked before it was passed. This
requires an examination of the liability standards for bookstores and
other distributors of content produced by third parties.
The foundations for distributor liability standards come from Smith
v. California,\1\ a 1959 Supreme Court opinion. In that case, the Court
reversed the conviction of a Los Angeles bookstore owner whose store
sold an obscene book. The ordinance under which he was convicted
imposed criminal liability on bookstore operators regardless of their
scienter or state of mind; in other words, the ordinance was one of
strict liability on any distributor of obscene content, regardless of
their intention or even awareness. Writing for the majority, Justice
Brennan recognized that obscenity is not protected by the First
Amendment, but he concluded that imposing strict liability on
booksellers nonetheless did violate the First Amendment because such a
rule would chill non-obscene speech.
---------------------------------------------------------------------------
\1\ Smith v. California, 361 U.S. 147 (1959).
---------------------------------------------------------------------------
``By dispensing with any requirement of knowledge of the contents
of the book on the part of the seller, the ordinance tends to impose a
severe limitation on the public's access to constitutionally protected
matter,'' Justice Brennan wrote. ``For if the bookseller is criminally
liable without knowledge of the contents, and the ordinance fulfills
its purpose, he will tend to restrict the books he sells to those he
has inspected; and thus the State will have imposed a restriction upon
the distribution of constitutionally protected as well as obscene
literature.'' \2\
---------------------------------------------------------------------------
\2\ Id. at 153.
---------------------------------------------------------------------------
The Court in Smith refrained from articulating the precise mental
state necessary to impose liability on distributors of third-party
content, only saying that strict liability is unacceptable.\3\ The
Supreme Court would provide a bit more guidance. For instance, in 1968,
the Court upheld a New York law that imposed criminal liability on a
newsstand that sold pornographic magazines to minors.\4\ The statute
applied to stores that have ``general knowledge of, or reason to know,
or a belief or ground for belief which warrants further inspection or
inquiry'' of both the character and content of material that is
``reasonably susceptible of examination by the defendant'' as well as
the minor's age.\5\ Writing for the majority, Brennan concluded that
this level of awareness satisfies the concerns that he articulated in
Smith, though he again refrained from setting a precise minimum
standard for all distributor cases.\6\
---------------------------------------------------------------------------
\3\ Id. at 154-55.
\4\ Ginsberg v. New York, 390 U.S. 629 (1968).
\5\ Id. at 646.
\6\ Id. at 644-45.
---------------------------------------------------------------------------
Following Smith v. California--but prior to the passage of Section
230--lower courts generally adopted a rule, rooted in the common law
and the First Amendment, that distributors cannot be liable for content
created by others unless the distributors knew or had reason to know of
the illegal content. This rule applies not only to criminal obscenity
cases, but also to civil claims such as defamation.\7\
---------------------------------------------------------------------------
\7\ See, e.g., Dworkin v. Hustler Magazine, 611 F. Supp. 781 (D.
Wyo. 1985); Osmond v. Ewap, 153 Cal. App. 3d 842 (Cal. Ct. App. 1984).
---------------------------------------------------------------------------
This common law rule was first applied to an online service in
1991, in a defamation action against CompuServe, one of the earliest
national online dial-up services. The suit arose from statements in an
online newsletter that CompuServe distributed. The district court
dismissed the lawsuit, concluding that CompuServe was ``in essence an
electronic, for-profit library that carries a vast number of
publications and collects usage and membership fees from its
subscribers in return for access to the publications.'' \8\ In other
words, CompuServe was a distributor, and therefore deserved the same
liability standards to which newsstands were held.\9\ Because the
plaintiff had not demonstrated that it knew or had reason to know of
the alleged libel in the newsletter, the court dismissed the case.
---------------------------------------------------------------------------
\8\ Cubby v. CompuServe, 776 F. Supp. 135, 140 (S.D.N.Y. 1991).
\9\ Id. Crucially, the court acknowledged that even a distributor
such as CompuServe could have some control over the content that it
distributed. See id. (``While CompuServe may decline to carry a given
publication altogether, in reality, once it does decide to carry a
publication, it will have little or no editorial control over that
publication's contents. This is especially so when CompuServe carries
the publication as part of a forum that is managed by a company
unrelated to CompuServe.'').
---------------------------------------------------------------------------
CompuServe's main competitor at the time was Prodigy, which sought
to distinguish itself from CompuServe by offering more family-friendly
services. Prodigy employed contract moderators and implemented detailed
user conduct rules. When Prodigy was sued due to comments made on a
Prodigy financial bulletin board, the company attempted to claim the
same distributor liability standard to which CompuServe was held. In
May 1995, a New York state trial court judge rejected Prodigy's
attempt, finding that Prodigy is not a distributor, but rather a
publisher that is liable regardless of whether it knew or had reason to
know of the allegedly defamatory content. Even though, by 1995, Prodigy
had loosened its user content policies, the Court focused on the fact
that Prodigy had at one point exercised substantial control over user
content. ``It is Prodigy's own policies, technology and staffing
decisions which have altered the scenario and mandated the finding that
it is a publisher,'' the judge wrote. ``Prodigy's conscious choice, to
gain the benefits of editorial control, has opened it up to a greater
liability than CompuServe and other computer networks that make no such
choice.'' \10\
---------------------------------------------------------------------------
\10\ Stratton Oakmont v. Prodigy Service Co., 23 Media L. Rep. 1794
(N.Y. Sup. Ct. May 24, 1995).
---------------------------------------------------------------------------
The Stratton Oakmont v. Prodigy case received significant media
attention. Although it did not create binding precedent, it strongly
suggested that online services could reduce their exposure to liability
by taking a hands-off approach to user content. If, like Prodigy, a
platform exercised significant control over user content, a court might
conclude that it does not receive the same ``distributor'' liability
standards as a bookstore or newsstand. I believe the ruling was flawed
because it ignored the fact that even the more hands-off CompuServe
could choose not to carry a publication in its electronic version of a
newsstand. And even if a platform received the liability standard of a
``distributor,'' it still could face liability if it knew of, or had
reason to know of, illegal content, creating another disincentive to
moderation.
When Reps. Chris Cox and Ron Wyden learned about the Prodigy case,
they agreed that it made little sense. Why subject an online service to
more liability simply because it took steps to moderate objectionable
content? This disincentive was particularly concerning as schools and
homes increasingly connected computers to the Internet. If the legal
system discouraged online services from moderation, the result could be
the exposure of children to pornography and other objectionable
material. A bill in the Senate, the Communications Decency Act of 1995,
sought to address this problem by imposing criminal liability for the
transmission of indecent content. The Senate attached this decency
proposal to its massive overhaul of U.S. telecommunications law.
Cox and Wyden believed that the online services--which are
accountable to their users--are better positioned than the government
to set user content policies. They saw the potential for the Internet
to be an engine for job growth. They did not want to stifle this
burgeoning new technology with regulation and litigation. Nor did they
want to impose a duty of pre-screening user content before it was
posted.
On June 30, 1995, Cox and Wyden introduced the Internet Freedom and
Family Empowerment Act, most of which would later become Section 230.
To address the prospect of government regulation, the bill initially
stated that the Federal Communications Commission does not have
authority ``with respect to economic or content regulation of the
Internet or other interactive computer services.'' \11\
---------------------------------------------------------------------------
\11\ This provision would not remain in Section 230 as signed into
law.
---------------------------------------------------------------------------
The centerpiece of the bill, however, focused on the liability of
online platforms for user content, and the need to eliminate any
disincentive to moderation. The provision that contains what I believe
are the 26 words that created the Internet states: ``No provider or
user of an interactive computer service shall be treated as the
publisher or speaker of any information provided by another information
content provider.'' \12\ The bill also prevents interactive computer
service providers and users from being liable for ``any action
voluntarily taken in good faith to restrict access to or availability
of material that the provider or user considers to be obscene, lewd,
lascivious, filthy, excessively violent, harassing, or otherwise
objectionable, whether or not such material is constitutionally
protected'' \13\ or providing the technical means to restrict
access.\14\
---------------------------------------------------------------------------
\12\ 47 U.S.C. Sec. 230(c)(1). As initially introduced, this
provision actually contained 25 words because it stated: ``No provider
or user of interactive computer services shall be treated as the
publisher or speaker of any information provided by an information
content provider.''
\13\ 47 U.S.C. Sec. 230(c)(2)(A).
\14\ 47 U.S.C. Sec. 230(c)(2)(B).
---------------------------------------------------------------------------
Cox and Wyden included exceptions for the enforcement of Federal
criminal law,\15\ intellectual property law,\16\ and Federal and state
electronic communications privacy laws.\17\ The bill was partly based
on a theory of user empowerment: the belief that users, with tools
provided by their platforms, should determine what content should be
available to them and their children.
---------------------------------------------------------------------------
\15\ 47 U.S.C. Sec. 230(e)(1).
\16\ 47 U.S.C. Sec. 230(e)(2).
\17\ 47 U.S.C. Sec. 230(e)(4).
---------------------------------------------------------------------------
To clarify their intentions, Cox and Wyden included findings at the
start of their bill. Among their findings: ``These services offer users
a great degree of control over the information that they receive, as
well as the potential for even greater control in the future as
technology develops.'' \18\ Cox and Wyden also wrote that the
``Internet and other interactive computer services offer a forum for a
true diversity of political discourse, unique opportunities for
cultural development, and myriad avenues for intellectual activity.''
\19\ The Internet has ``flourished, to the benefit of all Americans''
they wrote, ``with a minimum of government regulation.'' \20\
---------------------------------------------------------------------------
\18\ 47 U.S.C. Sec. 230(a)(2).
\19\ 47 U.S.C. Sec. 230(a)(3).
\20\ 47 U.S.C. Sec. 230(a)(4).
---------------------------------------------------------------------------
They also included statements of policy, including ``to promote the
continued development of the Internet and other interactive computer
services and other interactive media'' \21\ and ``to encourage the
development of technologies which maximize user control over what
information is received by individuals, families, and schools who use
the Internet and other interactive computer services.'' \22\ They also
wrote that it is U.S. policy ``to preserve the vibrant and competitive
free market that presently exists for the Internet and other
interactive computer services, unfettered by Federal or State
regulation.'' \23\
---------------------------------------------------------------------------
\21\ 47 U.S.C. Sec. 230(b)(1).
\22\ 47 U.S.C. Sec. 230(b)(3).
\23\ 47 U.S.C. Sec. 230(b)(2).
---------------------------------------------------------------------------
On Aug. 4, 1995, the House debated whether to add the Cox-Wyden
proposal to its version of what would become the 1996
telecommunications overhaul. The House members almost uniformly
welcomed the proposal as an alternative to the Senate's indecency
proposal, which many viewed as unconstitutional. ``Really it is like
saying that the mailman is going to be liable when he delivers a plain
brown envelope for what is inside it,'' Rep. Zoe Lofgren said of the
Senate proposal. ``It will not work.'' \24\
---------------------------------------------------------------------------
\24\ 141 Cong. Rec. H8471 (1995).
---------------------------------------------------------------------------
Rep. Robert Goodlatte spoke of the need to fix the perverse
incentive created by the Prodigy opinion. ``The Cox-Wyden amendment
removes the liability of providers such as Prodigy who currently make a
good faith effort to edit the smut from their systems,'' Goodlatte
said. ``It also encourages the online services industry to develop new
technology, such as blocking software, to empower parents to monitor
and control the information their kids can access.'' \25\
---------------------------------------------------------------------------
\25\ 141 Cong. Rec. H8471-72 (1995).
---------------------------------------------------------------------------
Cox spoke about the need to avoid Federal regulation of the
Internet. The bill, he said, ``will establish as the policy of the
United States that we do not wish to have content regulation by the
Federal Government of what is on the Internet, that we do not wish to
have a Federal Computer Commission with an army of bureaucrats
regulating the Internet because frankly the Internet has grown up to be
what it is without that kind of help from the Government.'' \26\
---------------------------------------------------------------------------
\26\ 141 Cong. Rec. H8470 (1995).
---------------------------------------------------------------------------
The House voted 420-4 to attach Cox and Wyden's amendment to its
version of the Telecommunications Act. As a compromise, the conference
committee included both the Senate's Communications Decency Act and the
House's amendment in the same Title of the telecommunications law.
Hence, the Cox-Wyden provision became known as ``Section 230 of the
Communications Decency Act,'' even though it had not been introduced
under with that title. Section 230 appeared largely as Cox and Wyden
proposed it, though it no longer contained the provision that banned
FCC regulation of Internet content. The final version also added an
explicit statement that ``[n]o cause of action may be brought and no
liability may be imposed under any State or local law that is
inconsistent with this section.'' \27\
---------------------------------------------------------------------------
\27\ 47 U.S.C. Sec. 230(e)(3).
---------------------------------------------------------------------------
From the relatively sparse legislative history, it is clear that
Section 230s drafters had two primary goals. First, they wanted to
ensure that the nascent commercial Internet was unburdened from
regulation and litigation. Second, they wanted to encourage online
providers to moderate as they (and their users) saw fit. In the short
discussion of Section 230 in the conference report for the
Telecommunications Act, the conferees wrote that they intended to
overrule the Stratton Oakmont v. Prodigy decision, and that ``such
decisions create serious obstacles to the important Federal policy of
empowering parents to determine the content of communications their
children receive through interactive computer services.'' \28\
---------------------------------------------------------------------------
\28\ H. Rep. 104-458 at 194.
---------------------------------------------------------------------------
On the day that President Clinton signed the Telecommunications Act
into law, civil liberties groups challenged the Senate's indecency
provisions, and the next year the Supreme Court would strike them down
as unconstitutional.\29\ The Supreme Court's ruling did not affect
Section 230. In fact, the civil liberties groups that challenged the
Communications Decency Act took care to not include Section 230 in
their litigation, recognizing the need to preserve Section 230.
---------------------------------------------------------------------------
\29\ Reno v. American Civil Liberties Union, 521 U.S. 844 (1997).
---------------------------------------------------------------------------
Section 230 received little attention in the months after it was
passed. This was in part because it was unclear how broadly courts
would interpret the 26 words. It was possible to read Section 230 as
merely conferring distributor liability standards to all interactive
computer service providers; in other words, a platform still could be
liable if it knew or had reason to know of the illegal user content. A
second, broader reading, would bar the platform from having any
liability for content provided entirely by third parties, unless an
exception applied.
This uncertainty ended on Nov. 12, 1997, when the United States
Court of Appeals for the Fourth Circuit adopted the latter, broad
reading of Section 230 in Zeran v. America Online. Distributor
liability, Judge J. Harvie Wilkinson wrote, ``is merely a subset, or a
species, of publisher liability.\30\ Thus, Wilkinson concluded, when
Section 230 states that an interactive computer service provider shall
not be ``treated as the publisher or speaker'' of information provided
by a third party, the statute also bars distributor liability.
``Section 230 was enacted, in part, to maintain the robust nature of
Internet communication and, accordingly, to keep government
interference in the medium to a minimum,'' Judge Wilkinson wrote.\31\
---------------------------------------------------------------------------
\30\ Zeran v. America Online, 129 F.3d 327, 332 (4th Cir. 1997).
\31\ Id. at 330.
---------------------------------------------------------------------------
Wilkinson recognized that subjecting an online service such as
America Online to notice-based liability likely would cause these
services to remove user content upon notice, even if the content was
not defamatory. ``Each notification would require a careful yet rapid
investigation of the circumstances surrounding the posted information,
a legal judgment concerning the information's defamatory character, and
an on-the-spot editorial decision whether to risk liability by allowing
the continued publication of that information,'' he wrote. ``Although
this might be feasible for the traditional print publisher, the sheer
number of postings on interactive computer services would create an
impossible burden in the Internet context.'' \32\
---------------------------------------------------------------------------
\32\ Id. at 333.
---------------------------------------------------------------------------
Because Judge Wilkinson was the first Federal appellate judge to
interpret Section 230, judges nationwide adopted his ruling in Zeran v.
America Online, and the broad reading of Section 230 became the law of
the land. Cox and Wyden--the authors of Section 230--told me as I was
researching my book that they agreed with Wilkinson's interpretation.
But it is possible to see how another judge might have concluded that
Section 230s scope if more limited.
The Zeran reading of Section 230 eliminates the Stratton Oakmont v.
Prodigy problem by preventing platforms from becoming liable for user
content they are unaware of simply because they have moderated some
other content. But it goes much further than that; it also allows
platforms to decide whether to keep up or take down content that they
are aware of without facing potential liability for that content. And
that is how Section 230 has created the legal framework for the
Internet that we know today.
Imagine how a social media site might behave had Judge Wilkinson
determined that Section 230 only means that all platforms be held to a
distributor liability standard. The site could face liability if it
knew or had reason to know of defamatory or otherwise actionable user
content. Such a liability regime might discourage the social media site
from actively moderating user content, as it might face liability for
content that it learned about but failed to remove. A social media site
with millions or billions of users is in no position to investigate
every user post and determine whether it is defamatory or otherwise
illegal. Section 230, as Judge Wilkinson interpreted it, removes that
disincentive to moderation.
Thanks to Judge Wilkinson's interpretation, Section 230 has
protected a wide range of platforms from many different types of
claims. As I detail in The Twenty-Six Words That Created the Internet,
this sweeping protection has been vital for consumer review sites,\33\
Wikipedia,\34\ social media,\35\ search engines,\36\ and countless
other sites that have built their business models around user-generated
content.
---------------------------------------------------------------------------
\33\ Kimzey v. Yelp! Inc., 836 F. 3d 1263 (9th Cir. 2016)
\34\ Bauer v. Glatzer, Docket No. L-1169-07 (Superior Court of
N.J., Monmouth County, 2008)
\35\ Doe v. MySpace, 528 F.3d 413 (5th Cir. 2008).
\36\ Fakhrian v. Google, No. B260705 (Cal. Ct. App. April 25, 2016)
---------------------------------------------------------------------------
Yet Section 230 also has shielded platforms in some cases in which
the plaintiffs have suffered serious harms. Among the lawsuits in which
courts held that Section 230 applies is one that involved a dating app
that was used to impersonate a man. The advertisements, posted by his
ex-boyfriend, claimed that the man wanted to engage in rape fantasies
or role play. This caused about 1,100 men to respond to the ads,
receiving the man's home and workplace locations via the app's
geolocation function. Many men visited his home and work, demanding sex
and drugs. The man said he contacted the app about 100 times, and only
received an automated response.\37\ He sued the app under a number of
theories of liability, including negligence, infliction of emotional
distress, products liability, and negligent design, but the district
court dismissed the claims on Section 230 grounds, and the Second
Circuit affirmed the dismissal.\38\ The district court also refused to
extend an earlier state court temporary restraining order that required
the app to ``immediately disable'' profiles that impersonated the
plaintiff. Section 230 has protected a gossip website that encourages
users to submit ``the dirt'' and selects which submissions to post and
highlight.\39\ And it has protected social media platforms used by
terrorists, even when the platform's algorithms helped make that user
content visible.\40\ As long as the website operator has not taken part
in the creation of the user content and an exception does not apply,
Section 230 will protect the website from liability arising from the
display and moderation of content created by others. Section 230 does
not block a plaintiff from suing the person who created the harmful
content, but there are a number of reasons why that might not be
practical, including the inability to track down the poster and fear of
retaliation.
---------------------------------------------------------------------------
\37\ Herrick v. Grindr, 306 F. Supp. 3d 579, 585 (S.D.N.Y. 2018).
\38\ Herrick v. Grindr, No. 18-396 (2d Cir. Mar. 27, 2019) (not
precedential).
\39\ Jones v. Dirty World Entertainment Recordings, 755 F.3d 398
(6th Cir. 2014).
\40\ Force v. Facebook, 934 F.3d 53 (2d Cir. 2019).
---------------------------------------------------------------------------
In short, Section 230 has fostered an Internet in the United States
that faces less regulatory and litigation burden than in other
countries, including other western democracies. This open Internet has
created many social benefits, but others have suffered real and serious
harms. For a broad perspective about the benefits and costs of the
Internet as governed by Section 230, I encourage you to read Hate
Crimes in Cyberspace by Danielle Citron, The Cult of the Constitution
by Mary Anne Franks, The Splinters of Our Discontent by Mike Godwin,
and Nobody's Victim by Carrie Goldberg.
II. Principles for Evaluating the Future of Section 230
Over the past year, Section 230 has been in the news more than any
other time in its nearly 25-year history. Often, the news is not
positive. Some critics argue that platforms have not adequately
moderated harmful content and have failed to achieve Congress's goal of
establishing content moderation systems that meet the needs of their
users. Other critics argue that some existing moderation policies and
procedures result in blocking certain political viewpoints.
Both criticisms have driven a number of proposals to change Section
230. I am not here today to endorse or propose any particular change to
Section 230. Rather, I hope to set forth some principles to guide your
evaluation of Section 230s future. I derive these principles from my
research into Section 230s history, and the impacts of courts'
interpretation of Section 230 over nearly a quarter century.
A. Not All Problems on the Internet Are Section 230 Problems
I recognize that this principle may sound odd coming from a
professor who asserts that Section 230 created the Internet. I maintain
that Section 230 provided the legal framework that allowed platforms to
structure their business models around user-generated content. But that
does not mean that every flaw in the current system is attributable to
Section 230. There is a lot to love about the Internet, but there also
is a lot not to love about the Internet. Some content is vile. Some
ruins lives. Some does lasting damage to society and our institutions.
But before placing all of the blame for this content on Section 230, it
is important to first examine whether a cause of action exists for that
harm. If a cause of action does not exist, then there is nothing for
Section 230 to block.
For instance, a big headline on the cover of the New York Times
business section last August proclaimed: ``Why Hate Speech on the
Internet is a Never-Ending Problem.'' Below the headline were the key
26 words from Section 230, followed by: ``Because this law shields
it.'' The Times later appended the following dramatic correction to the
story: ``An earlier version of this article incorrectly described the
law that protects hate speech on the internet. The First Amendment, not
Section 230 of the Communications Decency Act, protects it.'' Despite
the correction, later that month, a Federal judge cited this article
while describing the debate over ``Section 230s grant of immunity for
speech-based harms such as hate speech or libel.'' \41\ To be sure,
online hate speech is a serious problem, but the reality is that the
First Amendment protects hate speech, regardless of Section 230. Of
course, the Supreme Court's First Amendment jurisprudence could evolve
to treat hate speech differently--and some believe it should. And even
now, if that hate speech also is illegal for some other reason (for
example, because it is a true threat or defamatory), then it could fall
outside the scope of First Amendment protection. But hate speech,
standing alone, is constitutionally protected. Changing Section 230
would not change platforms' legal obligations in this area.
---------------------------------------------------------------------------
\41\ Papataros v. Amazon.com, Civ. No. 17-9836 (D. N.J. Aug. 26,
2019).
---------------------------------------------------------------------------
Many defamation claims that courts dismiss on Section 230 grounds
would also, if fully litigated, not survive common law and First
Amendment protections. These include the requirement for falsity, the
opinion privilege, and the actual malice bar for public officials and
figures. Because Section 230 provides strong procedural protections,
defamation lawsuits against platforms often are decided in the early
stages, eliminating the need for the parties to engage in extensive
discovery and for courts to decide fact-intensive questions about
defamation law. Additionally, as seen in the 1950s bookseller cases,
the First Amendment and common law provide some protection to
distributors of content created by others. As I describe in the next
subsection, there is uncertainty as to how extensive that protection
is.
In addition to hate speech concerns, large companies--including big
technology platforms--have been rightly criticized for their privacy
and data security practices. These are serious problems that I hope
Congress will address with comprehensive and effective laws that set
tough national standards for privacy and cybersecurity. Section 230,
however, is not at the root of these problems. Section 230 only
protects platforms from liability for third-party content; it does not
affect their liability after a data breach or generally shield their
data collection practices.
Nor does Section 230 have any link to copyright infringement. From
the beginning, Section 230 has had an exception for intellectual
property law. Platforms and content creators have long been engaged in
a spirited debate over the notice-and-takedown system established by an
entirely different law, the Digital Millennium Copyright Act.
Unfortunately, recent media reports have conflated Section 230 and the
DMCA. Likewise, Section 230 always has had an exception for the
enforcement of Federal criminal law, so user content that constitutes a
Federal crime is not covered by the statute's protections.
B. We Don't Know How Platforms Would React to a Repeal or Significant
Contraction of Section 230
Although there are not any legislative proposals to repeal Section
230, repeal has been publicly suggested in the media, and it is
important to examine what the Internet might look like without Section
230. Moreover, eliminating Section 230 protections for a particular
type of content might have a similar impact on some platforms.
Because Section 230 has been on the books since 1996, it is
difficult to know with certainty what the Internet would look like
without it. This uncertainty stems from the lack of caselaw that
extrapolates common law liability standards to modern online platforms.
We can only look at cases involving bookstores, and the few non-binding
opinions involving Prodigy and CompuServe that were decided before
Section 230s passage.
If courts were to adopt the Stratton Oakmont v. Prodigy line of
thinking, platforms would fear being dubbed ``publishers,'' who are
subject to the same liability for user content as the authors, rather
than ``distributors,'' who are liable only if they knew or had reason
to know of the illegal content. I believe that the judge in this case
got the law wrong, drawing an artificial line between a publisher that
exercises ``editorial control'' and distributor, when all distributors
exercise some degree of editorial control (for instance, a bookstore
could refuse to sell a certain book). Still, there is no guarantee that
courts would disagree with the Stratton Oakmont decision. If it were
widely adopted, this reasoning likely would discourage platforms from
engaging in any moderation, lest they be dubbed common-law
``publishers.''
If courts were to reject the Stratton Oakmont v. Prodigy holding
(as I hope they would), online platforms could face liability if they
knew or had reason to know of the illegal content. This might result in
a system in which anyone could complain to a platform about user
content, regardless of the merit of their complaints, at which point a
platform that did not take down the content would risk being forced to
defend it in court. Platforms also might avoid moderation, fearing that
once they encounter potentially defamatory or otherwise actionable
content, they would become liable for it. Complicating matters, it is
unclear when a platform would have a ``reason to know'' of illegal
content, as there is very little caselaw that articulates when a
distributor has ``reason to know.'' For instance, this might expose a
platform to liability if it generally knew that users posted defamatory
material, but had not seen the particular post in question.
In the landmark Zeran case, Judge Wilkinson warned that it was
``impossible'' to expect platforms to screen user content. ``Faced with
potential liability for each message republished by their services,
interactive computer service providers might choose to severely
restrict the number and type of messages posted,'' he wrote. ``Congress
considered the weight of the speech interests implicated and chose to
immunize service providers to avoid any such restrictive effect.'' \42\
Indeed, it is conceivable that some platforms might reduce or entirely
eliminate user content in a world without Section 230s protections.
---------------------------------------------------------------------------
\42\ Zeran v. America Online, 129 F.3d 327, 331 (4th Cir. 1997).
---------------------------------------------------------------------------
A change to Section 230--even short of full repeal--may have
significant impacts on platforms' operations. For instance, when
Congress amended Section 230 in 2018 to create an exception for certain
civil actions and state criminal cases involving sex trafficking,
Craigslist removed the personals section it had hosted for years. ``Any
tool or service can be misused,'' Craigslist wrote. ``We can't take
such risk without jeopardizing all our other services, so we have
regretfully taken craigslist personals offline.''
C. Section 230 is Designed to Encourage--Not Discourage--Moderation of
User
Content
In my decade writing about Section 230 and practicing Internet law,
I have encountered far too many lawyers who advise website operators
that if they moderate user content, they will lose their Section 230
protections. I fear that this has caused websites to take a hands-off
approach to user content that they otherwise would have blocked.
This advice is simply incorrect. Section 230s protections do not
disappear merely because a platform has engaged in content moderation.
As Section 230s legislative history makes clear, one of the main
purposes of Section 230 was to encourage online service providers to
moderate. Indeed, the title of the most important section of the
statute is ``Protection for `Good Samaritan' Blocking and Screening of
Offensive Material.''
To be sure, Section 230 does not require moderation. Rather, the
law leaves it up to the providers to determine what moderation--and
moderation tools--to provide to their users. Section 230 is very much a
market-based law, based on the assumption that user demands will
dictate platforms' moderation approaches.
Of course, Congress can and should determine whether the market-
based system continues to meet users' expectations in 2020, when a
handful of platforms have market capitalizations that are greater than
those of automakers. We are in a very different world than 1996, when
40 million people worldwide had Internet access, and being suspended
from Prodigy was unlikely to have significant consequences to one's
livelihood. Suspension from a large social media platform in 2020, on
the other hand, has a much greater impact.
In the debate over ``neutrality'' of platforms, I see a few
different questions. First: Does Section 230 currently require
neutrality? Second: Should Section 230 require neutrality?
The answer to the first question, as explained above, is ``no.''
The answer to the second question is up to you, as Congress is free to
amend Section 230 as it sees fit. If Congress were to attempt to impose
a neutrality requirement, I would ask what such a requirement would
look like, and how it would be implemented. Moderation often requires
difficult judgments about content being transmitted at a furious pace.
Could a platform block any content while still remaining ``neutral?''
Would a ``neutral'' Internet full of legal pornography, threats,
bullying, or encouragement of anorexia be an improvement? Even if this
neutrality requirement were limited to political speech, some political
debates can border on hate speech. If a platform were to moderate a
political discussion for violating its hate speech policies, would that
violate a neutrality requirement? These questions are tough to answer
in the abstract, and even more difficult when presented with the
torrent of choices that platforms must make every minute.
D. The Section 230 Debate Needs More Transparency
I am thrilled to see Section 230 suddenly receiving much-deserved
attention, as it is one of the most important technology-related laws
in the United States. Unfortunately, some of this attention has lacked
precision and accuracy. This is due to a number of problems, including
the nuances of Internet liability law and what I imagine is a
substantial amount of lobbying efforts on all sides.
But the debate also is muddled because the general public has
little insight into the possibilities--and challenges--of content
moderation at scale. Until recently, many large tech companies were not
terribly transparent about their policies and practices, though the
recent Section 230 debates have had the positive impact of shining a
bit of sunlight on content moderation. We need far more. Platforms
should continue to provide more information about how and why they
moderate content, and the possibilities and limits of human-based and
automated moderation. If platforms are not transparent, Congress should
consider whether to require or provide incentives for better
transparency.
Before we can develop new policies regarding intermediary liability
and content moderation, we need a more robust factual record. Section
230 is too important to overhaul in the dark. Last October, I suggested
the creation of a congressionally chartered commission to gather facts
and recommend a path forward.\43\ The commission would have a wide
range of stakeholders, including civil liberties groups, victims'
advocates, law enforcement, and technology companies and their counsel.
The Cyberspace Solarium Commission provides a good model for a
bipartisan group of experts who gather facts and develop well-reasoned
proposals.
---------------------------------------------------------------------------
\43\ Jeff Kosseff, Understand the Internet's Most Important Law
Before Changing It, Regulatory Review (Oct. 10, 2019).
---------------------------------------------------------------------------
A commission also could help to better identify the goals of
Section 230 reform and sort through the many current calls for reform,
some of which conflict with one another. The criticisms of platforms
vary widely, with some arguing that platforms do not moderate enough,
and others arguing that they moderate too much, at least for certain
political viewpoints. It is difficult to reconcile these criticisms,
let alone modify Section 230 in a manner that satisfies of them. Before
we identify a solution, we must agree on a problem.
* * *
I commend the Chairman and Ranking Member for the thoughtful
solutions that you propose in the Platform Accountability and Consumer
Transparency Act. The legislation addresses the need for more
transparency in content moderation policies and procedures, and begins
the process of identifying the most tailored and reasonable rules for
providing that transparency. The bill also provides people with a
mechanism to take down material that has been adjudicated to be
defamatory or illegal under Federal criminal or civil law. In my
experience, plaintiffs who are the victims of the most harmful
defamation campaigns are most interested in having the material removed
rather than recovering damages, and this legislation provides them with
an avenue. We must ensure that a take-down provision is not abused--for
example, via the falsification of court orders--but it also is
important to allow for the removal of material that is adjudicated to
be illegal.
As I routinely remind technology companies, Section 230 is not set
in stone, and can be repealed or significantly amended as easily as it
was passed. Congress may determine that it is in the public interest to
curtail some or all of Section 230s protections. I urge you to make any
such decisions with great care.
You likely will hear from many sides of the Section 230 debate
about the consequences of your action or inaction. They likely will
inform you of these consequences with great certainty. As I have
outlined today, there are many reasons to be uncertain about the
precise impacts of changes to Section 230. The best that we can do is
identify the problems, gather as much information as possible, and
address these problems in a focused and tailored manner.
Our online ecosystem relies on these 26 words. As I write in my
book, our modern Internet is a house that is ``built on the foundation
of Section 230.'' It is difficult to imagine how some of Silicon
Valley's largest companies could have emerged--at least in their
current forms--without Section 230. The challenge for all of us is to
determine how we want the Internet to look over the next 25 years and
what it takes to get it.
Senator Thune. Thank you, Mr. Kosseff.
Next up is Ms. Elizabeth Banker.
Please proceed.
STATEMENT OF ELIZABETH BANKER,
DEPUTY GENERAL COUNSEL, INTERNET ASSOCIATION
Ms. Banker. Chairman Thune, Ranking Member Schatz, and
members of the Subcommittee, thank you for inviting me to
testify at this important hearing.
My name is Elizabeth Banker, and I'm Internet Association's
Deputy General Counsel.
IA is grateful for the opportunity to appear before the
Subcommittee to discuss Section 230, the foundational law that
empowers the modern Internet. We appreciate the Subcommittee's
thoughtful approach to understanding the history and purpose of
Section 230.
IA also appreciates the focus on the twin goals of
promoting transparency and accountability in content moderation
that are at the heart of the PACT Act.
While we have feedback on the bill, it demonstrates that
not all problems related to online content can, or must, be
solved by amending Section 230. IA hopes to continue our work
with the authors to ensure that the bill can achieve its
objectives without hindering innovation and flexibility in
content moderation.
Section 230 empowers companies to offer innovative services
while simultaneously setting and enforcing policies for using
those services. The law carefully balances free expression and
protecting consumers in a way that serves their users and their
service, and also allowing them to respond to an ever-changing
set of challenges. Without Section 230, many individuals and
organizations would not be able to create spaces for
discussion, because of potential liability for every post.
Section 230 removes disincentives for companies to set and
enforce rules for the vast amount of content disseminated on
their platforms. As panelists have explained, Section 230
resolves what is called the ``moderator's dilemma,'' allowing
Internet companies to adopt and enforce community standards
without the fear that it will expose them to unnecessary and
often baseless lawsuits. IA members do exactly that, they set
and enforce rules for their services, working continually to
make them safer. From child sexual abuse material to terrorist
content, from self-harm to targeted harassment, IA members have
long track records of resource-intensive efforts to combat
objectionable online content and providing tools to allow their
users to control their online experiences.
Our member companies are constantly learning and adapting
their approaches to strike the appropriate balance between
allowing expression and protecting users. It's not easy, and
such action is frequently subject to criticism from all sides,
concerned either that too much or too little has been done.
Spam is a helpful example of how Section 230 works.
Providers must continually adjust their enforcement efforts in
realtime as spammers adopt new techniques designed to evade
detection. The scale of these efforts is staggering. Facebook
took action against 1.9 billion pieces of spam in a 3-month
period. In multiple cases, Section 230 has shielded providers
from lawsuits from spammers who sued over removing their spam
material. And the courts have applied Section 230 and allowed
the valuable work that the companies do to continue.
To better understand how the law works more broadly, IA
reviewed over 500 decisions involving Section 230. While the
national policy debate is focused on a few extreme examples
that break into national media or specific content moderation
decisions, the importance of Section 230 is best demonstrated
by the lesser-known cases that escape the headlines. These
cases show that the law continues to perform as Congress
intended, quietly protecting discussion boards operated by
soccer parents, nurses, police associations, and labor union
members, protecting them from lawsuits.
When applied by courts, Section 230 is far from a blanket
immunity. Only 42 percent of the decisions we reviewed relied
primarily on Section 230. Over a quarter of the decisions
involved claims that were dismissed for case defects that were
separate and apart from Section 230. Courts rejected Section
230 defenses when they did not apply. Further, courts looked
carefully at the provider's role in creating content, a
determining factor on whether or not Section 230 applies,
frequently requiring further investigation before making a
decision. Ultimately, our study supports the call for a
thorough and unbiased review of 230 to determine what, if any,
changes are necessary before legislating.
Great care should be taken when considering possible
changes to Section 230 or legislating on content moderation,
given the ever-evolving nature of Internet technology and the
complexity of law surrounding online speech.
Thank you. I look forward to your questions.
[The prepared statement of Ms. Banker follows:]
Prepared Statement of Elizabeth Banker, Deputy General Counsel,
Internet Association
Chairman Thune, Ranking Member Schatz, and members of the
Subcommittee, thank you for inviting me to testify at this important
hearing today. My name is Elizabeth Banker, and I am Deputy General
Counsel of Internet Association.
Internet Association is grateful for the opportunity to appear
before this Subcommittee to discuss Section 230--the foundational law
that has fostered the development and growth of the variety of online
services that consumers consider the best of the internet. We
appreciate the Subcommittee's thoughtful approach to understanding the
history and purpose of Section 230, and I hope my testimony will assist
in your efforts.
IA is the only trade association that exclusively represents
leading global Internet companies on matters of public policy. IA's
mission is to foster innovation, promote economic growth, and empower
people through the free and open internet. IA believes the Internet
creates unprecedented benefits for society, and as the voice of the
world's leading Internet companies, IA works to ensure policymakers and
other stakeholders understand these benefits.
Section 230 plays a critical role in empowering companies to offer
innovative services and set and enforce policies regarding the use of
those services. IA hopes, through our testimony, to explain: (1) how
Section 230 enables our members' services by allowing them to take
action against harmful activity when they find it; (2) how the law
strikes a careful balance by barring certain types of lawsuits and
encouraging moderation; (3) the role of the First Amendment in this
debate; and (4) considerations for policymakers looking at possible
amendments to Section 230 including IA's preliminary thoughts on the
PACT Act. This testimony also provides new research, based on our
analysis of more than 500 court decisions involving Section 230, that
sheds light on the wide variety of parties using the law, how the law
affects litigation, and how courts apply it.
Many of the things people consider to be the ``best of internet''
are possible because of Section 230. IA's research shows that consumers
value hearing from other consumers about their experiences before
making major purchases, booking travel, and ordering a ride-share.\1\
Consumers check online reviews more frequently than recommendations
from experts or friends. Section 230 allows users to access and share a
wide range of information, opinions, and experiences. This type of
sharing is at the core of many IA members' services and is what makes
them enjoyable, useful, and engaging for their users. It is difficult
to imagine a world where all of that would be possible if, for example,
a travel site could be held legally responsible for every word in every
review it hosts.
---------------------------------------------------------------------------
\1\ Internet Association, Best of the Internet Survey, June
26,2019. Available at: https://internetassociation.org/publications/
best-of-the-internet-survey/.
---------------------------------------------------------------------------
IA member companies recognize that in order to realize the full
benefits of the internet, it is critical that they take action to
prevent and respond to harmful online activities. This is essential to
building and maintaining both user and public trust. Today's world,
where we grapple with a global pandemic and a social justice movement
that is a reckoning with lives lost to systemic discrimination, has
shown both the tangible benefits of online services and the critical
role providers play in ensuring that their services are not undermined
and misused in ways that threaten individual lives or the public good.
IA members have played an essential role in helping society
transition into today's ``new normal.'' Their services allow us to stay
connected to loved ones, order takeout to support local restaurants,
conduct doctors' appointments via telehealth services, and even work
from home through video conferences.
While IA's members recognize that their platforms always have room
for improvement, they are consistently working to find ways to make
their services safer--whether by highlighting authoritative sources of
accurate information about COVID-19 and addressing dangerous
misinformation, or by working to make underrepresented and marginalized
groups feel that they have a safe place to express themselves. Many of
our members have made commitments as a result of recent events to do
more, and IA as an organization is also actively working to support
these efforts. IA has centralized and detailed member company efforts
in response to COVID-19 as a resource for the public and
policymakers.\2\ As part of its commitment to social justice, IA is
building on the work in its 2019 Diversity & Inclusion Benchmark
Report; helping underrepresented groups find employment opportunities
with technology companies through a soon-to-be-launched job portal; and
supporting social justice reform legislation.
---------------------------------------------------------------------------
\2\ https://covid19.internetassociation.org/industry/response/.
---------------------------------------------------------------------------
I. Section 230 Is Critical To Content Moderation And Content Moderation
Is Critical To Realizing The Value Of Online Services
In considering possible amendments to Section 230, it is vital to
remember the statute's history. Congress enacted Section 230, in part,
to encourage providers of online services to voluntarily adopt robust
content moderation policies and practices. Congress was reacting to two
lower court cases, Cubby, Inc. v. CompuServe Inc., 776 F. Supp. 135
(S.D.N.Y. 1991), and Stratton Oakmont, Inc. v. Prodigy Services Co.,
1995 WL 323710 (N.Y. Sup. Ct. May 24, 1995). Together, Cubby and
Stratton Oakmont created a powerful disincentive for Internet companies
to monitor and remove objectionable content by threatening to expose
companies to burdensome litigation and potential liability based on
their very efforts to moderate that content.\3\
---------------------------------------------------------------------------
\3\ In Cubby, a Federal district court held that an interactive
service provider, CompuServe, could not be held liable for allegedly
false statements that a third-party had posted in one of its online
forums unless CompuServe knew or had reason to know of the allegedly
false statements. 776 F. Supp. at 139-141. The plaintiffs had sought to
hold CompuServe liable for allegedly false and defamatory statements
contained in a third party's daily newsletter that CompuServe hosted.
Id. at 137, 140. The court noted that it would hardly be feasible ``for
CompuServe to examine every publication it carries for potentially
defamatory statements.'' Id. at 140. In granting CompuServe's motion
for summary judgment, the court analogized CompuServe to distributors
of third-party content such as bookstores and newsstands. Id. The court
explained that the requirement that such distributors ``must have
knowledge of the contents of a publication before liability can be
imposed for distributing that publication is deeply rooted in the First
Amendment.'' It therefore concluded that CompuServe could not be held
liable unless it knew or had reason to know of the allegedly false
statements. Id. at 140-141. Given the facts of the case--including that
CompuServe exercised ``little or no editorial control'' over the third-
party content available on its platform--the court held that the
plaintiffs had failed to set forth sufficient evidence that CompuServe
had the requisite knowledge, and the court thus granted CompuServe
summary judgment. Id.
---------------------------------------------------------------------------
Before the enactment of Section 230, these cases presented Internet
companies with a difficult choice. If they voluntarily adopted content
moderation policies and practices, they could end up like Prodigy--
treated as a ``publisher'' that could be held liable for user-generated
content. But if they sought to avoid this liability as CompuServe had,
they would be forced to take a hands-off approach and bury their heads
in the sand in an attempt to avoid acquiring knowledge of objectionable
third-party content. This dilemma is exacerbated by the immense and
rapidly increasing volume of third-party content that online platforms
host and are used to disseminate, which makes detecting objectionable
content exponentially more difficult. Pre-publication review cannot be
scaled to match the rate at which new content is posted, and
consequently, requiring it would undermine the core value of these
real-time, interactive services.
By contrast, in Stratton Oakmont, a New York state court held that
the interactive service provider Prodigy could be held liable for
allegedly defamatory statements posted on its message boards because it
employed staff and used software to monitor and police content in order
to attain a reputation as a ``family oriented'' service. 1995 WL
323710, at *2-4. The court agreed with the conclusion in Cubby that
mere ``distributors'' may be liable for defamatory statements of others
only if they knew or had reason to know of the defamatory statements at
issue. Id. But the court concluded that Prodigy was instead a
``publisher,'' liable as if it had itself made the statements, because
the court viewed Prodigy as analogous to a newspaper that is ``more
than a passive receptacle or conduit for news, comment and
advertising.'' Id. As a result, the court ruled Prodigy could be held
liable for defamatory content posted on its message boards even if it
lacked knowledge of that content. Id. The key distinction, according to
the court, was that unlike CompuServe, Prodigy ``held itself out as an
online service that exercised editorial control over the content of
messages'' on its platform. Id.
Section 230 provides a thoughtful solution to the so-called
``moderator's dilemma.'' It allows Internet companies to adopt and
enforce community standards without the fear that doing so would expose
them to an onslaught of burdensome lawsuits. In this way, Section 230
creates critical breathing room for online providers to voluntarily
undertake moderation of the unprecedented stream of content that users
disseminate through their platforms. It creates a middle ground between
the wild west of completely passive platforms and the closed-to-the-
public realm of newspapers and other media outlets that develop and/or
hand-select content for publication. That is why the statute plays such
a critical role in ensuring that companies of all sizes, including IA's
members, can operate the online services that the public finds so
valuable.
II. Section 230 Achieves The Careful Balance It Was Designed To Create
Section 230 has been successful in achieving the goals that led to
its enactment. IA member companies have adopted and enforced essential
content moderation policies, just as Congress intended in enacting
Section 230. In numerous areas--from child sexual abuse material (CSAM)
to terrorism-related content, and from self-harm to fake reviews--IA
member companies have undertaken decades-long and resource-intensive
efforts to combat objectionable online content. At the same time,
Section 230 has allowed the online economy to develop and prosper in
the United States in ways that simply have not been replicated
elsewhere around the globe. Section 230 has spurred the vibrant growth
of the Internet and a wide variety of diverse platforms, while also
permitting Internet companies to protect users, and to promote
healthier online discourse, through responsible domestic and
international content moderation.
A few examples can illustrate this point.
First, IA member companies take multifaceted approaches to
combating CSAM on their services and in the world that are enabled by
Section 230. For example, Microsoft donated PhotoDNA, image-matching
software that detects CSAM, to the National Center for Missing and
Exploited Children (NCMEC), so that it could be licensed for free to
other entities to identify versions of previously reported CSAM. The
use of existing and newly developed detection tools has significantly
increased, as is evidenced by the dramatic growth in the number of
CyberTipline reports in recent years. Today, IA member companies,
alongside governments, civil society, and other stakeholders,
continually work to stop bad actors from spreading CSAM online. They
take a variety of actions, including dedicating engineering resources
to the development and improvement of tools like PhotoDNA and Google's
CSAI Match, assisting in the modernization of the CyberTipline through
donations of engineering resources or funds, and engaging with law
enforcement agencies. Many companies also proactively detect instances
of CSAM and report to NCMEC.
IA member companies have also engaged in serious efforts to
eliminate content advocating or promoting terrorism. Twitter suspended
115,861 unique accounts for violations related to the promotion of
terrorism during the first half of 2019.\4\ Over 85 percent of those
accounts were flagged by internal tools developed by Twitter itself,
and many of the accounts were suspended before they ever issued even a
single tweet. In the first quarter of 2020, Facebook took action on 6.3
million pieces of content supporting terrorism, with 99.3 percent of
such content internally flagged before a third party reported it.\5\
During the same period, YouTube removed 258,908 videos for violating
its policies against violent extremism.\6\ IA member companies
consistently work to quickly remove any content that advocates
terrorism.
---------------------------------------------------------------------------
\4\ Twitter Transparency Report, Jan.-June 2019, Rules Enforcement.
Available at: https://transparency.twitter.com/en/twitter-rules-
enforcement.html.
\5\ Facebook Transparency, Community Standards Enforcement Report.
Available at: https://transparency.facebook.com/community-standards-
enforcement#dangerous-organizations.
\6\ Google Transparency Report, YouTube Community Guidelines
Enforcement, Video Removals by Reason. Available at: https://
transparencyreport.google.com/youtube-policy/removals?hl=en&
total_removed_videos=period:Y2020Q1;exclude_automated:human_only&lu=tota
l_removed_videos.
---------------------------------------------------------------------------
IA member companies also employ a multitude of general-purpose
technologies to support their content moderation efforts. IA members
provide ``report abuse'' buttons and other mechanisms so that users can
flag problematic content or contact the companies with complaints. The
companies also provide specific community guidelines that provide
standards for third-party content, and they devote significant staff
and resources to enforcing those policies. Broad collaboration with
civil society groups and other experts informs and deepens our members'
commitment to safety and security. In addition, the companies have
developed sophisticated software and algorithms to detect and remove
harmful content. In many instances, they have shared these technologies
to help others eradicate that harmful content as well. Some companies
also dedicate large teams of staff that can provide quick responses to
evolving problems, including responding to user complaints and removing
objectionable and unlawful content. These efforts are the types of
activities that Section 230 was designed to promote. It is because of,
not in spite of, the law that IA members are able to take action to
create safe experiences for their users.
Section 230 has played a particularly important role in creating
space for online platforms to refine their approaches to content
moderation over time. Moderating content is not easy given the enormous
volume of content online and the sometimes-nuanced distinctions that
platforms must make to strike the right balance between which content
to remove and which to leave up. Our member companies recognize that
they do not always achieve the perfect balance, but they are constantly
learning, adapting, and updating their approaches.
Section 230 allows online companies the room to experiment in this
way without having to worry that they will face the heavy costs of
litigation each time a mistake is made or someone is unhappy with a
moderation decision. Companies can learn and make adjustments--an
essential process that they engage in constantly.
The difficulty of content moderation and the importance of Section
230 is best demonstrated using an example of content that is
universally hated--spam. Since the advent of the commercial internet,
spammers have been intent on finding ways to flood online services with
unwanted commercial messages. Their business is one of volume--if
enough messages go out, even if only a small percentage are acted upon,
it is profitable. The high volumes of spam messages can operate as a
literal or figurative ``denial of service attack.'' They can choke
capacity of even large providers and render services of minimal value
to their users by obscuring the content users want to see. It is for
these reasons that spam was among the earliest targets of proactive
content moderation efforts and exemplifies the challenges providers
face in keeping pace with bad actors who are determined to misuse their
services.
Spam detection has evolved over time from simple techniques, such
as spam block lists and rate limiting on accounts to prevent any one
account from sending too many messages at once, into something
altogether more sophisticated. While many of the early techniques
remain important tools, new algorithmic approaches that pull signals
from a variety of sources are essential today. These more sophisticated
tools are able to assign risk based on numerous indicators and then
apply any one of a variety of interventions, including pausing account
activity, requiring further account verification or passing reCaptchas
to verify it is not automated activity, demoting suspect content,
blocking or deleting content, and closing accounts of violators. The
battle between spammers and service providers can be characterized as
an arms race, as spammers quickly adapt to detection techniques and
providers must continually respond. The automated systems that protect
providers' services from spam may be changed on a daily, if not a more
frequent, basis.
The volume of spam activity actioned by IA members is staggering.
For example:
Facebook: In the three-month period from July to September
2019, Facebook took action against 1.9 billions pieces of
content for spam.\7\
---------------------------------------------------------------------------
\7\ Facebook Transparency Report, n. 5.
Twitter: During the first six months of 2019, Twitter
received over 3 million user reports of spam and challenged
over 97 million suspected spam accounts.\8\
---------------------------------------------------------------------------
\8\ Twitter Transparency Report, n. 4.
YouTube: In the first quarter of this year, 87.5 percent of
channel removals were for violations that were related to spam,
scams, and other misleading content resulting in 1.7 million
channels being removed. In addition, in the same period,
YouTube removed over 470 million spam comments.\9\
---------------------------------------------------------------------------
\9\ Google Transparency Report, n. 6.
Section 230 is critical to these content moderation efforts.
Indeed, service providers sued by spammers for removing spam have
asserted Section 230 as a defense.\10\ Section 230 is even more
critical to efforts to address content for which there is no general
global agreement that it is harmful or should be restricted. Providers
develop policies across a range of issues that are extremely nuanced
and uniquely tailored to their services, addressing a broad range of
behaviors that are disruptive to the goal of the service they provide.
There are frequently contrasting views about whether individual content
moderation decisions were correct or flawed. No single solution could
ever balance all of the competing visions of how content moderation
ought to work. Instead, Section 230 protects a critical equilibrium
that safeguards free expression and promotes user safety, while
allowing providers the flexibility to respond to an ever-changing
landscape of challenges in a way that best serves their users and their
unique services.
---------------------------------------------------------------------------
\10\ See, infra, fn. 16.
---------------------------------------------------------------------------
III. IA's Review of Section 230 Decisions
Over a year ago, IA began reviewing court decisions involving
Section 230 with a goal of developing a better understanding of how the
law works in practice. Having now reviewed more than 500 decisions, IA
is sharing its observations which demonstrate the need for in-depth
study of this case law to inform the public policy debate over Section
230. In recent years, the national policy debate around Section 230 has
focused on a few cases that garnered national media attention or
specific content moderation decisions by particular providers.
Employing a holistic approach will ensure that all stakeholders have a
comprehensive understanding of Section 230 before advocating for
changes to the careful balance that it strikes.
IA's findings are further described in the attached paper, along
with a description of our methodology and the list of decisions
reviewed. IA acknowledges that the review was not comprehensive and
that there are inherent limitations in attempting to draw broad
characterizations from the outcome of any stage in litigation. However,
we found clear patterns and observations of important note for
policymakers based on judicial decisions reviewed where Section 230
immunity was implicated. IA believes that this initial effort provides
a sufficient basis to support a call for a comprehensive and unbiased
review of Section 230 before any action is taken to change the law. I
would like to share some of our observations with you today.
A. Section 230 Benefits A Wide Range Of Entities.
IA's review of Section 230 decisions revealed that it is not only
large social media companies that assert Section 230 as an affirmative
defense. The importance of Section 230 is best demonstrated by the
lesser-known cases that escape the headlines. Online users; Internet
service providers and website hosts; online newspapers; universities;
libraries; search engines; employers; bloggers, website moderators and
listserv owners; marketplaces; app stores; spam protection and anti-
fraud tools; and domain name registrars have all asserted Section 230
immunity. These decisions show the law quietly protecting soccer
parents from defamation claims, discussion boards for nurses and police
from nuisance suits, and local newspapers from liability for comments
posted by trolls.
It is critical to keep these smaller entities in mind when
evaluating the value of Section 230. For example, in Joyner v.
Lazzareschi,\11\ Lazzareschi, a soccer parent and the operator of a
local online messaging board for youth soccer called SoCalSoccerTalk,
was sued by Joyner, a disgruntled soccer coach, for allegedly
defamatory comments that parents made on the regional messaging board.
While Lazzareschi would have fallen under the Section 230 definition of
a ``provider'' of an ``interactive computer service'', the case was
ultimately dismissed and the decision was upheld on appeal for Joyner's
failure to meet the requirements for a defamation claim. Another
example of a lesser-known entity to assert Section 230 is
Allnurses.com, in the case of East Coast Test Prep LLC v. Allnurses.com
Inc.\12\ In this case, Allnurses.com, was sued by East Coast Test Prep
(ECTP) because two nurses made negative remarks about ECTP's services.
While this case was dismissed at the summary judgment phase for a
variety of shortcomings in the plaintiff's case, Allnurses.com also
successfully argued that Section 230 protected its service from
liability for allegedly defamatory statements made by the nurses that
used their message board to discuss topics important to the nursing
field, including the relative merits of test prep providers. It is
these small fora and communities, local soccer messaging boards and
discussions of nursing exam courses, that would be silenced by
crippling litigation without Section 230.
---------------------------------------------------------------------------
\11\ No. G040323, (Cal. App Jul 10, 2012).
\12\ No. Civ. 15-3705 (JRT/SER), (D. Minn. Jan 26, 2018).
---------------------------------------------------------------------------
These examples represent just two of the seldom discussed entities
that are among the wide-cross section of Section 230 beneficiaries.
They are joined by local newspapers, labor unions, police associations,
individuals, and others who provide spaces for users to discuss topics
of interest. These entities and individuals make important
contributions to the online ecosystem that exists today. During the
pandemic, many of these online communities that support sharing of
hyperlocal information, like the length of the line at the local COVID
testing site, the health and safety measures employed by a favorite
neighborhood restaurant, or resources for assistance such as food
banks, play a critical role in helping us cope and recover. It is
important for this Subcommittee to keep in mind the impact that
changing Section 230 may have on a variety of entities and individuals
within the online platform space.
B. Courts Dismiss Many Cases In Which Section 230 Is Raised As A
Defense Based On Unrelated Defects In Plaintiffs' Claims.
Our review found that Section 230 is far from a ``blanket
immunity'' \13\ when it comes to the law's application in the courts.
Instead our research demonstrates that only 42 percent of decisions
reviewed were decided primarily based on Section 230 immunity. In over
a quarter of the decisions (28 percent), the courts dismissed claims
without relying on Section 230 because the plaintiff failed to state a
claim upon which relief could be granted, or because of other defects
in their case. Courts rejected attempts to rely on Section 230 when it
was not applicable--whether because the party asserting 230 was not a
covered entity, an exception applied, or the party asserting 230 was a
content provider of the information at issue. Courts carefully consider
the issue of the service provider's role in the creation of the
problematic content, a determining factor on whether Section 230
applies.\14\ When rejecting complaints based on Section 230, judges
frequently explained in detail the requirements to adequately allege
that the provider developed--in whole or in part--the content at issue,
and gave plaintiffs multiple tries to amend their complaints. When
plaintiffs did raise factual issues as to the service provider's role
in content development, courts required discovery to allow further
investigation before rendering a judgment as to whether Section 230
applied.\15\
---------------------------------------------------------------------------
\13\ See, e.g., Executive Order on Preventing Online Censorship,
May 28, 2020. Available at: https://www.whitehouse.gov/presidential-
actions/executive-order-preventing-online-censorship/; Department Of
Justice's Review Of Section 230 Of The Communications Decency Act Of
1996, at 4(a). Available at: https://www.justice.gov/ag/department-
justice-s-review-section-230-communications-decency-act-
1996?utm_medium=e-mail&utm_source=govdelivery.
\14\ See, e.g., Enigma Software Group v. Bleeping Computer, 194 F.
Supp. 3d 263 (2016); Tanisha Systems 3v. Chandra, 2015 U.S. Dist. LEXIS
177164 (N.D. Ga. 2015); Perkins v. LinkedIn, 53 F. Supp. 3d 1222
(2014); Brummer v. Wey, 2016 NY Slip Op 31021(U); Dimetriades v. Yelp,
228 Cal. App. 4th 294 (2014).
\15\ See, e.g., General Steel v. Chumley, 840 F.3d 1178 (10th Cir.
2016); Samsel v. DeSoto County School District, 242 F.Supp.3d 496 (N.D.
Miss. 2017); Pirozzi v. Apple, 913 F. Supp. 2d 840 (N.D. Cal. 2012);
Cornelius v. Delca, 709 F. Supp. 2d 1003 (D. Idaho 2010); Best Western
v. Furber, No. CV-06-1537-PHX-DGC (D. Ariz. September 5, 2008); Energy
Automation Systems v. Xcentric Ventures, Case No. 3:06-1079 (M.D. Tenn.
May. 25, 2007); Hy Cite v. Badbusinessbureau.com, 418 F .Supp. 2d 1142
(D. Ariz. 2005).
---------------------------------------------------------------------------
C. Section 230 Protects Providers Who Engage in Content Moderation, But
Typically Through The Application Of Section 230(c)(1)'s
``Interactive Computer Service'' Provision Not Section
230(c)(2)'s ``Good Samaritan'' Provision.
In our review, only 19 of the 516 court decisions in which Section
230 was raised as a defense were resolved on the basis of Section 230s
(c)(2) ``good Samaritan'' clause, which provides immunity for actions
taken ``voluntarily'' in ``good faith'' to restrict content that is
``obscene, lewd, lascivious, filthy, excessively violent, harassing, or
otherwise objectionable.'' Furthermore, the majority of these cases
involved provider efforts to block spam.\16\ In other such decisions,
courts resolved claims based on Section 230(c)(1),\17\ Anti-SLAPP
motions,\18\ the First Amendment,\19\ or for failure to state a claim
or other deficiencies.\20\
---------------------------------------------------------------------------
\16\ See. e.g., Holomaxx Technologies Corp. v. Yahoo!, Inc., No.
10-cv-04926 JF (PSG) (N.D. Cal. August 23, 2011), E360INSIGHT, LLC v.
Comcast Corp., 546 F.Supp.2d 605 (N.D. Ill. 2008); Pallorium v. Jared,
G036124 (Cal. Ct. App. Jan. 11, 2007); America Online, Inc. v.
GreatDeals. Net, 49 F. Supp. 2d 851 (E.D. Va. 1999).
\17\ See, e.g., DeLima v. YouTube, 2019 WL 1620756 (1st Cir. Apr.
3, 2019); Green v. AOL, 318 F.3d 465 (3d Cir. 2003); King v. Facebook,
3:19-cv-01987 (N.D.Cal. Sept 5, 2019).
\18\ See, e.g., Sikhs for Justice v. Facebook, 144 F. Supp. 3d 1088
(N.D.Cal. 2015); Johnson v. Twitter, No. 18CECG00078 (Cal. Superior Ct.
June 6, 2018).
\19\ See, e.g., Davison v. Facebook, 370 F. Supp. 3d 621 (E.D. Va.
2019); Estavillo v. Sony Computer Entm't Am., 2009 WL 3072887 (N.D.
Cal. Sept. 2, 2009).
\20\ See, e.g., Roberson v. YouTube, 2018 DNH 117 (D. N.H. 2018);
Young v. Facebook, 790 F. Supp. 2d 1110 (N.D. Cal. 2011); Lewis v.
YouTube, No.H041127 (Cal. App. January 25, 2016)
---------------------------------------------------------------------------
Another reason (c)(2) has not been invoked more often is that, when
providers are sued for removing content, many of those lawsuits are
based on assertions that the provider has violated the First Amendment
rights of the user whose content was removed.\21\ As the First
Amendment applies to only government actors, these cases have been
dismissed for failure to state a claim without the necessity of
defendants asserting or a court analyzing Section 230. In fact, courts
have found that service providers' decisions regarding whether and how
to display content are protected by the First Amendment.
---------------------------------------------------------------------------
\21\ See, e.g., Tulsi Now, Inc. v. Google, LLC; Prager University
v. Google LLC, 951 F.3d 991 (9th Cir. 2020); amango v. Facebook, Inc.,
No. 3:11-CV-0435, 2011 WL 1899561 (N.D. NY April 19, 2011); Davison v.
Facebook, Inc., 370 F.Supp.3d 621 (E.D. Va. 2019); Federal Agency of
News LLC v. Facebook, Inc., 2020 WL 137154 (N.D. Cal. Jan. 13, 2020;
Zhang v. Baidu. com Inc., 932 F. Supp. 2d 561 (S.D.N.Y. 2013); Buza v.
Yahoo!, Inc.,, No. C 11-4422 RS (N.D. Cal. 2011).
---------------------------------------------------------------------------
IV. Considerations For Policymakers
It is of the utmost importance that policymakers tread carefully
when considering possible changes to Section 230 or enacting any other
laws targeting content moderation. This caution is necessary because of
the ever-evolving nature of harmful content and Internet technology, as
well as the complexity and variety of potential legal liability for
online speech. This caution is also essential in light of foundational
First Amendment principles.
A. Maintaining The Careful Balance Struck By Section 230.
Section 230, in its current form, supports a diverse Internet
ecosystem that provides users with reviews, places for discussion and
lively debate, and opportunities to expand their knowledge. Without
Section 230s protection, Internet companies would be left with a strong
disincentive to monitor and moderate content. Section 230 removes this
disincentive to self-regulate, creating essential breathing space for
Internet companies to adopt policies and deploy technologies to
identify and combat objectionable or unlawful content--or to develop
other innovative solutions to address such content. Society benefits
from the rules that providers voluntarily set and enforce to enhance
user experiences as well as safety, goals that would be challenging--if
not impossible--for the government to achieve directly due to the First
Amendment. Through exceptions and carefully crafted language limiting
Section 230s protections to only third-party content and activities,
bad actors can still be held accountable when they participate in, or
materially contribute to, illegality.\22\ Over the more than two
decades since Section 230s enactment, the Internet continues to thrive
due to the carefully crafted language balancing the fostering of online
innovation with ensuring there are proper ways to hold content
providers accountable for their actions.
---------------------------------------------------------------------------
\22\ Fair Housing Council of San Fernando Valley v. Roommate.com,
521 F.3d 1157, 1168 (9th Cir. 2008) (en banc). In addition, Section
230(e) outlines the criminal law, intellectual property, state law,
communications privacy law, and sex trafficking law exemptions from 230
immunity.
---------------------------------------------------------------------------
B. Ensuring That Any New Requirements Recognize The Flexibility
Required For Effective Content Moderation.
Some proposals to change Section 230 have the potential to impact
how service providers conduct content moderation by limiting the
protections in the statute to just certain types of content, or setting
new rules for how companies engage in content moderation activities.
Policymakers considering making changes to Section 230 must recognize
the wide cross-section of online services that rely on the law and keep
in mind the need for flexible and non-prescriptive language. There are
important reasons why the broad group of entities and individuals who
qualify as providers of interactive computer services should be able to
retain discretion and flexibility to set and enforce their rules. For
example, content moderation teams should be encouraged to be nimble
enough to respond to unanticipated events quickly. The urgent nature of
the response to the video of the Christchurch attack is a good example
of how world events can impact content moderation efforts. The
circumstances of the Christchurch attack are precisely why providers
need to be able to make adjustments to the techniques they use to
battle policy violations to adapt alongside the ever-evolving nature of
threats. Imposing overly prescriptive and burdensome requirements
through legislation or regulations will negatively impact the Internet
ecosystem. Without flexibility, service providers are unable to
effectively moderate content on their platforms, which could
dramatically reduce the quality of their services. Furthermore, online
platforms are not uniform in their breadth, construction, business
models, or approach to content hosting. Changes to Section 230 intended
to address concerns with a particular platform or type of platform,
will impact all platforms. Given the discrete but important differences
among Internet platforms, changes to Section 230 must carefully
consider the broad and varied impacts of legislative language on
different platform models.
C. Aligning With Established First Amendment Principles That Apply To
Content Moderation.
Any amendments to Section 230, and any other laws pertaining to
content moderation, should take careful account of three First
Amendment guardrails.
First, platforms are not state actors and consequently need not
refrain from moderating speech protected by the First Amendment. The
First Amendment only limits the actions of state actors--that is,
governmental entities--not private companies merely because those
companies provide forums for speech. Courts have consistently held that
Internet platforms are not state actors bound to follow the strictures
of the First Amendment.\23\ Plaintiffs cannot bring suit against
platforms alleging that the platforms somehow violated the plaintiffs'
right to express particular speech under the First Amendment. Some have
suggested that social media sites should be treated as public forums
subject to First Amendment restrictions. However, most users would not
want the First Amendment to dictate Internet platforms' content
moderation practices as though they were state actors. If that were to
happen, platforms would be prevented from blocking or screening a wide-
range of problematic content that courts have held to be
constitutionally protected including pornography, hate speech, and
depictions of violence.
---------------------------------------------------------------------------
\23\ See, e.g., Freedom Watch, Inc. v. Google Inc., 2020 WL
3096365, at *1 (D.C. Cir. May 27, 2020) (per curiam); Prager Univ. v.
Google LLC, 951 F.3d 991, 996-999 (9th Cir. 2020).
---------------------------------------------------------------------------
Second, the First Amendment protects the rights of the platforms
themselves. When platforms determine what kind of platform to be and
what kinds of content to host or prohibit, those are forms of free
expression protected by the First Amendment. It is bedrock First
Amendment doctrine that such editorial decision-making is
constitutionally protected. In Miami Herald Publishing Co. v.
Tornillo,\24\ for instance, the Supreme Court held that a statute
requiring newspapers to provide political candidates with a right of
reply to critical editorials violated the newspaper's First Amendment
right to exercise ``editorial control and judgment'' in deciding the
``content of the paper.'' \25\ Several courts have applied this
reasoning in the online context, holding that platforms possess the
First Amendment right to decide what content to carry.\26\ Recognizing
this principle has never been more important. It is critical to
allowing online communities and services to develop around common
interests, shared beliefs, and specific purposes. It is also critical
to allowing online services to cater to different audiences, including
the ability to design rules to make their services age-appropriate or
purpose-appropriate.
---------------------------------------------------------------------------
\24\ 418 U.S. 241 (1974).
\25\ Id. at 258.
\26\ See, e.g., Jian Zhang v. Baidu.com Inc., 10 F. Supp. 3d 433,
436-443 (S.D.N.Y. 2014); Langdon v. Google, Inc., 474 F. Supp. 2d 622,
629-630 (D. Del. 2007).
---------------------------------------------------------------------------
Third, the First Amendment sets a constitutional floor that ensures
that online platforms that carry vast quantities of third-party content
cannot be held liable for harms arising from that content based on a
standard of strict liability or mere negligence. Applying such non-
protective standards of liability to entities that distribute large
volumes of third-party material would violate bedrock First Amendment
principles. The Supreme Court examined this issue over six decades ago,
in Smith v. California.\27\ There, a city ordinance prohibited
bookstores from selling obscene or indecent books regardless of whether
the store owners knew the books were obscene or indecent.\28\ The
ordinance violated the First Amendment, the Court explained, because it
would cause a bookseller to ``restrict the books he sells to those he
has inspected'' and thus ``impose a severe limitation on the public's
access to constitutionally protected matter.'' \29\ This principle--
that the First Amendment gives special protection to those who act as
clearinghouses for large quantities of third-party content--applies
with especially great force to Internet platforms, given the
exponentially greater volumes of content that they host and the
important role they play in societal discourse. Were these platforms to
face liability for distributing unlawful third-party material absent
circumstances in which they both knew of that particular content and
yet failed to remove it, Internet users' access to vital
constitutionally protected speech would be severely stifled.
---------------------------------------------------------------------------
\27\ 361 U.S. 147 (1959).
\28\ Id. at 148-149.
\29\ Id. at 153.
---------------------------------------------------------------------------
This trio of First Amendment principles provides important
constitutional guardrails that protect free expression on the internet.
Along with Section 230, they have contributed to making the Internet a
vibrant medium that benefits so many. Policymakers addressing content
moderation must therefore carefully consider the interaction between
these principles and new policies before enacting new laws that could
threaten to undermine the constitutional foundation of our dynamic
internet.
V. The PACT Act
Given the many crucial considerations implicated by any proposal to
amend Section 230, IA appreciates the thoughtful approach taken by
Senators Schatz and Thune in the ``Platform Accountability and Consumer
Transparency Act'' or the ``PACT Act.'' IA and its member companies
appreciate the focus in the bill on the twin goals of promoting
transparency and accountability in content moderation. Over the last
several years, IA member companies have been working continuously to
enhance transparency with their users and the public about their
community rules; how they are enforced; and how often they are
enforced. These efforts include expanding transparency reporting to
cover a wider range of topics including content removals for terms of
service violations, making the rules of the service easier to
understand and more detailed, providing additional user education and
guidance through examples of potential rule violations, and explaining
in more detail how rules are enforced and the potential consequences
for violations. These efforts are just one part of how IA member
companies approach content moderation and supplement measures such as
easy reporting of violating content, user notices and appeals, and
proactive efforts to find violating content. The PACT Act's focus on
these aspects of content moderation in many ways align with IA member
company efforts, and for that reason, IA hopes to work with the
sponsors to ensure that the bill is able to achieve its goals without
inhibiting flexibility and innovation in content moderation.
IA would like to highlight two areas of concern related to the
bill's broad scope and highly detailed requirements. With regard to the
bill's scope, the requirements for transparency and accountability in
Section 5, as drafted, would apply to all ``interactive computer
services'' (ICSs), which is essentially the same group of entities that
benefit from the protections of Section 230. As discussed above,
Section 230 applies to a wide-range of interactive services and
platform models including those offered by individuals, informal clubs
or groups, and member associations. In addition, the types of services
that fit within the term ``interactive computer service'' are likewise
broad, covering not only social media services, but also private
messaging, search, message boards and listservs, dating services, job
search platforms, review sites, and more. IA is concerned that the
requirements of Section 5 would prove too large a burden for those ICSs
as hobbyists, volunteers, and as adjuncts to other activities. Such
requirements may force these individuals and entities to shut down
their projects and may discourage similarly situated individuals and
groups from engaging in important expressive activity. There are also
services for which transparency requirements may not make sense given
the type of service or the purpose for which it was created. For
example, under Section 5, review platforms would have to disclose to
fraudsters (e.g., rival businesses or competitors of the business being
reviewed) that their fake reviews had been detected and give fraudsters
an opportunity to appeal the takedown of their fake review. While
transparency is often a benefit, the burdens associated with these
particular requirements should be carefully weighed against the
benefits they would likely achieve. For this reason, IA hopes to work
with the sponsors to consider potential changes to the scope of the
bill.
The other potential issue IA would like to share regarding the bill
relates to the negative ramifications the highly detailed requirements
in Section 5 could have. First, the detailed nature of the requirements
would be extremely burdensome for all ICSs, and it would be a struggle
for all but the most highly resourced providers to comply. For example,
in order to comply with transparency reporting requirements, providers
would need to rebuild the systems they use to process and track user
reports, as well as any systems that operate independently to moderate
content, to ensure that all of the required types of information are
collected for future reports. Absent a long window to ramp up,
providers may need to manually review each individual report to collect
information for backwards-looking reports. The difficulty of complying
could adversely impact content moderation as providers may choose to
narrow their content policies to limit the scope of issues that would
have to be addressed by the requirements of Section 5. Therefore, these
requirements could unintentionally result in less moderation, rather
than more.
As with the scope of the bill, the Section 5 requirements would
likely limit the diversity and richness of the different types of
individuals and entities that are part of the online ecosystem. This in
turn would limit consumer choice and access to information. Small
providers in particular would feel the impact of these requirements. In
addition, the detailed nature of the requirements would significantly
diminish the essential flexibility providers have today to constantly
adjust their approaches to content moderation to keep pace with bad
actors, respond to emergencies, and focus their efforts on the
activities that pose the highest risks to users of their services and
the public.
IA appreciates the opportunity to discuss the value of Section 230
to the modern Internet and looks forward to continuing these
conversations around our concerns and other feedback pertaining to this
bill with the members of the Subcommittee.
APPENDIX: Internet Association, A Review Of Section 230's Meaning &
Application Based On More Than 500 Cases (July 27, 2020).
Senator Thune. Thank you, Ms. Banker.
Next up is Mr. Sylvain.
Mr. Sylvain, please proceed.
STATEMENT OF OLIVIER SYLVAIN, PROFESSOR,
FORDHAM LAW SCHOOL
Mr. Sylvain. Thank you, Committee Chairman Wicker, Ranking
Member Cantwell, Subcommittee Chairman Thune, Ranking
Subcommittee Member Schatz, members of the Senate Subcommittee
on Communications, Technology Innovation, and the Internet.
As vexing as the problems posed by Section 230 are, I am
honored to discuss potential reforms of the statute with you
today. Let me get straight to it.
Section 230 is not adapted to the ways in which many online
intermediaries control practically all aspects of consumers'
online experiences. The doctrine presumes that any given
interactive computer service is no more than a publisher or
distributor of user-generated content, unless that service
materially contributes to the content at issue. Sir, courts
have been generous to intermediaries under this rule, generally
dismissing claims before discovery begins. Courts and
plaintiffs, accordingly, never really get to learn how
implicated intermediaries are in their distribution and
targeted delivery of either the user content or data.
The most powerful online intermediaries today are anything
but publishers and distributors of user-generated content. They
illicit, harvest, sort, and repurpose user posts and personal
data to attract and hold consumer attention and, more
importantly, to market this valuable data to advertisers. They
do this unfettered by law. The result is too often lived harm
to the everyday people whom consumer protection and civil
rights laws and remedies exist, but, because of the protection,
are unavailable.
Recent civil rights litigation against Facebook's Ad
Manager provides a vivid example of this unfair and
underregulated political economy for user data. The social
media giant relies on sophisticated, but imperfect, automated
decisionmaking tools to make sense of the troves of consumer
data that it collects. Through these processes, it creates
hundreds of new categories in which advertisers may, as they
wish, refine campaigns by including or excluding potential
recipients across hundreds of dimensions. For example,
distributors of Urdu-language music or WNDA gear can target
specific audiences by gender, age, ethnicity, and language, to
list just a few variables. Facebook generates these categories
through automated algorithmic processes. Consumers, meanwhile,
generally have no idea where they fall in the scheme. Facebook,
for example, does not ask its consumers to identify with any
ethnicity, and yet it slots consumers into proxies for these
identities.
Here's the rub. Under civil rights law, Congress forbids
discrimination in ads on the basis of race, ethnicity, age, and
gender in the markets for housing, education, and consumer
credit. But, that is exactly what Facebook allowed building
managers and employers to do. A series of blockbuster stories
by ProPublica in late 2016 revealed that the service-enabled
advertisers, to include and exclude consumers by proxies for
race and age, and consumers were not the wiser for it, because
they do not see each other's news feed.
Soon after the ProPublica stories ran, civil rights groups
and aggrieved plaintiffs brought cases against Facebook in
civil and in Federal courts across the country. They alleged
that the Ad Manager made discrimination in violation of civil
rights laws possible. They also asserted that Facebook targeted
ads, on behalf of the advertisers, in ways that were
discriminatory. Through its lookalike feature, in particular,
Facebook replicates any given advertiser's customer list, and,
based on that list, delivers the ads far more widely to people
who fit the profile. With regards to these recent civil rights
cases against Facebook, the lookalike feature entrenched extant
disparities at the expense of members of historically
marginalized groups. None of this resembles the news groups and
bulletin boards that Congress had in mind in 1996.
The social media giant, nevertheless, moved to have these
civil rights claims dismissed, pursuant to Section 230, as it
routinely and formulaically does in practically all cases. And
most of the time, they win. But, here the courts never got a
chance to weigh in, because the parties settled in March 2019.
Among other things, Facebook agreed to create a new ad portal
for housing, employment, and credit markets to protect against
unlawful discrimination in those settings. HUD filed a charge
against the company soon afterwards. And, as far as I know,
that action remains unresolved.
Importantly, however, researchers from Northeastern
University, Upturn, and ProPublica late last year found that,
in spite of the settlement, Facebook's Ad Manager continues to
distribute job ads that discriminate against women and older
people at alarming rates. That is, even as Facebook no longer
allows advertisers, in the first instance, to use unlawful
proxies in hiring, the Ad Manager nevertheless continues to
discriminate.
This is just one example in one area of law, but it
underscores that Section 230 has done nothing to cultivate any
demonstrable urgency about protecting consumers in this setting
even as intermediaries milk that data for ad revenue.
The PACT Act is a good start at reform. The one piece I
will emphasize here, with the limited time that I have, is the
expansion of the exceptions under 230(e) to include civil
government agency enforcement. I would like to see this
expanded to include claims by private individuals under Federal
and State law, but, even in its current form, the bill would
allow courts to scrutinize intermediaries' practices more
closely. This would be all the more important in cases in which
plaintiffs allege that automated decisionmaking systems cause
harm. I can see agencies like the FTC and HUD proceeding
pursuant to their statutory mandate in this regard.
Much of the PACT Act would represent the sensible reform
that would help to vindicate consumer protection, civil rights,
and public law protections for everyday people.
Thanks again for the invitation. I look forward to your
questions.
[The prepared statement of Mr. Sylvain follows:]
Prepared Statement of Olivier Sylvain, Professor, Fordham Law School
Committee Chairman Wicker, Ranking Member Cantwell, Subcommittee
Chairman Thune, Ranking Subcommittee Member Schatz. Members of the
Senate Subcommittee on Communications, Technology, Innovation, and the
Internet,
Thank you for inviting me to today's hearing about reforming 47
U.S.C. Sec. 230. I have been thinking and writing about this provision
and its attendant judge-made doctrine for the past few years. I have
argued that the law is not adapted to the ways in which online
intermediaries today effectively control practically all aspects of
consumers' online experiences. The so-called ``immunity'' under Section
230, developed in the late 1990s, continues to presume that online
intermediaries (or ``interactive computer services'') are no more than
than mere publishers or distributors of user-generated content. The
courts have read the provision as a shield from liability unless a
plaintiff in any given case successfully pleads that a defendant
intermediary ``materially contributes'' to the content at issue.
Pursuant to this standard, courts today have been unwilling to see
anything but glaring direct contributions of substantive content as
counting as ``material contribution.'' In 2020, this presumption--this
benefit of the doubt--makes no sense. The biggest of these companies
today design their services to elicit, collect, harvest, sort, analyze,
redistribute, and altogether repurpose their consumers' data in service
of their business objectives and the interests of the advertisers who
have come to rely on them. And they use sophisticated but demonstrably
imperfect automated decisionmaking systems to do this. (They actually
purport that these systems, even as they structure the entirety of our
online experiences, help them to moderate user content.)
None of this--especially the companies' pecuniary designs on user
data--resembles the considerations at work when Congress enacted 47
U.S.C. Sec. 230. The relatively quaint and romantic motivations of the
Usenet newsgroups or AOL bulletin boards were on the minds of
legislators or judges in the 1990s. They could not foresee Big Tech's
industrial designs on controlling and consolidating information flows
to achieve their own commercial objectives.
Nor does the statute's titular claim that online intermediaries'
``Protection for `Good Samaritan' blocking and screening of offensive
material'' do any real work. One could reasonably read Section
230(c)(2) as the operative ``safe harbor'' under the statute, in which
case courts would only immunize ``interactive computer services'' that
voluntarily take good faith steps to moderate illicit or illegal user-
generated content. But, in practice, that is not what defendants, Big
Tech companies, or their advocates have asserted. Rather, they have
projected 230(c)(1)'s passive-voice and indirect mandate about how
providers of ``interactive computer services'' should be ``treated''
onto Section 230(c)(2) in ways that effectively overshadow and
essentially eliminate the mechanism for courts to consider whether an
online intermediary is reasonably trying to moderate content. Because
of this doctrine, defendants raise the Section 230 defense at the
motion to dismiss phase, well before plaintiffs and courts ever get to
find out how implicated intermediaries are in their control and
administration of user content or data.
It is time that the law and doctrine reflect our currently reality.
This is why I am honored and eager to engage your consideration of the
PACT Act and other reforms of the statute. For what it is worth, the
following are, in reverse chronological order, the recent pieces I have
written on the topic:
Solve the Underlying Problem: Treat Social Medias as Ad-
Driven Companies, Not Speech Platforms, Knight Foundation (June
16, 2020)
Recovering Tech's Humanity, 119 Columbia Law Review Forum
252 (November 2019)
A Watchful Eye on Facebook's Advertising Practices, N.Y.
Times (March 28, 2019)
Discriminatory Designs on User Data, Knight First Amendment
Institute ``Emerging Threats Series'' (April 2018)
Intermediary Design Duties, 50 Connecticut Law Review 203
(March 2018)
AOL v. Zeran: The Cyberlibertarian Hack of Sec. 230 Has Run
Its Course, Law.com (November 2017)
My views of the prevailing doctrine precede the emergent du jour
argument that social media companies have a liberal coastal urban bias.
I will answer inquiries from you on this question if you have them, of
course, but, at the outset, please know that I do not believe that
online intermediaries' editorial moderation decisions are unlawful or
even imprudent. As I have written elsewhere, Facebook and Twitter, for
example, have sensibly developed tools that enable their users to
control the ways in which trolls and bigots slide into online
``conversations'' and user-groups. They have used their
constitutionally protected editorial prerogative to flag user content
that they find hateful or dangerously misleading. The principal
question I have in this context is whether these efforts are enough,
since illegal content and advertisements continue to proliferate on
their services.
More pertinently, I believe that framing the question of Section
230 reform in terms of political viewpoint or even free speech obscures
what is truly at work: the political economy of online advertising. The
biggest and most popular online intermediaries today are not simple
``platforms'' for user-generated content as much as commercial services
for targeted advertising to consumers. These companies design their
applications and the automated decisionmaking systems that power them
to maximize advertising revenues. Social media companies in particular
are keenly committed to designing services and products that keep users
viscerally engaged in service of their bottom-line. We are well past
the discounts and coupons that retail chains include in their
circulars.
The current debate about Section 230 should focus instead on Big
Tech's unprecedented power to control consumer behavior. The
beneficiaries of the protection under Section 230 are not in the
business of promoting free speech as much as designing services that
optimize user engagement which, in turn, maximizes the scope and depth
of their advertising revenue. They do this more or less unmoored by
settled legal conventions because of the broad protection under Section
230, a doctrinal protection that I do not think any other species of
company in the United States has ever enjoyed.
To be sure, their pecuniary motivation, unfettered by the threat of
liability under the courts' broad reading of the protection under
Section 230, has allowed an array of innovative applications for user-
generated content to proliferate. But the current legal protection
under Section 230 has also cultivated in application developers a cool,
above-it-all indifference to (1) public law norms and (2) the immediate
lived harms that so much of the content and data that they distribute
causes. Dangerously misleading public health related information,
disinformation about elections, nonconsensual pornography, and
discriminatory advertising, all of which may be illegal in any given
circumstance, proliferate still because online intermediaries do not
have to bear the responsibility for designing systems that carefully
distribute them. The question for you is whether there is something
legislation can do to cultivate and engender a demonstrable sense of
social responsibility.
If you were to ask me to make any recommendations today about
Section 230 reform, it would first be that courts should read the
protection for interactive computer services far more carefully than
they have. We have seen slow but steady improvement on that front since
the Ninth Circuit's decisions in Fair Housing Council of San Fernando
Valley v. Roommates.com in 2008 and Barnes v. Yahoo! in 2009 and, more
recently, perhaps, after the Second Circuit's decision last summer in
Oberdorf v. Amazon. My humble recommendation to courts is that they
should be far more searching than they have been in determining whether
a defendant interactive computer services' designs materially
contribute to the distribution of illegal content. At a minimum,
opening the standard up in this way would allow plaintiffs to engage in
discovery on colorable claims--a prerogative that litigants in other
legislative fields generally have. Today, most Section 230 defenses are
decided at the motion to dismiss, before any discovery can be had. I
urge courts to be far more open to the pleaded claims that online
intermediaries' designs materially contribute to illegality. But this
is for the courts to sort out under current doctrine.
My humble recommendation to you, as legislators, must be different,
as you bring a different, more generalizable and prospective
institutional authority: the exceptional legal protection that online
intermediaries now enjoy under the statute is ripe for narrowing
because, today, it directly causes consumer harms and sometimes
entrenches racism, misogyny, and discrimination against members of
historically marginalized groups. Public laws and regulations exist to
protect people from these kinds of injuries. But, because of the
prevailing doctrine, the entities most responsible and capable of
protecting people bear no legal responsibility to do so.
Thanks again for the generous invitation to testify. I look forward
to engaging your questions as best I can.
Senator Thune. Thank you, Mr. Sylvain.
Well, we'll dive right into questions.
Representative Cox, one of the most common misconceptions
about CDA 230 is that it requires platforms to be politically
neutral with respect to its content moderation decisions. Does
CDA 230 require platforms to be politically neutral? And, if
not, why not?
Mr. Cox. No, it does not. Section 230 was never meant to
require neutrality. To the contrary, the idea is that every
website should be free to come up with its own approach, to let
1,000 followers bloom. If it were otherwise, we could imagine,
you know, how, you know, very quickly things would not work.
The Democratic National Committee could not have its own
website. The Republican National Committee could not. It's not
a question of everybody having to be neutral. It is, however,
important to distinguish that it is possible for websites to
decide that they want to have a business model of political
neutrality. And when they do so, they can and should be held to
that. If they promise this to the public through their terms of
service, through their advertising, through their, you know,
rules of content moderation that are published, that ``We will
do this,'' then, you know, by all means, hold them to it. And I
think, you know, many existing laws are available for this
purpose, including the Federal Trade Commission Act and the
little FTC Acts in all of the States. There is a well-known
case in the Ninth Circuit that used the common law promissory
estoppel to require Yahoo! to honor promises to take down
material. And so, it is in this framework that I think we
should look at political neutrality. If these platforms are
holding themselves out as neutral, then they should be held
accountable.
Senator Thune. For those who are concerned about systemic
viewpoint discrimination by the content moderation decisions of
platforms, wouldn't rigorous transparency requirements for
platforms help to reveal whether there is, in fact, systemic
viewpoint discrimination on a given platform?
Mr. Cox. Yes. I'm a big fan of transparency. I think the
devil always is in the details. The question is, you know, How
steep is the compliance burden, and is it, you know, such that
it will, you know, expose platforms to, you know, liability
that they will need to structure around and try to protect
themselves from, which could have some unintended consequences
for the availability to all of us of user-generated content.
But, the ambition of using the law and all the other tools that
we have at our disposal to encourage and get, where
transparency is a good one, you can imagine, for example, that
if the rules of the road were much more explicit and
understandable in every case, and if, when decisions were made,
for example, to cancel somebody's account or to take down
material, that the decision was directly tied to, you know,
specific provisions in those rules that people could
understand, and the decisionmaking was publicly available so we
could all learn from it, we would all be, you know, more likely
to abide by those terms of conduct online, because we would
understand them. I hold we would be better off, it would be
achieving its aims more successfully, and then, hopefully in
the process, the Internet would be a little more civil.
Senator Thune. And you stated in your testimony that the
PACT Act's content moderation transparency component, and as I
quote from your testimony, ``unquestionably constructive and
consistent with Section 230 in its ultimate aims,'' end quote.
What content moderation transparency requirements do you think
are most important to apply to Internet platforms? And how do
transparency requirements benefit consumers?
Mr. Cox. Well, just further to the points I was making, you
know, consumers sometimes operate in the dark. And when
takedown decisions are made and then people get up in arms
about it and talk about it on the Internet--we can all see what
people are saying--you know, a lot of the complaint is, ``What
the hell are they doing? What are they thinking? Their terms of
service say X, but they did Y,'' and so on. So, the more
information that is provided so that people can understand
decisionmaking, the better off we're all going to be.
In my written testimony, I've provided some extensive
commentary on very specific aspects of the PACT Act that bear
on this question. I think that, you know, ultimately what you
need to focus on and be concerned with is, How is this going to
operate in the real world? Is it actually doable? Is it
workable? Are, you know, the data-collection requirements or
the reporting requirements going to be such that they might
actually interfere with the realtime aspect of Internet
communications who don't wish to do that, because that's an
essential feature of how the Internet operates and what its
great benefits are? You know, is it going to put platforms in a
position where the liability that we thought we protected them
from for using user-generated content just, you know,
evaporates? You know, we don't want to be in that position,
either, because we know what they would do. Many of them will
feel the need to jettison user-generated content, or to vastly
restrict it.
So, those are the guardrails that I think we want to
observe. The objectives are equally clear.
I just don't see what's in it for platforms to be opaque.
It's not in their interest to do so. And I think that many of
them have been making big steps in this regard. I also think
it's important that more resources be devoted to this. The
bigger platform, the more resources we can expect to be devoted
to this.
Senator Thune. On that----
Mr. Cox. And I would----
Senator Thune. On that----
Mr. Cox. Yes, go ahead.
Senator Thune. On that point, I would turn to Ms. Banker,
because--you made this point, Representative Cox, but, Ms.
Banker, you noted that IA's member companies have been working
to enhance content moderation transparency with their users.
Today is there a minimum transparent requirement platforms have
to meet? And it seems like, as Representative Cox noted, it
would be in their interest not to be opaque, to be as
transparent as possible. What's happening in that space right
now?
Ms. Banker. Thank you, Senator.
IA member companies are voluntarily enhancing their
transparency around content moderation activities. The--
Internet companies actually have a long history of being
transparent. Early transparency reports generally focused on
topics like the disclosure of user information to law
enforcement. But, in recent years, the companies have been
working steadily to expand that into content moderation topics,
as well. There is not a minimum standard. And, you know, we
appreciate the focus of the PACT Act on this area, but do want
to make sure that, should new requirements be put in place,
that they're requirements that all of IA members would be able
to fulfill. And we represent both large and small companies and
a multitude of business models that are other than social
media, which tends to be the focus of most of this discussion.
Senator Thune. Yes. Thanks.
Senator Schatz.
Senator Schatz. Thank you, Mr. Chairman.
I'm going to start with Professor Sylvain. You went very
quickly on a very important point, and I want you to flesh it
out a little bit.
Can you describe exactly what happens when a company
invokes Section 230 and tries to win on summary judgment, avoid
discovery, avoid a lawsuit? Regardless of how the statute,
whether it be in civil rights or the extension of credit or
education or housing, that statute would normally carry the
day, or at least get a plaintiff a day in court. But, with
Section 230, they just invoke it, and there's--and it's the end
of the conversation. Could you flesh that out a little bit?
Mr. Sylvain. Thank you for the question, Senator Schatz.
And just to be clear, the way--the moment it's invoked in
litigation is at the motion to dismiss, generally, not in a
summary judgment. At summary judgment, there will be--you know,
they will have--discovery will have--some discovery will have
happened already. It's invoked at the motion to dismiss, before
any discovery. And what the courts are doing, as you say, is,
they're not reviewing the substantive statute under which any
plaintiff is alleging some harm. What they're doing is
evaluating whether the defendant is--meets the qualifications
of the safe harbor, or the--as people call it, the immunity
under Section 230----
Senator Schatz. In other words, whether they're online.
Mr. Sylvain. Whether--that's right--whether they're
interacted computer service, which is, basically, whether
they're online.
And then, the further question, which is a point that
Jeff--Professor Kosseff started with, is, they evaluate--you
know, moving away from the old case law on this, evaluate
whether they're being treated as a publisher or distributor
in--under the claims--under the legal theory brought by
plaintiffs. And if they are, that effectively shuts down the
litigation. And the things that count as publishing are pretty
expansive, so I invoked the advertising case--the Ad Manager
case, because it underscores how broad this claim is. It
includes the sorts of things that social media companies do
when they distribute content. It might also include
advertising. Indeed, it does include advertising. Right? So--
and I think----
Senator Schatz. But, would it matter--I'm sorry to
interrupt, but would it matter whether the platform was
intentionally violating a--say, a civil rights statute or a
housing statute or a banking statute? Or would that Section 230
immunity sort of obviate the inquiry into the question of
whether it was sort of like them hatching a plan or whether or
not they were just simply allowing it to happen as a result of
their systems and algorithms?
Mr. Sylvain. That's a really good question. Could a company
that intentionally wanted to perpetuate some harm under public
law do so, and be comfortable doing so because they would be
treated as a publisher in--a distributor under a claim, and
would never--we'd never find out what their intent is? I think
that would be--I mean, I would be curious to see what that case
looks like--I think it would be hard to imagine, if they
weren't also content--contributing content in that
circumstance, if they weren't also creating and developing
content in that circumstance. But, I suppose you might envision
such a thing. And the closest I can think of is the Jones case,
wherein the Sixth Circuit reviewed a plaintiff's argument--this
young woman who alleged that she was being defamed by a site
that elicited comments that were abusive about young women.
And, you know, the--one of the things the Sixth Circuit is
entertaining is whether another standard, not the material
contribution standard, would be relevant. But, in any case,
that comes close to the scenario you've just mapped out. The
Section 230 defense was victorious in that case.
Senator Schatz. In other words, we'll probably never find
out what was in the minds of the people who were the--whether
it's de facto discrimination or it was intentional. We would
just not--not know, because you wouldn't get past the motion to
dismiss.
I have a question for all of the panelists. Everybody
understands, who's participating in this hearing, that a series
of rulings have established that platforms are not obligated to
remove or address illegal content, a 1997 decision and then a
2010 decision. And so, the question for the panel--I'll start
with Mr. Sylvain, followed by Ms. Banker, Mr. Cox, and Mr.--and
Professor Kosseff. Do you think the platform should be required
to remove content that has been found to be illegal?
Mr. Sylvain, yes or no?
Mr. Sylvain. Yes.
Senator Schatz. Ms. Banker?
Ms. Banker. I think that, much like in the PACT Act, you've
made provisions for that. I think that's a fruitful area of
inquiry, but we'd want to make sure that there are safeguards--
--
Senator Schatz. What does that mean? I'm sorry. It's a yes-
or-no question. Should illegal content--we're talking about not
supposing that something is illegal. A court determines that
this is illegal content, and you represent the Internet
Association here. Should there be a statutory requirement that
illegal content--court-determined illegal content be removed
from websites?
Ms. Banker. Our companies have no interest in hosting known
illegal content, and most of them would remove it voluntarily.
But, I think, if there is a requirement, we'd just want to be
sure that there is sufficient guardrails to make sure that the
court engaged in a thorough review. Unfortunately, we've also
seen why--the variety of instances where these types of
mechanisms have been subject to abuse.
Senator Schatz. Where a court has determined that content
is illegal, and somehow we need additional guardrails? I mean,
it's not like that happens all the time, right?
Ms. Banker. No, but it--frequently, plaintiffs do go into
court seeking that type of ruling, with an eye to having
content taken down. Unfortunately, not all of those cases, you
know, are brought in good faith. We are lucky. Some states have
tools, like anti-SLAPP Act provisions, that can be used to make
sure that individuals who are exercising the right to comment
on matters of public interest are able to do so. But, not in
every case is a defendant able to show up and defend themselves
in court. So, we just want to make sure that, you know, to the
extent we are covering things like defamatory content, that
there are safeguards.
Senator Schatz. Congressman Cox.
Mr. Cox. Yes, I think that a well-crafted statute could do
a lot of good here. I see no reason that court order and
certainly final judgments requiring the takedown of content
already adjudged to be defamatory, for example, couldn't be
enforced. If the law were to provide clear standards for
platforms, telling them how they should handle defamation
judgments, just to use the paradigm situation, especially in
cases in which the platform hasn't been sued, such as is the
case in the California Supreme Court with Hassell against
Bird--I think that was 2016--and the platform isn't the party
to the litigation, and so it's essentially succeeding in
keeping these disputes out of its hair without fear of
liability. This is consistent with the aims of Section 230----
Senator Schatz. Thank you.
Mr. Cox.--injured party would receive justice. That's a
positive outcome for everybody except the fellow who broke the
law and lost in court. So, it's exactly the way Section 230
should work.
Senator Schatz. Professor Kosseff.
Mr. Kosseff. Yes, absolutely, if something has been
adjudicated to be illegal, I think it should be taken down,
just with the caveat that there have been cases of court orders
being falsified because there are many platforms that already
honor the court order. So, we want to just make sure we have a
provision to take care of that.
And I would also add that, in addition to defamation
judgments, we might want to look at, also, family court
judgments, because defamation can be a very expensive process
to go through, and, in some of the most harmful content--
stalking, harassment--people are not going through the full
defamation process. So, I think we'd want to look at that, as
well.
Senator Schatz. Thank you very much.
Senator Thune. Thank you, Senator Schatz.
Senator Wicker.
The Chairman. Thank you very much.
Very, very good panel. Very helpful.
Critics of the Communications Decency Act argue that the
words ``otherwise objectionable'' allow social media companies
to remove content they simply dislike or disagree with, without
facing liability. So, that raises the question how social media
platforms inform their users about what constitutes ``otherwise
objectionable content.''
As I mentioned in my opening statement, the content that is
available to be removed already includes ``obscene, lewd,
lascivious, filthy, excessively violent, harassing,'' those
words. So, the question is, Would it be helpful if we removed
the somewhat ambiguous term ``otherwise objectionable''?
So, Mr. Cox, let me start with the very beginning. What
does ``otherwise objectionable'' mean to you, as an author? And
what was the intent behind the inclusion of that term in the
original statute?
Mr. Cox. Of course, what the statute says is, you know,
most important, and that's the standard to which people need to
conform their content, that's what vendors need to interpret,
and so on. I think what I intended, what Ron intended when
originally we wrote this, and what Congress intended when it
considered the legislation and voted for it, were all
consistent with how we would want it to work. And that is that
this string of words that's followed by ``otherwise
objectionable,'' you know, has to be read as a whole. It's a
well-established rule of statutory interpretation that when
general words follow specific words in a statutory enumeration,
the general words are construed to embrace only things similar
in nature. It's a legal rule known as ``ejusdem generis,''
which is Latin for ``of the same kind.'' It's a rule of long
standing that's been reaffirmed by the U.S. Supreme Court in
the 21st century. So, the words ``otherwise objectionable''
have to be understood with reference to the list of specific
things that precedes them. It's not an open-ended grant of
immunity for editing content for any unrelated reason a website
can think of.
But, it's necessary not to be too stingy in, you know,
setting out in the statute, you know, what is covered, because
there are a lot of really bad things we'd like to see taken
down from the Internet, some of them maybe not invented yet,
that would be, you know, just as objectionable and just as, you
know, categorically fitting as what we've listed specifically.
So, we don't want to deprive websites the opportunity, you
know, to keep their sites clean and civil and so on. But,
that's the rubric. You know, it's--it doesn't take you far
afield into anything that I personally idiosyncratically as a
website don't like. That clearly is not what the statute says.
And one hopes----
The Chairman. What--Mr. Cox, what about it--political
speech? Does ``otherwise objectionable content'' include
political speech?
Mr. Cox. Well, you know, political speech sounds like a
broad category, but it's easy enough to imagine that somebody
decides to take something down because they disagree with it.
You know, ``You say that taxes should be higher, I say they
should be lower. I don't like what you said, and I'm taking it
down.'' You know, that's not a good reason to invoke Section
230. That's not what that Good Samaritan piece is all about.
And I think that you would have to find some other reason to be
protected for doing that. And there are plenty of other good
reasons that a website can say, ``We don't want to put your
political point of view up here.''
The Chairman. OK.
Mr. Cox. But, not Section 230.
The Chairman. Let me quickly move to Ms. Banker.
What type of content do your member companies consider to
be ``otherwise objectionable'' that--where the content is not
already covered by the preceding terms of the statute, which I
just quoted in my question to Representative Cox?
Ms. Banker. Thank you, Senator.
There's a wide range of types of content that are covered
by our member companies' policies, and they are often dictated
by what the business model is of the particular company. So, in
addition to social media members, we also have companies that
have travel information, you know, dating sites, job sites. And
what's appropriate for one of those platforms may not be
appropriate for one of the others. So, it's really important
for our member companies that they have broad protections to be
able to engage in the type of content moderation that helps
users have a positive experience on their services and also
builds trust and confidence in their services.
One example is one I used in my testimony, which is spam,
which is something that courts have applied ``otherwise
objectionable'' to. And I think, you know, obviously, it has a
broad impact on the usability and enjoyment that people have
around online services.
The Chairman. Why don't we just add ``spam'' to the
definitions, and take out ``otherwise objectionable''?
Ms. Banker. I think you can certainly do that. I think that
that would be narrow, and there are many things that we cannot
necessarily predict today. Many of our member companies, for
example, have taken positions against hate speech on their
platforms, and a narrow definition of ``otherwise
objectionable'' could very well, you know, inhibit their
ability to feel like that's something that they can do.
The Chairman. Thank you.
Thank you, Mr. Chairman.
Senator Thune. Thank you, Senator Wicker.
In my day, ``spam'' was a meat product, so we'll have to
define that, probably, in the amendment, too.
Next up is Senator Klobuchar.
STATEMENT OF HON. AMY KLOBUCHAR,
U.S. SENATOR FROM MINNESOTA
Senator Klobuchar. Thank you very much.
And thank you, to you, Senator Thune and Senator Schatz,
for not only holding this hearing, but really walking into this
and trying to take this on. I truly appreciate it. I think we
all know that Section 230 has played an important role in
allowing the Internet economy to develop from its humble
beginnings, but now these aren't just scrappy startups anymore.
They are, at times, trillion-dollar companies and some of the
largest companies in our global economy, and raises many
questions of how Section 230 is not working, and changes that
must be made. And I have some other views on the antitrust side
that I'll save for Judiciary, but this is our focus today.
So, I've done a lot of work in the area of misinformation
in elections with the Honest Ads Act. And, Mr. Sylvain, one of
the most heartbreaking unpaid ads that went out and was
targeted on African-American pages in the 2016 election was a
picture of an African-American woman, and it said, ``Why wait
in line in this election? You can text your vote for Hillary at
83456,'' or pick some numbers. To me, it was a crime, but it
spread all over the Internet and was targeted at their pages.
In your testimony, Mr. Sylvain, you note that Section 230
has helped perpetuate discrimination and racism against members
of historically marginalized groups. Do you believe that
proposals to reform it should consider the spread of election-
related disinformation and how it is targeted at certain
groups?
Mr. Sylvain. Thank you for the question, Senator Klobuchar.
Yes, I do. The trick is, of course, trying to sort out how
to do that. One would be allowing the Federal Elections
Commission, presumably, to entertain enforcement actions
consistent with what the PACT Act sets out. But, you might
imagine civil rights groups also being able to initiate actions
if this is the kind of electioneering that is against the law.
Section 230 does pose an obstacle for that.
One question that I'd like to ask in this context is
whether the law does, indeed, engender a sense of obligation--
of civic obligation. And in a circumstance, the one you've just
described, I suppose it might, after the fact; but, in the
first instance, it obviously didn't. And so, there are a
variety of circumstances under which we can envision a statute
being written in ways that do encourage intermediaries to
intervene before the damage is done. And the conventional way
we do that is--as Senator Schatz, earlier on, said--is
articulated in law. Online intermediaries are immune from that
obligation, because they----
Senator Klobuchar. Right.
Mr. Sylvain.--might act, under the doctrine, as a publisher
or a distributor.
Senator Klobuchar. And, you know, the other way I think
about it is just with this pandemic and the misinformation
going out. Last month, one study found that 38 percent of
Americans believe that the coronavirus outbreak has been
overblown. And, just last week, Facebook had to suspend a group
for spreading disinformation about wearing masks. And I just
think, again, it's just a visceral reaction that I have, and I
think many lawmakers should have, regardless of party, when
these platforms are used in that way.
Let me turn to you, Mr. Cox.
Thank you, Mr. Sylvain.
I noticed--I listened carefully in your testimony, and you
talked about transparency, and you talked about--Congressman,
about how you wrote the law to consider the future, which I
truly appreciate. And if you knew then what you know now about
the importance of large platforms in our election, the
oversized importance, and how they've been exploited by foreign
adversaries, which has been validated by Trump intelligence
officials, to influence and undermine our democracy, would you
have written any protections for our democratic system into
Section 230?
Mr. Cox. Well, I think that watching the case law develop
over the last quarter century has been quite an eye opener. And
when all of you retire from Congress and have a chance to watch
what the courts do with your handiwork, it will, in many cases,
make you proud, and, in other cases, disappoint you. It's--so,
some of what has gone in interpretively with Section 230 has
shocked me. And I would say most of the courts have read the
statute and gotten it right.
I think one of the most important parts of the statute is
the definition in (f)(3) that covers the situation in which a
platform, itself, is actually involved in the illegal content
or activity or conduct. This is not something that, you know,
came to our attention in later years, after we wrote the law.
We thought about this at the time, and we wanted to make sure
that if the website was actually the problem, that the website
could be prosecuted criminally and it could be sued civily.
Senator Klobuchar. Mr.----
Mr. Cox. So, it's taken away. So, what I would do, Senator
Klobuchar, if I were, you know, back then in 1996, and knowing
what I know now in the 21st century, is probably write a 30-
page essay around section--subsection (f)(3).
Senator Klobuchar. OK. But, we don't have--the 30-page
essay isn't going to stop the Internet mess that has been
created when it comes to these political ads. And our problem
is that we don't have any requirements in place right now for
disclaimers or disclosures, not just for the campaign ads. And
some of the platforms have either stopped running political ads
or they have done it themselves. But, it is a complete
mishmash, what is going on. And we have a situation where they
don't have to put disclaimers, disclosures. I'm not just
talking about campaign ads. I'm talking about issue ads. And
billions and billions of dollars have migrated from TV, radio,
and newspaper that have those requirements over to online
platforms that don't. And that is our issue. And we--Senator
McCain did the bill with me. Senator Graham is now a cosponsor
of the bill. And it is separate and apart from 230, but it
would help with this type of illegal activity. And that's
what--why I bring it up to you now.
Mr. Cox. ----
Senator Klobuchar. OK. All right.
Thank you.
Senator Thune. Thank you, Senator Klobuchar.
Senator Fischer. Senator Fischer? Senator Fischer, you want
to turn your mic on?
STATEMENT OF HON. DEB FISCHER,
U.S. SENATOR FROM NEBRASKA
Senator Fischer. Am I on?
Senator Thune. There you go. Yes.
Senator Fischer. OK. Thank you.
Congressman Cox, in your testimony, you described the First
Circuit's dismissal of BackPage.com as an outlier. And you
noted that the court did not reach the sensible decision on the
law in the first place. Yet, we continue to hear concerns from
law enforcement that Section 230 limits their ability to
undertake enforcement actions in cases dealing with illegal
content online. Clearly, we're dealing with daylight between
these takes on the law, here. So, I know that Senator Schatz
touched on this in his question and also Senator Klobuchar. In
your answer to her, you addressed some of the concerns on
Section 230, the effects on hindering case discovery in
prosecution. I'd like to ask you how we can best support
enforcement actions to keep illegal content off of the
Internet.
Senator Thune. Senator Fischer, who was that directed to?
Senator Fischer. Congressman Cox.
Senator Thune. Chris, are you there?
Mr. Cox. Thank you, Senator. Yes, I'm here. And excellent
question. And it's the right focus. Because, as is noted, it
was the original intent of this law to be completely consistent
with law enforcement aims and also with civil redress. So, what
we want to do is, again, punish the guilty and protect the
innocent.
The law gives us tools that aim in the right direction.
First of all, Federal criminal law is completely unaffected by
Section 230. And State law that's consistent with Section 230,
likewise. What we'd like to do, on the law enforcement side, I
think, is get the states more deeply involved. And one way to
do that and not lose the benefits of a uniform national policy
that Section 230 represents would be to extend the limited
opportunities that we have in law right now for State attorneys
general to be deputized by the Department of Justice to enforce
Federal law and/or State laws that are, you know, run in
parallel with the Federal law when Federal law is also
violated. And the PACT Act, you know, gets, to a certain
extent, some of this going. I think there's even more that we
could do, and I've mentioned, you know, specifically how in my
written testimony.
On the civil side, just a quick word about how the law
ought to operate. I think it is vitally important that there be
objective standards that judges could apply at the pleading
stage, where you have to assume that all the allegations in the
complaint are true, and determine whether or not a website, you
know, has to, you know, enter upon a long litigation involving,
you know, civil discovery and so on. I don't think that every
case should go into discovery ``just because.'' And so, the
nice thing about Section 230 is, it has an objective test: Were
you a content creator, or did you contribute or develop it? I'd
like to keep that in anything that we do, because otherwise the
sheer volume of third-party information on the Internet would
mean that every single piece of content now, you know, is a
potential 3- to 7-year litigation.
Senator Fischer. Right. Thank you.
Mr. Kosseff, would you also comment on the tension that we
see with Section 230 and with law enforcement?
Mr. Kosseff. Yes. So, I fully agree with Congressman Cox
about providing state--the states with the ability to enforce
both Federal law as well as State laws that parallel Federal
law. The only concern might be if there were 50 different
significant variations of State laws applying to Internet
platforms, which we see in other areas of the law, like data
security. I also think that the use of the existing Federal
criminal exception is very important. For example, Backpage
having been shot down a few days before FOSTA was signed into
law because of that Federal criminal exception. But, I fully
agree that there should be some balance to allow some State
enforcement.
Senator Fischer. You know, we're seeing case law that's
been developed that's protecting a wide range of platforms for
many different types of claims there. How do you address that,
then? What do we need to do?
Mr. Kosseff. Just to clarify, civil claims? Yes. I mean,
obviously, because of the increase in the complexity of the
Internet, there is a wider variety of claims that go far beyond
defamation. And I think, as Congressman Cox said, the key is,
Did the platform contribute, in whole or in part, to that
content? And if it did, then then Section 230 is not going to
apply.
Senator Fischer. OK. Thank you.
Thank you, Mr. Chairman.
Senator Thune. Thank you, Senator Fischer.
Next up, Senator Blumenthal.
STATEMENT OF HON. RICHARD BLUMENTHAL,
U.S. SENATOR FROM CONNECTICUT
Senator Blumenthal. Thanks very much, Mr. Chairman.
Senator Thune. And in person. In person.
Senator Blumenthal. In person.
I really want to thank you and Senator Schatz for having
this hearing and for your efforts to take a broad look at the
continued viability of Section 230 and the need for reform. And
I think if there's a message to the industry here is, it is:
the need for reform is now. There's a broad consensus that
Section 230, as it presently exists, no longer affords
sufficient protection to the public, to consumers, to victims
and survivors of abuse, and others who deserve greater
protection.
Earlier this month, the Judiciary Committee overwhelmingly
approved the EARN IT Act. In fact, it--the vote was unanimous.
It's a reform to Section 230 specifically crafted to fight
online child sexual exploitation. I have worked with Senator
Graham on it, to go back to Senator Schatz's comment at the
beginning. We actually engaged in bipartisan legislating on the
EARN IT Act to modify it in committee with a manager's package
that met just about all of the potential real objections,
including the impact on encryption.
While working hand-in-hand with survivors on the EARN IT
Act, I was struck by the predicament that they face in
combating sexual abuse of children. And I really want their
voices to be heard today. And so, I'm going to ask that we make
a part of the record a statement from Nicole, whose mother--
whose daughter was a victim of sexual abuse, and she said--and
I'm quoting, if there's no objection, Mr. Chairman----
Senator Thune. Without objection.
[The information referred to follows:]
Senator Blumenthal. Thank you. ``Child sexual abuse is one
of the worst things that anyone can endure. The production,
distribution, and possession of CSAM make it so the abuse is
relived every time the material is viewed or shared.'' She
fought, for years, to rid the Internet of those abusive images,
and many of them are still there. Her child was abused for 7
years.
So, we know that these images can spread like wildfire, in
milliseconds, each time victimizing the survivor again. The
images of Nicole's daughter were shared 168,474 times by
thousands of predators on social media, each time a new trauma,
decade after decade.
I appreciate the Chairman and the Ranking Member's work,
but I'm very concerned about the burden that's placed on the
victims and survivors like Nicole. If we take Nicole's
nightmare, the images of sexual abuse of Nicole's child were
shared on Facebook. Under the PACT Act, to achieve any kind of
remedy against Facebook, Nicole would have to undertake a
number of steps. First, she'd have to go to court and obtain an
order declaring the images illegal, an unnecessary step, or it
should be, when it's plainly clear her child was sexually
abused. And then Nicole would have to submit to Facebook a
sworn statement containing the court order and, quote,
``information reasonably sufficient to permit Facebook to
locate the content.'' Again, the burden on the victim. She
would have to find all the abuse images on Facebook, herself, a
tremendous burden. And only then, if Facebook failed to act,
could Nicole go to court to guarantee that the image is taken
down. Again, the burden on the victim.
The PACT Act does not provide any incentive for Facebook to
police its own platform. Instead, it puts the obligation on
Nicole. It is, unfortunately, cumbersome, costly, time-
consuming, and it would offer Nicole no real relief in ending
her daughter's nightmare. And, in the meantime, those images
would spread as she had to undertake the burden and the
obstacle of seeking a court order. Going to Facebook, going
back to court, and then seeking additional litigation to
guarantee that the image is taken down, the images would be
spreading.
Under the EARN IT Act, Nicole could go straight to court,
stop the images, and, if Facebook has failed victims, a court
can require real changes in behavior and impose actual
remedies, with real teeth, that provide incentives for Facebook
to do better. And what applies to Facebook here would apply to
all of the platforms. They can no longer be regarded simply as
a independent and immune means of conveying a message without
any responsibility, any accountability. To quote Nicole, ``The
EARN Act is about the safety of kids, and there is nothing more
important than protecting them.'' I encourage my colleagues to
listen to her words.
Thank you.
Senator Thune. Thank you, Senator Blumenthal.
Senator Moran.
STATEMENT OF HON. JERRY MORAN,
U.S. SENATOR FROM KANSAS
Senator Moran. Chairman Thune, Ranking Member Schatz, thank
you very much for this hearing.
Thank you, for our witnesses, for joining us.
This is a question intended for all of our witnesses.
Increased automation in the form of artificial intelligence to
improve content moderation and other functions used by
technology companies, along with many other sectors, many
critics and policymakers, including me from time to time, have
called for certain degrees of algorithmic transparency to
ensure increased consumer insight in how automated
decisionmaking is informed, but there is also an inherent
question of preserving the proprietary value of innovative
algorithms that fuel the competitive drive among industry
actors.
How should we, as policymakers, consider improving
algorithmic transparency for consumers without upsetting the
economic competitiveness and innovation pursued by the
industry?
It's odd asking questions to no one I can see.
Senator Thune. Who wants to take the first stab at that
one?
Mr. Cox. Well, I'd like to go second----
[Laughter.]
Mr. Cox.--but, since nobody is speaking up, I will put my
oar in the water.
You know, this is the future. Increasingly, we're going to
see artificial intelligence taking over, not just social media
decisionmaking about how they promote material and so on, but
really all of the management of data across the Internet, in
all manner of contexts. And it raises questions, in
jurisprudence, about liability and responsibility, because if
people can, in the future, say, ``The computer did it, and none
of us knew that that was going to happen, we had no actual
knowledge,'' then it would become, you know, the greatest
loophole ever invented into law.
I think we need to not be distracted by the complexity--and
there is a good deal of it--of code-writing and the way that
algorithms are deployed, and the way that artificial
intelligence increasingly is being deployed, and stick with
enduring principles of the law, as best we can.
I would take this all the way to a principle of writing
statutes and drafting to try to avoid all the jargon, to the
extent you can, because, when you put that jargon in the
statute, one of the unintended consequences is that compliance
is now going to adhere to whatever tech you put in the statute,
and it prevents them from advancing and coming up with the next
best thing. So, stick with general principles.
And what's the general principle here? It is that human
beings write the code. So, ultimately, I think the simple
answer is, imagine a room full of 50 people that did the same
thing that the algorithm did, and what would you do? What would
the legal result be in that case? And I think that that will
lead us to the right answer.
Senator Moran. Anyone want to take the second position that
Congressman Cox wanted to have?
Mr. Sylvain. If Representative Cox doesn't mind, I'll take
the second position. I'm grateful that he went first.
First, I may be wrong on--I'll be happy to be corrected--I
thought there was a Algorithmic Accountability Act in
consideration on the Hill. So, I take Senator Moran's question
to be addressed to that, as well.
There are a couple of ways to--I--so, I agree that it has
to be part of the way we think about intermediaries. My
testimony is addressed to automated decisionmaking systems, and
I'm particularly interested in the question of how we render
them accountable. And one of the ways we might do that is
requiring some kind of--what people who write in this,
including Andrew Selbst, who's written about this a lot, is an
algorithmic impact statement evocative of the sort of things we
envision in other areas of law. So, that's one way of making
this work.
But, for what it's worth, it also speaks to the question of
what we are rendering accountable. For those people who write
and study in this area, there is a question of the extent to
which any given algorithm or automated decisionmaking is
explainable and understandable to most people. And generally
they're not. Most of us--Representative Cox, right?--are not
code-writers. But, on the other hand, there might be a way to
convey it in ways that are meaningful to the public, and an
impact statement might be in that vein.
Senator Moran. Thank you.
Mr. Kosseff. I'm happy to go next, just to echo what was
said before. And what I would just add is that behind
algorithms, there are policies that are informing them, and I
think the first principle is to make policies as transparent as
possible. And we're seeing some progress on that. I would point
to Facebook, I believe, last year, releasing a far more
detailed list of its community standards, not only with just
principles, but with some fairly specific examples of what is
allowable and what's not allowable. And I think that's really
at least a first step in any transparency, is understanding
what policies are driving this.
Senator Moran. Thank you.
Senator Thune. Thank you, Senator Moran.
Senator Udall. Remotely. Senator Udall.
STATEMENT OF HON. TOM UDALL,
U.S. SENATOR FROM NEW MEXICO
Senator Udall. Yes, you got me?
Senator Thune. There you go. Got you.
Senator Udall. Great. Thank you.
Chairman Thune, thank you so much, you and Ranking Member
Schatz, for calling this hearing and focusing on this really
important issue.
President Trump's recent executive order on preventing
online censorship directs various Federal agencies to protect
against online censorship, under the claims that, quote,
``online platforms are engaging in selective censorship that is
harming our national discourse,'' end quote.
Just yesterday, the NTIA filed a petition for rulemaking
with the FCC, asking the agency to develop rules to moderate
online content under Section 230. As Commissioner Rosenworcel
has said, the FCC shouldn't take this bait.
This question is to all of you on the panel. Can you speak
to the legal grounds of the President's executive order? And
what recommendations would you make to the Department of
Justice and the FCC for how they should evaluate this
information to maintain the constitutional protections under
the First Amendment?
Mr. Sylvain. I guess everyone wants to go second. I'll take
a shot, here.
I am deeply skeptical about the nature of the way in which
the FCC could proceed, although, you know, I can envision it
happening, but the FCC would have to do a lot of work to
reverse the ways in which it has, basically, discounted its
obligations or authority under law to interpret and apply
Section 230. This is to say nothing of whether there is a--
there are First Amendment concerns with regards to the SEC
regulating speech.
And I want to go further and say, in my opinion, the focus
on user speech is a distraction from actually the way in which
the constitutional protections for intermediaries works.
Ultimately, they are the ones to resolve what their interests
are with regard to what they allow online, even consistent with
230(c)(2). And that's my own view.
I also think the focus on speech, as though these platforms
are speech platforms, as such, is distracting. Consistent with
my point in the opening, these are commercial entities that
traffic in user data. And they have the prerogative to do so,
but there are few constraints on what they can or can't do.
Senator Udall. Thank you.
Others, jump in, please.
Mr. Cox. Well, I'll take the opportunity to go second,
which I was, you know, always hoping for.
[Laughter.]
Mr. Cox. Because now that Professor Sylvain has laid the
groundwork, it's much easier for me.
A couple of things. When originally we wrote what--became
Section 230, when Ron and I introduced a freestanding bill, it
was called the Internet Freedom and Family Empowerment Act, and
mostly what became Section 230. It contained a provision that
explicitly denied the FCC authority in the area of regulating
the content of speech. And what I said on the floor at the time
is, we don't want to turn the FCC into the Federal Computer
Commission.
I would like to see the FTC be more active in this area.
I'd like to see the FTC, you know, holding platforms to their
promises. You know, when they advertise that they're neutral
platforms, by all means hold them to it. I think the error in
the executive order that President Trump issued, in analysis--
the error in analysis--is in thinking that there's something
about Section 230 that requires all websites, including the
Republican National Committee's website, the DNC website, to be
politically neutral. That's just wrong. The law can't--
shouldn't--require that, and it makes no sense to do so.
But, when platforms adopt business models, when they make
promises to consumers, when they have rules of the road of
content moderation, when they have terms of service, those are,
you know, essentially, advertisements and contractual promises
that can be enforced. That's why we have an FTC. It's why we
have State consumer protection laws. Let's enforce them. Let's
use----
Ms. Banker. I would just add to both of those comments and
say I think it's incredibly important, when we think of how--
availability of Section 230 or it going away completely, what
is the law going to do then? And I think there have been a
number of cases brought against providers, in the last several
years in particular, over content removals. And in those cases,
many times what the users whose content was removed are arguing
is that their First Amendment rights were violated. And we have
repeatedly--you know, without any involvement in Section 230--
and we've repeatedly seen courts say, ``No, these private
companies are not--they're not State actors, so they cannot
violate First Amendment rights.''
So, I think the focus on Section 230 to address the problem
that the EO aims to address is a little bit misplaced, because
I don't think it's necessarily a Section 230 problem, I think
it's more--if you consider it a problem, it's probably a First
Amendment problem.
Senator Udall. Professor Kosseff?
Mr. Kosseff. Yes, I would just echo what's been said, and I
would also just add--and this isn't--there have been a number
of proposals--this isn't specific to the executive order, but
the one challenge in all of this--and there have been a lot of
discussion about conditioning Section 230 protections on
certain editorial behaviors. And that runs into the additional
First Amendment consideration of unconstitutional conditions.
We've seen, in other areas of the law, where, if you withhold a
government benefit, that--in exchange for a specific type of
speech--that can run into constitutional problems. It's kind of
a unique situation, so it requires more exploration, but I
think that's at least one thing to look out for.
Senator Udall. Thank you.
You know, this executive order came out after the
President's many public complaints that various online outlets
are not favorable to him. And Twitter, notably, flagged one
post for inciting violence when he tweeted, quote, ``When the
looting starts, the shooting starts,'' end quote.
So, I think I'll end there, Mr. Chairman.
Thank you so much, to all the panelists.
Senator Thune. Thank you, Senator Udall.
Senator Gardner.
STATEMENT OF HON. CORY GARDNER,
U.S. SENATOR FROM COLORADO
Senator Gardner. Thank you, Mr. Chairman.
Thank you, to the witnesses, for appearing today.
According to Pew Research, approximately 5 percent of
Americans used a social media platform in 2005. Five percent,
the first year they started collecting that data. As of last
year, that number had grown to at least 72 percent. Globally,
social media and Internet adoption numbers continue to rise.
Most of the world's biggest social media platforms--things like
YouTube, Facebook, Twitter, WhatsApp, and more--are companies
founded right here in the United States. This growth in
activities spurred by American innovation that I hope will be a
boon, and should be a boon, to free speech, democratic values,
and greater understanding of differences around the world.
You certainly see that reflected in the mission statements
of these companies today. YouTube says, ``Our mission is to
give everyone a voice.'' Facebook says their platform exists
to, ``give people the power to build communities and bring the
world closer together.'' Twitter wants to, ``give everyone the
power to create and share ideas and information instantly,
without barriers.'' And WhatsApp says their goal is to ``let
people communicate anywhere in the world, without barriers.''
And yet, the most recent Freedom on the Net report from Freedom
House concluded that Internet freedom declined yet again for
the ninth year in a row, even as social media adoption
continues to rise.
So, Congress and the world should ask itself, Why is that?
For one, government censorship is alive and well around the
world. In countries with repressive regimes, questions of
intermediary liability always involve that repressive regime at
the table. Each of these major social media platforms I've
mentioned--YouTube, Facebook, Twitter, and WhatsApp--are banned
and behind the great firewall in China. Their missions of
empowering people and sharing information, without barriers,
don't exactly align with the Chinese Community Party or the
Chinese Community Party platform.
Even in allied countries like Germany, international
organizations like Human Rights Watch have lambasted local
content moderation laws as unaccountable over-broad censorship
rather than sensible public empowerment. The United States
should be pushing back on this global tide of government
regulation, censorship, and blocking.
There is no question. Any social media platform who claims
to be a forum for all and then engages in politically motivated
content moderation is not a platform living up to its stated
mission. And those platforms must be more transparent. But,
should Congress involve itself in enforcing so-called neutral
content moderation? What does ``neutral content moderation''
even look like?
Congressman Cox, we're lucky to have you here today
testifying as one of the two major authors of Section 230,
someone who can give us that clarity about what Congress really
intended when it drafted this statute, insights obviously
critical to policymakers as we consider next steps and whether
or not this statute needs amending at all.
On August 4, 1995, going into the wayback machine, you took
to the House floor and declared, ``Our amendment will do two
basic things. First, it will protect computer Good Samaritans
from taking on liabilities such as occurred in the Prodigy
case. Second, it will establish as the policy of the United
States that we do not wish to have content regulation by the
Federal Government of what is on the Internet, that we do not
wish to have a Federal Computer Commission with an army of
bureaucrats regulating the Internet, because, frankly, the
Internet has grown up to be what it is without that kind of
help from the government.''
So, Congressman Cox, do you think Congress has paid enough
attention to the second tenet of what you said in 1995 was a
fundamental aspect of Section 230? Or do you believe Congress
should still view Section 230 as what helped establish the
policy of the U.S., that we do not wish to have content
regulation by the Federal Government?
I'll turn it over to you, and then I've got another
question for you.
Mr. Cox. Well--and that's great history. I have a twofold
answer to your question. The first is, I do think that Congress
has paid attention and observed that policy as the policy of
the United States in practice, because it has not, to date,
imposed, you know, stifling regulation on the Internet. But, I
will say that the public discussion right now around these
issues is paying little heed to that same norm. And so, as we
reevaluate where we're headed next, I think the question's very
much on the table.
You noted the Freedom House annual reports on how we're
doing across the planet with social media. And what we see in
their report is not just that China and Russia and Iran and
Saudi Arabia, all known cases, are using social media as a
means of social control, reading people's, you know,
communications and punishing them, you know, if they're online,
and so on. And then you have got a system of social credit, you
know if you don't agree, you can't go to school, you can't get
an apartment or get in an airplane, leave the country, what
have you. The point----
You know, our model--different, but--this is a problem in
38 countries covered in their last report----
Senator Gardner. Yes. And I have a----
Mr. Cox.--around the world.
Senator Gardner. I'm out of time. I want to follow it up
with a quick question, here.
As the world continues to consider aggressive Internet
regulation and Goulburn at Freedom continues to weigh in, it's
more important than ever that we are communicating American
ideals about fundamental human rights and First Amendment
protected free speech abroad. How should Congress better enlist
the support of technology companies in that mission? What more
can Congress and American tech companies be doing to make the
Internet a truly free, or a freer, and safer place for all?
Mr. Cox. Well, you know, I'll--go--I'm sorry, go ahead.
Senator Gardner. No, no, that's it. Go ahead.
Mr. Cox. You know, the challenge that our global companies
face in the 38 countries highlighted by Freedom House, for
example, where they're trying to control the Internet and use
it as a means of control are very significant, because the
companies themselves obviously can't conduct foreign policy;
they need the assistance of the U.S. Government. But, you know,
one of the questions that has been widely debated this past
year is whether, when we negotiate our trade agreements, we
should be pushing for this principle, that we want to protect
user-generated content, because, you know, the ability that we
all have to post our things on public forums is derivative of
Section 230 in, you know, protecting the platform from
liability for that. And liability will be a reason that they
won't host our speech, and we won't have that avenue if we take
Section 230 protections away.
So, should we try and put this in our trade agreements? I
believe we should. But, you know, people that are concerned
about, you know, some of the issues that we have here in the
United States, which they don't have in other countries because
they don't have any freedom of speech, in many cases, or it's,
you know, deeply suppressed because of the government's
exercise of its regulatory control over social media and the
internet itself. So, they've taken those concerns that we have
here in America, which I think we can work out and use that as
reasons, and not put--the broad principles into our trade
agreement.
So I think number one, use our trade leverage and use our
diplomacy, and put the Federal Government's, you know, foreign
policymaking as the wind at our back so that our--platforms,
you know, can----
Senator Gardner. Thanks.
Mr. Cox.--and so that the Internet can be visible,
globally, and people can have access to more information around
the world.
Senator Gardner. Thanks.
Senator Thune. Thank you, Senator Gardner.
Next up is Senator Peters.
STATEMENT OF HON. GARY PETERS,
U.S. SENATOR FROM MICHIGAN
Senator Peters. Well, thank you, Mr. Chairman, for this
hearing.
And, to each of the witnesses, appreciate your testimony
today.
I--Mr. Sylvain, the first question I have is for you.
African Americans make up a little less than 13 percent of the
United States population, but accounted for over 38 percent of
U.S.-focused ads purchased by the Russian Internet Research
Agency, and almost half of the user links. The social media
accounts generally built a following by first posing as being
African-American-operated, and by paying for ads that social
media companies then distributed largely to African--or to
African-American users. And near election day, the accounts
urged African Americans to, quote, ``boycott the election.''
So, my question to you, sir, is--African Americans make up
14 percent of the population in Michigan. What recommendation
do you have to prevent these targeted ads from disenfranchising
voters? And what kind of oversight mechanisms could Congress
possibly implement?
Mr. Sylvain. This is one issue that I find extremely
difficult to wrestle with and think about. I do think that
Twitter and Facebook have put a good step forward insofar as
they have asserted that disinformation in regards to elections
is the sort of thing that they were going to try to be
proactive about.
You're asking whether or not Congress can do more. In my--
and consistent with my testimony, I wonder whether there is a
mechanism that Congress can undertake, apart from what Section
230 reform, as set out in the PACT Act, would do.
Let me just speak about the PACT Act. I talked, a moment
ago, about the possibility for civil enforcement. I--you know,
I--again, I'm not an election expert, but my sense is that the
sort as content that you're describing is fully inconsistent
with public law in the context of elections, and so there might
be some opportunity for intervention by an agency in that
setting.
But, the--you know, the other way I think about this--and
this is an indirect question--in a way--indirect way of
answering your question--I think that the big social media
companies, the ones through which a lot of this content has
been distributed, and also through WhatsApp, by the way, which
is even more difficult to track, is pursuant to an interest in
maximizing user engagement, and in spite of whatever the
content is. And sometimes there's a kind of disregard for the
content until an alarm bell is rung. And I think part of the
reason that the content--that misleading information about
elections could flow to African Americans, whomever, in
election years, is because, in the first instance, it's a--it's
drawing attention, apart from whether it's disinformation. So,
the arguments I've made is that you--that companies might be
more alert to their social obligation if they are already
attuned to the obligation to attend to law. So, this is the
indirect answer. The salutary effect of imposing obligations on
companies to attend to law is to attend to things like election
tampering through social media.
Senator Peters. Well, thank you.
Congressman Cox, you noted that the ideal way to screen
wanted information from unwanted information is not for the
government to regulate that flow, but rather for each
individual user to have, basically, the greatest possible
control over what they actually receive. Right now, as you
know, private companies have significant control over screening
what an individual sees or does not see through their
algorithms.
Question for you is, What are your thoughts on providing,
perhaps, individual users with algorithm options that they can
take greater control over what is presented to them?
Mr. Cox. Well, I think that sounds like a wonderful idea. I
mean, the more tools that we have as users of the Internet, all
of us, to customize the content that we receive, the better
that's going to be, from our perspective, for sure. You know,
whether or not those tools will be, you know, freely available,
whether they'll be expensive and we'll have to buy them, I
don't know, but I'm all for it. I hope that, increasingly, such
tools are available. As you know, you know, going back to the
original law, that hope is expressed as one of the reasons that
we adopt a law is to make sure that we have, you know,
continued technological development, that regulation doesn't
stifle it, and so on.
But, just last, I would say that, insofar as, you know,
parsing who should be in charge of deciding, you know, what's
on the Internet, you know, only if we're talking about material
that is legally acceptable in the first place, you know, does
it then become the primary responsibility of the Internet user
to determine whether he, she, or it wants that, you know, on
their computer and in their premises. But, if it's illegal
material per se, then that's the government's business, and it
should be the platform's business to be concerned with that, as
well. And so, we want to get as much of that off of those--the
servers in the first place.
Senator Peters. Great, thank you.
Mr. Sylvain. Senator Peters, do you mind if I jump in one
last--for one last point?
Senator Peters. Yes, please.
Mr. Sylvain. I know your time's up.
Just real quickly. I mean, as Representative Cox was
talking, it struck me that there is another specific tool that
we haven't yet raised or discussed but that, I admit, is more
controversial, and that is looking at targeted advertising. And
maybe there's a way to think about restrictions in targeted
advertising in the context of politics and elections. This--it
is, in many ways, probably a--it's a constitutionally thorny
question, but it's--because we already have rules against
electioneering that are addressed to speech, I--you know, for
example, what you can do outside of a voting station--I wonder
whether that's another possibility.
Senator Peters. Great.
Senator Thune. Thank you, Senator Peters.
Senator Cruz.
STATEMENT OF HON. TED CRUZ,
U.S. SENATOR FROM TEXAS
Senator Cruz. Thank you, Mr. Chairman.
Thank you, to each of the witnesses, for being here today
on this important topic.
Representative Cox, let me start with you. Thank you for
your long service and for your friendship. You and I have had
interesting and substantive conversations on this specific
topic of Section 230 and speech online. And I appreciate your
expertise.
When you were part of drafting Section 230, it was 1996. It
was a long time ago. And, as you stated in your testimony, the
Internet looked dramatically different than it does today. At
the time Section 230 was written, there were roughly 20 million
American adults who had access to the Internet, which compares
now to more than 7-and-a-half billion globally, which is
roughly 375 times as many people have access to the Internet as
they did when Section 230 was drafted.
In 1996, the principal means of access was dial-up,
something, as you talk to young people today, that they don't
even know what dial-up is. But, the speeds were so mind-
numbingly slow that it would take 10 minutes to download a
single low-quality song, and anywhere from 3 to 5 days to
download a single low-quality movie. The world has changed
dramatically, and the money and power that has been
concentrated in Big Tech is altogether different. In 1996, the
Internet was still a nascent technology, and a technology that
was very much being incubated from the ground up.
Representative Cox, when you helped draft 230, was Google a
company? Was Facebook a company? Was Twitter a company?
Mr. Cox. No, not at all.
Senator Cruz. So, none of them existed. There was no player
that had the dominant monopoly position that we see,
particularly at Google, in terms of controlling search,
controlling access to videos, and being in a position to
potentially manipulate search outcomes and silence views with
which they disagree.
In your opinion, should we be concerned about Big Tech
censorship? Is it a problem that should trouble, not just
Congress, but the American people?
Mr. Cox. Well, thank you, Senator. And there's a lot to
unpack, here.
You know, first, you're looking back to 1996, and even
1995, which was older still, and that's when Ron and I started
on this. The concerns that we had at the time, I think are even
more powerful in today's 21st-century context, because what we
were concerned about was this essential distinction between the
new technology, the Internet, and all that had preceded it,
that government had found a way to regulate newspapers,
television, and radio, where there was one group of content
creation--the TV station, the radio station, the newspaper
editorial staff--and then, you know, millions of passive
recipients. With the Internet, you had millions--it was
millions even back then, in the 1990s--of content creators who
would then, you know, instantaneously broadcast to the whole
world. What we could see, if the platforms were given the legal
responsibility to monitor all of that content constantly and
take legal responsibility for it, well then you couldn't have
the Internet, you couldn't have realtime communication and--
among, you know, millions of people around the planet. So that
today the fact that it is billions, and not millions, makes the
problem all the more stark. And if you impose that liability on
the platforms----
Senator Cruz. So, Representative Cox, if I could ask if you
could focus your answer on the question I asked, which is,
Should the American people be troubled that a handful of
Silicon Valley billionaires have monopoly power to silence
speech they disagree with and amplify views that they agree
with?
Mr. Cox. Well, so I don't want to miss the point of your
question, so may I answer this question within the framework of
Section 230, or do you want me----
Senator Cruz. I'm just asking if you agree it's a problem,
not--I think the solution is complicated, but I'm just trying
to start with, Do you agree this is a real problem?
Mr. Cox. Yes, so I think it--first of all, if a platform is
holding itself out as neutral, and, in fact, is not, and, in
order to accomplish its objective of not being neutral, it is
covering up the fact that it is, you know, doing things behind
the scenes that it disclaims, then I think you have, you know,
a violation of a lot of existing laws, and that it is
absolutely something that we should be troubled about. Whether
or not that's exactly what's happening--you know, this is
something of a hypothetical question--but, whether or not
that's exactly happening obviously is the fulcrum of what we do
about it.
Senator Cruz. So, Representative Cox, my time is expiring,
here, but I want to make an observation. One document that I
would encourage you to take a look at is a document that Google
drafted and that was a major focus of a subcommittee hearing
that I chaired some time ago. And the document is entitled
``The Good Censor.'' Now, this is a Google document they wrote,
in which they described how the old view of the Internet was
what they described as laissez-faire free speech. It was the
model you were talking about, of lots of content producers, let
them speak, let's have a free-speech forum. And then they
described the new model of Big Tech, which is what they called
``European-style censorship,'' active censorship. You know, you
talked about companies purporting to be neutral. According to
Google's document, they listed four companies that had moved
from laissez-faire free speech to European-style censorship.
Those four companies were Google, YouTube, which Google owns,
Facebook, and Twitter. By their own terms, they're not being
neutral. They are actively censoring and, given the monopoly
power they have over free speech, I view that as the single
greatest threat to our democratic process that we have today.
Thank you.
Senator Thune. All right. Thank you, Senator Cruz.
Senator Lee is up next. And I'm going to hand off to
Senator Schatz and go vote.
So, Senator Lee, recognized.
STATEMENT OF HON. MIKE LEE,
U.S. SENATOR FROM UTAH
Senator Lee. The United States light-touch regulatory
approach to the Internet has been good. It's been good for the
world economy, and for the American economy, in particular.
It's produced, for us, an incredible success story of free
speech and of innovation that really no other country in the
world can boast.
Section 230 is, I think, a significant part of that
success. But, the Internet, like all other things, is not free
of bad actors, bad actors who may be to try to use this
resource for harmful and damaging, in some cases deceitful,
purposes. But, in order to find the solution to those evils, we
need to understand the problem.
So, let me start by asking Mr. Kosseff and Ms. Banker, Are
today's Internet problems--that is, political bias, illicit
content, et cetera--caused by Section 230 immunity?
Mr. Kosseff. I guess I'll go first. That's an excellent
question. And I think it's hard, because I think the success of
the Internet and the prevalence of the Internet has certainly
amplified problems. I wouldn't say it's necessarily directly--
--
Senator Lee. Not the cause.
Mr. Kosseff.--Section 230.
Senator Lee. Right.
Mr. Kosseff. But, Section 230 being responsible for the
success of the Internet in the United States, clearly the
Internet has more of an impact, but I wouldn't say that it's
Section 230 itself that's causing the--there are a number of
different factors.
Senator Lee. Ms. Banker, what do you think? Do you agree?
Ms. Banker. Thank you. I'd actually say something perhaps a
little bit different, which is, I think that, in terms of the
types of harms we see on the Internet today and the voluntary
measures that the Internet companies engage in to try and
address those harms, we're actually enabled by Section 230. It
is the protection that companies receive under the law for
their voluntary efforts to try and enforce rules around
illegality or just objectionable content that, you know, has, I
think, put us in a better place today than we would have been
without Section 230.
Senator Lee. All right. That's good to know. Thank you.
I want to talk a little bit about viewpoint discrimination
among Big Tech, particularly among the big social media
platforms. It is concerning. It's concerning on a number of
levels, that you see what has become, I think, increasingly
blatant viewpoint discrimination going on within these
platforms. Now, to be sure, these platforms are not
instrumentalities of the government. They're not government
property. And yet, there is something concerning; in part,
because, from my vantage point, centralization of power leads
to centralization of government power. It makes it easier for
government to take control of these levers, and it can cause
other problems.
Now, if I've been informed correctly, Donald Trump, Jr.,
the President's son, has, as of today, been suspended from
Twitter. Why? Well, I'm told because he posted something
containing content posted yesterday by a number of medical
doctors who were speaking their mind as to what they view as
the appropriate course of treatment for COVID-19. Now, I
strongly suspect there might be an ideological angle, here, or
it might have something to do with the fact that he's the
President's son. Either way, this is concerning; in part,
because these kinds of things tend to influence public policy
and public debate.
I don't think I've ever seen that kind of action taken by a
social media platform when preferring one view on how to treat
strep infections over another, or one view about how to treat
cancer over another. So, that is concerning here.
Nonetheless, I'm not so sure that Section 230 reform, or
repeal, is necessarily the answer we're looking for. I think we
need to find an answer to these questions. And I hope that that
answer can be found within the market, itself, and not through
government. But, I don't think 230 is the way to go.
Now, Mr. Cox, in your 1995 floor speech, which Senator
Gardner alluded to earlier on why Congress needed to adopt
Section 230, you noted the relationship between adoption of
Section 230, on the one hand, and avoidance of Federal
regulation of the Internet. You argued that Section 230, quote,
``will establish as the policy of the United States that we do
not wish to have content regulation by the Federal Government
of what is on the Internet, and that we don't want to have a
Federal Computer Commission with any--with an army of
bureaucrats regulating the Internet.'' So, I think you're on to
something here, but I'd like to ask you. Since your 1995 speech
in Congress, has Section 230 helped the United States avoid
government regulation of the Internet, and of Internet content,
and allowed for Americans to have access to many viewpoints at
the click of a button?
Mr. Cox. Umm, well the answer to that is yes, I think so.
And, you know, with the benefit of hindsight, looking back over
a quarter century of experience, not just in the United States
but around the world, we can see the other model. It exists now
in technicolor. So, take a look at the great firewall in China,
take a look at the way that Russian media is organized, and
take a look at Iran and Saudi Arabia, and then go down the list
of the 38 countries that have been flagged by Freedom House as
using social media for population control, and you can see, you
know, the other model in its extreme. I don't know that, you
know, the United States, with its, you know, democracy and, you
know, for the most part, benign government looking after, you
know, people's rights and so on, would have ended up there in
any case, but if you're looking for the extreme dystopian
alternative to the system that we did choose, it exists now
today in the world. So----
Senator Schatz [presiding]. Senator, if I--we'd like to
move on. We're over time. We have a number of members to speak.
Let's move on to Senator Baldwin.
STATEMENT OF HON. TAMMY BALDWIN,
U.S. SENATOR FROM WISCONSIN
Senator Baldwin. Thank you.
I really want to appreciate Chairman Thune and Ranking
Member Schatz for having this hearing today.
You know, it's important, and it's a thoughtful
conversation, for the most part. And it goes without saying
that the Internet of today looks very different from what
existed when Section 230 was first authored and adopted in
1996. But, as we've already heard from many of our witnesses,
that is, at least in part, because of Section 20--230. So, I
think it makes a lot of sense for us to take a hard look at
this law in light of all of the changes we've seen. But, I also
think we have to approach any potential reforms with
thoughtfulness and humility.
Section 230 allows online platforms to moderate the third-
party content that they host. Its Good Samaritan provision
arguably encourages platforms to do so proactively. But, there
has been a great deal of criticism of companies that have
fallen short in their content moderation, whether by failing to
promptly remove content that violates their own policies,
lacking clarity and transparency about those policies and their
enforcement, or appearing to apply those policies in a biased
manner.
There are a number of proposals, including the measure
introduced by the Subcommittee Chair and Ranking Member, that
would take steps to incentivize, encourage, or require better
content moderation practices, including more transparency and
due-process protections for those whose content is removed. And
I think there's a lot of value in having platforms improve
their content moderation efforts. But, I'm also concerned that
a more proscriptive requirement or set of requirements could
lead companies fearing litigation to overcompensate and push
more content off their platforms.
So, I'd like to get the whole panel's views. Is there an
approach by which we can incentivize active, clear, and
consistent content moderation without the negative consequences
of less-open platforms and fewer new entrants into the Internet
ecosystem?
And let me begin with Professor Kosseff.
Mr. Kosseff. Thanks so much for the question.
And I think you really hit the nail on the head, in terms
of what the challenge is, here. Because, I mean, the balance
that Section 230 has at least attempted to strike is to give
the platforms breathing room to develop their own content
moderation and both not become liable for everything that's on
their system because they're doing some moderation, and also
not become liable if they're aware of something and make a
judgment call to leave it up. Because content moderation at
scale is incredibly difficult.
So, I think that moving more toward transparency, rather
than being proscriptive and saying, ``You must do this,''
because it's hard to get a one-size-fits-all approach for every
platform, but giving the public a much better--much better
insight into how the platforms are working, I think that's a
very positive first step, and I think some of the platforms are
really taking steps by themselves to move in that direction.
Senator Baldwin. Mr. Sylvain? Professor?
Mr. Sylvain. Yes, I actually agree with a lot of what
Professor Kosseff just said. I think--and you hit the nail on
the head, that the question of incentives matters, and we might
have--this is a circumstance where, on the one hand, we do want
to encourage moderation, but, the other hand, we may not want
to chill too much speech. And so, on that principle, I think we
can all agree.
The only difficult, for me, is that the--as embodied in
current law, that's just not good enough with regards to
content that is harmful to historically disadvantaged groups.
And so, I do think we need more than, you know, standardized
conceptions of how to do moderation. I think the PACT Act sets
out a really interesting transparency map and a--process for
takedown. But, I do think we need more. And that's where law
comes in.
You know, there's a--I'm not the first person to observe
that--you know, that law engenders public regarding, to take a
language that Senator Schatz used at the outside, kind of
public-interest, public-regarding norms. Law does that. And so,
how do we formalize that, as a matter of course, rather than
rely on companies' moderation standards? That's the tough
question, and I think I err on the side of seeing more law play
a role, here.
Senator Baldwin. Congressman Cox? Good to see you, by the
way.
Mr. Cox. Well, one of--yes, thanks--one of the reasons that
I am so enthusiastic about seeing more enforcement around
content moderation policies, you know, getting the FTC
involved, getting a consumer protection enforcement involved,
is that I think Section 230 already provides plenty of space
for companies to do things voluntarily; and the maddening thing
is when they don't, anyway. So, when someone brings an
egregious case of law violation to a platform, if the platform
says, ``You know, we have no legal responsibility to do
anything here,'' that's--that, to me, you know, is an outcome,
you know, we should dearly wish to avoid. And so, how about
looking to see whether that platform ever promised that that
kind of illegal stuff was not going to be allowed? And if they
said, ``We take these sorts of things down,'' and don't, then I
think you've got, you know, plenty of grounds for enforcement.
But, more than, you know, a Section 230 all by itself is
necessary, here. There's got to be, you know, some backbone
provided to the very sensible steps that I think common sense
suggests that, you know, platforms should be taking. And most
do, to their credit. But, the cases where they don't, you know,
immediately reach our attention. And when I was in Congress, as
you all still are, you know, the nice thing was, I'd pick up a
newspaper, and if there's an outrage, and say, ``Ought to be
law,'' then, you know, we'd go fix it.
We have laws that I think are even being used that we can
apply to this situation, because there are promises being made,
and they extend even sometimes to political neutrality and so
on, as Senator Cruz was mentioning. When those promises are
violated, the courts have held that they can be enforced.
Senator Schatz. Senator Blackburn.
STATEMENT OF HON. MARSHA BLACKBURN,
U.S. SENATOR FROM TENNESSEE
Senator Blackburn. Thank you so much.
And, Mr. Cox, welcome over to the Senate side.
Mr. Cox. Yes.
Senator Blackburn. It is good to see you.
You and I have talked about Section 230 for years. And, as
you were working on it in the mid-'90s, of course, I was in
Tennessee and chairing the Tennessee Film Entertainment Music
Commission and watching closely what you all were doing here.
And, of course, through this time, the Internet has grown.
I think that in 2018, when we did FOSTA and SESTA, and I
was in the House and chairing Coms and Tech, we were pretty
much looking at the fact that there were reforms that were
needed, there were things that needed to be done, updates that
needed to be given. And our concern is, you have some of our
social media companies that have grown to the point that they
function, in the words of Facebook's founder, more like a
government. And they ought not to be making those decisions as
to prioritization of content or availability of content on
their sites.
So, let me ask you this. And raise your hands. As you look
at post-230 cases, are there any of those that any of you
panelists would say have been wrongly decided?
Mr. Cox. Well, I will jump in, if I may. Oh. Raise my hand.
Senator Blackburn. OK. Go ahead.
Mr. Cox. Yes, in my written testimony, I laid out, in some
detail, how I think the First Circuit got it wrong in the
BackPage litigation. The First Circuit, at least at the
District Court level, subsequently in 2018 rectified its
mistake and interpreted Section 230 correctly, I think, as
against the pleadings that it was looking at in that case.
Senator Blackburn. All right.
Professor Kosseff?
Mr. Kosseff. So, Congressman Cox took my answer that I
think that case actually pointed more toward the discovery
problem that we've discussed earlier, the lack of the ability
to get discovery in extraordinary cases on whether the platform
contributed to the content.
Senator Blackburn. OK.
And let me ask--and I'll start with you, Professor Kosseff,
since your mic is open--as we look at the statute--of course,
Congressman Cox helped to author that statute--do you believe
that it is time, and that it is actually necessary, for us to
revisit and to reform this statute based on concerns that are
in the marketplace now?
Mr. Kosseff. I think revisit, in terms of gathering better
facts--as I said earlier, having a commission that gathers
facts about, for example, what's possible with content
moderation at such a large scale. I don't think we're at the
point of being able to reform, because we have so many
competing viewpoints about what platforms should be doing on
top of what we could require them to do because of the First
Amendment and other requirements. So, I think Section 230 is so
important that we really have a duty to always look at how it's
working, but I think right now we really need to be gathering
more facts.
Senator Blackburn. OK.
Congressman Cox?
Mr. Cox. Yes, I agree with that. I think the real challenge
is figuring out what's doable in the real world. When I look at
some of the legislation that's been introduced, such as the
EARN IT Act, you know, which contains, you know, 16 separate
categories of new, you know, regulatory requirements that are
going to apply to websites, you know, these websites are all
very different. They're not all Facebook. They're not even in
that business. They're not all social media companies. And
there are big differences between, you know, Twitter and
Google, even at the Big Tech level. So, you know, what's
doable, and what's going to, you know, on the other hand, you
know, put at risk the availability of the user-generated
content that we all rely on? And I think that, you know, we
need to be careful of doing things like the EARN IT Act does,
which is to say, well, we're----
Senator Blackburn. OK. Let me ask you----
Mr. Cox.--we're just not going to create a commission such
as Professor Kosseff suggests, but we're going to give that
commission legal authority that Congress won't even have an
opportunity to check on, and it won't be something to
Administrative Procedure Act public input or anything else.
It'll just happen. So, I think we need a little more humility
than that.
Senator Blackburn. Should it be left with the FTC?
Mr. Cox. Well, I think the FTC has a role. But, as I said,
I think the FTC can do a lot of enforcement around false
advertising, around, you know, the consumer abuses that have
been identified. But, I think, you know, there are a lot of
different laws and law enforcement levels, not just the Federal
Government, but States, as well, can be our allies here.
Senator Blackburn. OK. My time has expired, so I will yield
back, but I'm going to send to each of you a question on what
you think we should do so that we're protecting free speech
online and we are still making certain that the social media
outlets do not practice censorship or prioritization.
Thank you so much.
Yield back.
Senator Schatz. Senator Tester.
STATEMENT OF HON. JON TESTER,
U.S. SENATOR FROM MONTANA
Senator Tester. Yes, I want to start by thanking you,
Senator Schatz and Senator Thune, for your good work on this
issue.
And I want to express my appreciation to everybody's who's
testified here during this hearing.
I want to shift the conversation a little bit, at least I
don't think it has been here. I had to get off for about a half
hour. But, when a broadcast station, whether it be TV or radio,
hears false or misleading advertising, they can be held
accountable by the FCC or a court of law. However, I don't see
the same consequences for Internet platforms.
So, I have two questions. Number one, Is this right? Is
this the way it should be? Is this the way it was intended when
you wrote it up, Congressman Cox? And, number two, How did we
get here? How did we get to this point?
And I'll start with you. I'd like to have everybody address
it, but I'll start with you, Congressman Cox.
Mr. Cox. Yes. Thank you, Senator.
The challenge that this poses is only matched by its
importance. I think it's--important for us to get it right,
because this goes to fundamental operation of our democracy.
The difference--there are big differences between the radio
and TV and newspaper paradigm, and the Internet. And that's
what creates the disparity, because the essential functioning
of the Internet is based on not just, you know, 10 or 50 or
even 100 advertisers coming to the TV station, saying, ``Here's
our ad. You know, check it out and put it on the air.''
Instead, it's millions of people, or billions of people, around
the world, because it's a planetary potential user base. And
the question is, Is it reasonable for the law to presume that
the platform has the opportunity to review all of this
material? And, if they were to have the obligation to review
all of it, is it even conceivable that that could be done in
realtime? And, if not, are we willing to say that----
Senator Tester. Yes.
Mr. Cox.--the regulation can do away with that essential
aspect of the Internet, which is realtime communication----
Senator Tester. The point is, though, Congressman, on paid
advertising--in paid advertising, do you believe that they have
an obligation to review that, just as TV and radio and
newspaper do?
Mr. Cox. Yes, well, and this may be an area where the
existing rules could reasonably apply, because it is reasonable
to expect that a company, a platform, has the opportunity to
review its paid advertising before putting it up. You know,
that may or may not be the case, depending on the business
model. But, if that's the way it actually operates--it's why,
you know, the real-world aspects of this are so important--then
I think the way at least 230 was conceived, you're not risking
an unreasonable burden on the platform that would cause it to
dump user-generated content all together.
Senator Tester. Mr. Sylvain, could you respond to that,
also? I'd love to get your perspective as to, when it comes to
paid advertising, should the same rules apply to the Internet
platforms as does the TV, radio, and newspaper?
Mr. Sylvain. Thank you for the question, Senator Tester.
Senator Tester. I'm talking about from an information
standpoint----
Mr. Sylvain. Yes.
Senator Tester.--being misleading.
Mr. Sylvain. Yes. I do think so. And I'd like to start by
getting to your second question first: How did we get here? The
business model on which a lot of the most prominent
intermediaries rely really is focused on maximizing or
optimizing user engagement. And often that is at the expense of
veracity. In some cases, that's OK. But, there are certain
circumstances where lies and misleading and information and
disinformation are costly for the operation of democracy. And
our electoral system is one of them. So, we have a--the
prevailing business model accelerates the distribution of
provocative content, because it's consistent with the interest
of engaging users. And often this is the kind of content that
might misdirect people with regards to our political system.
So, that's how I think we've gotten here. And so, I'm more
alarmed--and I think many members have already expressed in
this hearing--about disinformation online, so much so that I do
think this is an urgent opportunity to align electoral law with
what we see in other communications platforms.
Senator Tester. Thank you.
--you've only got about 20 seconds. Could you respond, to
the best of your ability, quickly?
Ms. Banker. Yes, I'll be quick.
I think the platforms understand that, with regard to
advertising and its importance, that they have stricter
policies and heightened enforcement. But, I just--I want to add
a caution, too, about the tremendous benefit that the new ad
ecosystem has had for small businesses, because it has,
essentially, lowered costs so that businesses that previously
could not have afforded to advertise are now able to. And, as a
former restaurant owner, I can tell you that, for small
businesses that are cash-strapped, the ability to, like, really
target advertising to a key demographic, and be able to do that
quickly and easily, is a huge benefit. So, we'd want to make
sure that, if any changes are made, that, you know, the costs
don't end up again making advertising out of the reach for
critical small businesses.
Senator Tester. Thank you all.
And thank you, Chairman Schatz.
Staffer. Senator Rosen, you're next.
STATEMENT OF HON. JACKY ROSEN,
U.S. SENATOR FROM NEVADA
Senator Rosen. Thank you so much. Thank you, Chairman
Shune, Ranking Member--Thune--Ranking Member Schatz.
Thank you, to all the witnesses, for being here today, and
your work in this area.
I'd like to speak a little bit about the bias in
algorithms, and particularly in hate speech. Because one of the
issues commonly raised regarding content moderation across
multiple platforms is the presence of bias in artificial
intelligence systems that are used to analyze the content.
Decisions made through AI systems, including for content
moderation, run the risk of further marginalizing and censoring
groups that already face disproportionate prejudice and
discrimination, both online and offline.
As a former computer programmer and systems analyst who
began my career when the field was even more dominated by men
than it is today, I find this particularly troubling, because
we've had previous hearings, where the witnesses--they've
discussed the harms that algorithmic and AI systems potentially
pose, including those from biased training data, algorithms,
other system flaws that reproduce historical and existing
social inequities. We see this when firms have tried to
automate their hiring through machine learning, but then find
that it merely perpetuates existing biases toward women.
But, what I want to do, besides sexism, is to talk about
another alarming issue that's happening today. Another
challenging--another challenge that we're facing is hate
speech. And so, when we use algorithms, and we deal
specifically with the growing prevalence of anti-Semitism and
the hate speech that's perpetuated online, this is bypassing
content moderation, if such moderation exists on these sites at
all.
Last year, we saw the deadliest attack in the Jewish
community in American history, when 11 people were killed at
the Tree of Life Synagogue in Pittsburgh. Then, perhaps
unsurprisingly, the shooter was linked to numerous anti-Semitic
postings on fringe social network sites. This one, in
particular, called Gab.
So, for all the panelists, knowing that our online
platforms are running into issues with content moderation
specifically as regards to algorithms and AI systems, how
should mainstream social networks interact with these fringe
sites to stop the spread of manifestos, letters, other hateful
writings?
So, Congressman Cox, perhaps you can start; go then to Ms.
Banker, Professor Sylvain, and then Professor Kosseff, please.
Mr. Cox. Yes. So, there is so much work to be done in this
area, because, despite the best efforts of even the most well-
motivated social media platforms, you know, we see examples
where the algorithms don't work, sometimes the algorithms even
serve up, you know, opportunities for bad people to, you know,
get together. So, we've got to be constantly vigilant.
We've talked about the legal incentives that exist. We've
also talked about what more law enforcement could do in cases
where platforms are actually interfering with the right
outcomes. I think the most troubling challenge for writing law
in this area is, you know, what about the great middle ground,
where people are not--that have the platforms, they're not bad
actors, they're trying to do the right thing, but it just
doesn't amount to enough? And you suggested something that I
think can be a way forward, and that is that there be
collaboration among all the people with good hearts here to
advance best practices and so on.
The term ``best practices'' has been bandied about in some
ways that are constructive, and others not. For example, in
connection with the EARN IT app. But, I think the development
of best practices in some ways, along the lines that Professor
Kosseff has suggested, and then, you know, pushing those out to
smaller websites, and so on, could be an absolutely
constructive way to help.
Senator Rosen. Appreciate that.
Ms. Banker, can you speak to the process with, perhaps,
algorithm and AI systems that we have to moderate content with?
Ms. Banker. Absolutely. Algorithms are obviously incredibly
important today. You know, they help us from everything to
finding the fastest route to get to work to things like content
moderation.
The specific area you were asking about, you know, hate
speech, is an area where--you know, algorithms alone are not
going to be able to get us there. There's a critical role for
human reviewers, and many of our companies, you know, use that,
or rely on that, you know, primarily for these things where
context is incredibly important.
And I would also note that, in terms of industry
collaboration, we've seen that work well in other contexts. You
know, in 2006, you know, Internet companies came together to
form a--kind of a coalition working together to share
technology, know-how, and information to address the issue of
child sexual abuse material. And, more recently, they've done
something similar to address terrorism.
So, I think there's precedent for that, and it's certainly
an issue that the companies care about a great deal.
Senator Rosen. Thank you.
Professor Sylvain, quickly. I know my time's expired.
Perhaps I'll leave you with the last word. Anything to add on
AI and algorithms?
Mr. Sylvain. Well, I would have loved for Professor Kosseff
to jump in. I'll just say, quickly, that----
Senator Rosen. Well, if Ranking Member Schatz----
Mr. Sylvain.--align this point----
Senator Rosen.--would let him, we'll be glad to, yes.
Mr. Sylvain. Just one quick point, and that is, I would
associate--I don't have anything to add with--on--based on what
everyone else has said. I will say that the question of what is
objectionably offensive, I very much appreciated Representative
Cox statutory interpretation lesson earlier. You know, we think
about what is within the subjective decisionmaking prerogatives
of intermediaries. And I think this is squarely in it, for what
it's worth.
Senator Rosen. We'll give you, then, the last word,
Professor Kosseff.
Mr. Kosseff. Thanks so much.
I agree with everything that's been said. I would just also
add that there have been some significant failures, and we
can't ignore that. But, Section 230 does provide the
flexibility for experimentation with the best and most
innovative procedures. We have to look at whether that's
working. But, that is kind of the underpinning of Section 230.
Senator Rosen. Thank you. I appreciate y'all being here
today.
Senator Thune [presiding]. Thank you, Senator Rosen.
We're about ready to wind up, here, guys, so thanks for
your indulgence and your patience, and even for, while I was
off voting, since we have so many people who are appearing
virtually, remotely, including members of the Committee,
sometimes it's hard to cover the gaps when we've got votes
going on, so I understand staff had to step in. So, thank you
for very ably picking up the slack here.
I'm going to ask just, if I might, perhaps one question,
sort of close it out, and then we'll wrap things up.
But, Representative Cox, Jeff Kosseff notes, in his
prepared testimony, that CDA 230 has sometimes shielded
platforms from plaintiffs who've suffered serious harms, and
that, while Section 230 does not block a plaintiff from suing
the person who created the harmful content, there are a number
of reasons why that might be impractical, including the
inability to track the person down or fear of retaliation.
And one very recent example is the case of Force v.
Facebook. And, in that case, the Estate of Taylor Force, who
was killed in a Hamas terrorist attack in Tel Aviv, alleged
that Facebook's algorithm provided the Hamas terrorist group
with the means to find each other and to share content
celebrating terrorist attacks. The Federal Court shielded
Facebook from liability, on the basis of Section 230.
So, the question is, How should Congress deal with this
particular scenario, where a plaintiff, who has unquestionably
been harmed, is unable to seek justice?
Representative Cox.
Mr. Cox. Yes, thank you.
You know, that was a very tough case that you mentioned.
It's an excellent example of the kinds of novel questions that
the courts are going to face increasingly as technology, and,
in particular, artificial intelligence, is going to play a
greater role, not just in social media, but the--you know,
their overall exploitation of data on the Internet.
This case came out of the Second Circuit, which I noted in
my written testimony has followed the national trend of
applying the Ninth Circuit's rule in Roommates.com, which I
credit for, you know, really taking us a great step forward in
the development of the common law under Section 230. I think
Roommates.com got it right, particularly when the Ninth Circuit
considered that case en banc. In that case, the court held that
a website was not entitled to protection from liability under
Section 230 because it contributed content that was at issue in
the case. And an important part of that holding was that the
automated features of the website that were written into the
code, as opposed to any direct human involvement, were
themselves enough to constitute content development under
subsection (f)(3) of Section 230.
In the Second Circuit case that I noted in my testimony--
that was FTC against LeadClick Media--the court followed the
Ninth Circuit precedent, Roommates, and denied Section 230
immunity for the Internet marketer by deeming it a content
developer, even though it didn't create the illegal content at
issue; it merely provided advice to those who did create the
content. That was deemed enough.
So, the fact that the same Second Circuit Court of Appeals
reached the opposite result in the Force case shows how
difficult that decision was, and how difficult it must have
been, you know, on the facts. The plaintiffs argued that
Facebook's algorithms had introduced terrorists to one another,
and then they would have had the further burden of proving that
Facebook was a knowing conspirator with the terrorists. That's
what the Federal statute--it's a similar statute, 18 U.S.C.
2333--requires, or that Facebook, you know, knowingly
participated in that situation. The Court seemed not to be
persuaded that those facts had been adequately pleaded. But,
they hung their hat on Section 230, as you point out. The facts
alleged in the complaint seemed to make it appear that this was
all the work of algorithms and the people at Facebook didn't
themselves know that this happened, though they certainly found
out about it eventually.
These are the kinds of cases involving the work of
algorithms that I think are going to become more frequent. And
so, in the future, it won't be enough, I think, to say that
people didn't know that the algorithm did it. We were talking
about this earlier. I think as AI takes over more and more
responsibility for what we see on the Internet, the law is
going to have to go one way or the other. Unless the people who
write the algorithms are going to be responsible for what the
algorithms do, everybody's quickly going to realize that that
will be the easiest way to avoid any liability at all.
Whether it is possible to write a statute that cleanly
parses when human knowledge is necessary, on the one hand, and
when the work of an algorithm will be deemed to be the work of
a human who could be held liable, on the other hand, that's a
tough question. I think, in the medium- to short-run, this is
going to be left for the courts to continue to sort out. And my
hunch is that, as they've done in the Second Circuit, in the
main, and the Ninth Circuit and elsewhere, that they're going
to continue to look hard to find ways to ensure that wrongs
don't go without remedies.
Mr. Sylvain. Can I answer your question, Senator Thune,
about the Force case?
Senator Thune. Yes.
Mr. Sylvain. So, if I had more space in the 5 minutes, I
was going to talk about the Force case, for a couple of reasons
I had already mentioned. It's a really interesting case out of
the Second Circuit, for reasons Representative Cox mentions.
But, I think it's no less interesting because of Chief Judge
Katzmann's concurring opinion, in which he recognizes that the
reasoning on which the majority of the panel bases its
evaluation of whether or not Facebook should be subject to some
liability in connection with the material support of terrorism
because of recommendation algorithms, he doesn't agree with the
conclusion, and he says that it's an invitation to Congress, an
explicit invitation to Congress, to redress the emergent
realities of automated decisionmaking systems for online
intermediaries. Because, for him, he's not sure, as a matter of
the way they're pled, as I think Representative Cox just said,
that plaintiffs would have won, but this is a question that
should have been resolved by the District Court after discovery
had been gathered, to find out how deeply involved the
recommendation algorithm was in the process. And I completely
agree with Chief Judge Katzmann's approach to this.
I urge the members to look at the concurrence and think
about ways of thinking about, you know, amending the statute.
My impression is that the PACT Act does start this process,
because, particularly in the context of complicated automated
decisionmaking systems, regulators can be in a good position to
evaluate consonance with public law.
Senator Thune. Thank you, Mr. Sylvain.
Well, we would welcome your ongoing thoughts and
suggestions and input with respect to some of these, as you
described, novel situations, where, you know, we obviously want
to get to see justice served. And I think the challenge that we
face in all this, as I mentioned earlier, is protecting
consumers and making sure that platforms, companies, have the
ability to continue to innovate, and give them room to
maneuver, too, but in a way that ensures that consumers are
protected.
So, that's a balance that we're trying to strike. I think
the bill--as I said before, we've put a lot of thought and
deliberation into trying to come up with a discrete set of
solutions that we think get at some of these issues, but there
are certainly exceptions, which will probably have to be, at
some point, addressed, too, simply because of the continuing
evolution of, as you all have pointed out, AI and algorithms
and all these other ways now in which materials are being
curated and made available to consumers.
So, we'll continue this discussion. I appreciate all of you
contributing to it. Thank you.
And we will ask that members will have some time, a couple
of weeks, to submit written questions for the record, and we
would ask all of you, upon receipt of those questions, if you
could submit your answers to the Committee as soon as possible.
It would be greatly appreciated.
I would also like to submit a letter for the record from
CCIA, EPIC, and OTIA. And those will be included in the hearing
record, without objection.
[The information referred to follows:]
Senator Thune. So, thank you all very much. I think this
was very constructive and helpful, and it certainly will inform
our decisions as we figure out how to proceed here. And I hope
that we can proceed. I do believe it's time, and I think there
are a lot of good ideas out there that we can include in a
potential solution.
So, thank you all very much.
This hearing is adjourned.
[Whereupon, at 12:42 p.m., the hearing was adjourned.]
A P P E N D I X
Response to Written Question Submitted by Hon. Shelley Moore Capito to
Jeff Kosseff
Are you aware of any instance where an interactive computer service
has profited from illegal activity online? If so, which services? What
have they done to remedy these instances?
Answer. My response is only my personal view, as a Section 230
scholar, and does not represent the views of the Naval Academy,
Department of Navy, Department of Defense, or any other party.
Online crime is a serious problem. Just as in the offline world,
illegal activity can and does occur in cyberspace. For instance, the
U.S. Senate Permanent Subcommittee on Investigations concluded in a
January 2017 report that online classified advertising site
Backpage.com ``knows that advertisers use its site extensively for
child sex trafficking, but the company has often refused to act swiftly
in response to complaints about particular underage users--preferring
in some cases to interpret these complaints as the tactics of a
competing escort.'' \1\ The U.S. Court of Appeals for the First Circuit
ruled in 2016 that Section 230 barred civil claims against Backpage.com
by victims who were trafficked on the site, causing Congress in 2018 to
amend Section 230 to allow certain civil claims and state criminal
enforcement.\2\ It is important to note, however, that Section 230
always has had an exception for the enforcement of Federal criminal
law,\3\ allowing the Justice Department to seize Backpage.com a few
days before the Section 230 amendment was signed into law.
---------------------------------------------------------------------------
\1\ United States Senate, Permanent Subcommittee on Investigations,
Committee on Homeland Security and Government Affairs, Backpage.com's
Knowing Facilitation of Online Sex Trafficking (Jan. 9, 2017) at 3.
\2\ Doe No. 1 v. Backpage. com, LLC, 817 F. 3d 12 (1st Cir. 2016).
\3\ 47 U.S.C. Sec. 230(e)(1).
---------------------------------------------------------------------------
Many other online crimes also pose a constant challenge for law
enforcement. Among the most serious is the distribution of child sex
abuse material. Online platforms must file reports if they obtain
actual knowledge of apparent violations of child sex abuse material
laws.\4\ The service providers are not legally required to monitor for
child sex abuse material,\5\ and such a requirement could raise Fourth
Amendment concerns.\6\ Nonetheless, many platforms voluntarily use
technology such as PhotoDNA to scan for illegal content. The challenge
for all of us who are concerned about online crime is to encourage the
development and deployment of such technologies without disturbing the
very delicate Fourth Amendment balance that allows these systems to
detect online crime.
---------------------------------------------------------------------------
\4\ 18 U.S.C. Sec. 2258A.
\5\ 18 U.S.C. Sec. 2258A(f).
\6\ See United States v. Ackerman, 831 F.3d 1292 (10th Cir. 2016).
---------------------------------------------------------------------------
______
Response to Written Question Submitted by Hon. Rick Scott to
Jeff Kosseff
Mr. Kosseff, do you think social media platforms, like Twitter or
Facebook, should be able to restrict free speech and expression online
to a greater extent than our First Amendment? And if so, why should
they benefit from Section 230s legal immunity provision that was
intended to promote free speech online?
Answer. My response is only my personal view, as a Section 230
scholar, and does not represent the views of the Naval Academy,
Department of Navy, Department of Defense, or any other party.
Yes. Some content, such as hate speech, is constitutionally
protected yet quite harmful. Online platforms should be able to protect
their users from such harm. Prohibiting a platform from engaging in
certain forms of moderation could raise independent First Amendment
concerns, as would tying a government benefit such as liability
protection to a certain form of moderation.
One of Section 230s findings was that the ``Internet and other
interactive computer services offer a forum for a true diversity of
political discourse, unique opportunities for cultural development, and
myriad avenues for intellectual activities,'' \7\ a recognition of the
role that Section 230 could play in enabling online speech. Yet
Congress also explicitly stated that Section 230 was intended to
overturn a New York state trial court decision that suggested that a
platform faces increased liability for all user content by engaging in
some content moderation.\8\ In other words, Congress wanted platforms
to determine how to moderate content in the manner that their users
demanded. To that end, Section 230 assumes that if platforms fail to
meet users' expectations (either by over-moderating or under-
moderating), users will seek other platforms.
---------------------------------------------------------------------------
\7\ 47 U.S.C. Sec. 230(a)(3).
\8\ Stratton Oakmont v. Prodigy Service Co., 23 Media L. Rep. 1794
(N.Y. Sup. Ct. May 24, 1995).
---------------------------------------------------------------------------
It is fair to question whether the Internet created by Section 230
works fairly when a handful of platforms have billions of users and
control a great deal of online speech. Are other platforms available if
users are unhappy with a company's moderation practices? Being
suspended or banned from one of these platforms can have a dramatic
impact on a person's ability to speak and, in some cases, earn a
living. A suspension from Prodigy or CompuServe in 1996 simply would
not have the same impact.
I cannot say with any degree of certainty what impact a repeal of
or significant amendment to Section 230 would have on moderation. It is
unclear how courts would apply the common law and First Amendment
protections for distributors of third-party content to modern
platforms, as Section 230 has been on the books since 1996. Until we
know the rules in a Section 230-free world, we do not know how
platforms would respond. There is a chance that at least some platforms
would reduce the avenues for user generated content, fearing increased
liability, or they might increase their moderation and block more
content that is on the margins. Conversely, platforms might engage in
less moderation, fearing that any involvement might make them liable
for all user content. It also is conceivable that only the largest
platforms could survive in a world without Section 230, eliminating
smaller competitors and further consolidating venues for online speech.