Matthew B. Lawrence
On July 27 the Senate passed
the Kids Online Safety Act. The bill, a
major federal public health measure regulating social media platforms, now
moves to the House, but it is dogged by opponents’ questions about its
constitutionality.
For years, uncertainty has surrounded state and federal
efforts to regulate social media platforms. Last month’s decision in Moody v. Netchoice
endorsed a framework for assessing the constitutionality of laws regulating
platforms that substantially clarifies the law in this space, although major
questions remain. As described below, what
might be called the neutrality triangulation framework endorsed in Moody
embeds Balkin’s free speech triangle, looking not only to how a law treats
platform conduct but also to how the regulated platform conduct treats user
content.
Readers are no doubt aware of concerns that platforms have
used their power to construct and oversee digital
spaces in ways that produce unnecessary harm. For example, plaintiffs in a growing number
of addictive design cases allege that
platforms have used slot machine design tricks to foster compulsion in
kids—that the platforms have designed
the spaces in which we interact to be more like a casino than a public square,
complete with flashy machines dispensing unpredictable rewards on schedules structured
to exploit human psychology and get people “hooked.” The
bipartisan sponsors
of the Kids Online Safety Act highlight youth mental health concerns, as does an
important Surgeon General’s advisory.
Jonathan Haidt’s “The
Anxious Generation” is a #1 New York Times bestseller. And, of course, concerns about social media harms
go far beyond these direct public health impacts to include worries that social
media supercharges the spread of misinformation about elections
and epidemics, among other harms.
These concerns have fueled regulatory efforts aiming to
mitigate social media harms, but—as Gaia Bernstein’s book Unwired describes—these efforts have been clouded by unresolved First
Amendment questions. Due to the novelty
of the technology, it has not been clear how courts would or should think about
how the First Amendment applies to laws that regulate platforms. For example, would courts view Tik Tok as more
akin to a slot machine (themselves usually assumed to be beyond
the coverage of the freedom of speech), to a “matchmaker” (perhaps subject
to distinctive
rules), or to a novel (subject to strict
constitutional protections)?
The Supreme Court went a long way toward resolving these
issues in Moody v.
Netchoice. The case was focused
on the constitutionality of Florida and Texas laws restricting discrimination by
platforms in their censorship and content prioritization choices (what they
recommended to users or put at the top of ther “feeds”) against certain
political content (for more on the case see Teachout’s summary
in the Nation). But observers and a host
of amici (including a brief
by a group of law and history scholars and the American Economic Liberties
Project that I joined) watched the case closely in the expectation that the
Court might use it to develop a broader framework for evaluating the
constitutionality of platform regulation, and that turned out to be true.
In Moody, the court endorsed a neutrality
triangulation approach to resolving questions about the applicability of the
First Amendment to content moderation. Several
years ago, Balkin pointed out that in the platform era, “Free Speech
is a Triangle” because speech governance is not simply a bilateral
relationship between the state and the speaker, it is a trilateral relationship
between the state (which can regulate the platform or the speaker), the
platform (which regulates the speaker), and the speaker (who is governed by
both the state and the platform).
This triangle concept is embedded in Moody. Justice Kagan’s majority opinion explains
that the “expression” entailed in content moderation that may be protected by
the First Amendment is a “particular edited compilation of
third-party speech.” Op. at 11, 14, (emphasis
added). In other words, “expressive
activity[] include[es] compiling and curating others’ speech. Op. at 17 (emphasis added). And the opinion elaborates on what it means by
the key words “compiling” and “curating,” namely, “choices about whether—and if
so, how—to convey posts having a certain content or viewpoint.” Op. at 24.
Note how the key coverage question under Moody is not
about the relationship of the government regulation and the platform conduct
only; it is about the interaction of the platform regulation and user content—as
indicated in Figure 1 below, the other vertex of Balkin’s triangle. The court indicates that state regulation of
a platform function is protected by the First Amendment if the platform
function itself discriminates (entails “choices about whether—and if so, how—to
convey”) among user speech based on its “content or viewpoint.”
Figure 1: Neutrality triangulation
This is a function-by-function, choice-by-choice level
determination. It is not platforms that
might be covered by the First Amendment (or not), nor is it content
moderation (loosely defined) that might be covered. Rather, it is individual functions of
particular platforms that might or might not be covered (with the key
question being whether those functions entail discriminating among user content). As Justice Kagan puts it, courts must “ask[]
as to every covered platform or function, whether there is an intrusion on
protected editorial discretion.” Op. at
11; see also op. at 18 n. 4 (“an entity engaged in expressive activity
when performing one function may not be when carrying out another”). Technical details matter.
In other words, what Moody seems to be saying is that
assessing the coverage of the First Amendment when it comes to the broad set of
activities understood as “content moderation” is a question of neutrality
triangulation: Whether a platform activity is itself protected by the First
Amendment depends on whether that platform activity is content neutral vis a
vis user speech. When a platform
operator is actively discriminating among user speech based on its content—making
choices to prioritize some content and to censor others—it is protected by the
First Amendment. When a platform
operator is not actively discriminating among user speech based on its content it
is not covered (unless it is otherwise expressive). Just as states have more discretion to
regulate speech when they do so in a content-neutral way (because
constitutional tests for content-neutral regulation are more forgiving), states
have more discretion under Moody to regulate platform activity when the
activity they regulate itself is content neutral vis a vis user
speech.
As I explain in a recent article
describing this neutrality triangulation approach, this way of understanding
the First Amendment’s applicability to particular platform content moderation
choices has been percolating in the lower courts recently. In the Social Media Cases pending in
California, for example, Facebook, Instagram, Snapchat, TikTok, and YouTube
moved to dismiss all plaintiffs’ addictive design cases on First Amendment
grounds. Judge Kuhl granted that motion
in part but denied in part last October in an opinion that focuses on the content
neutrality vel non of the regulated platform conduct vis a vis user
content. Judge Kuhl thus found many of
the claims to lie outside the coverage of the First Amendment because “[t]he
allegedly addictive and harmful features of Defendants’ platforms are alleged
to work regardless of the third-party content viewed by the users.” Op. at 38.
An advantage of the neutrality triangulation approach is
that it produces workable lines. Those
lines are illustrated in as Judge Kuhl’s opinion, as well as a
hot-off-the-presses Ninth Circuit opinion in Netchoice v. Bonta applying
Moody (and so neutrality triangulation) to affirm in part and vacate in
part a lower court’s ruling enjoining in toto the California
Age-Appropriate Design Code Act. Proponents
of regulation allege that infinite scroll encourages compulsive use by
eliminating natural stopping points and thus opportunities for self-governance;
that design feature regulates the manner in which users consume content so it
would not be covered by the First Amendment.
Proponents of regulation also allege that platforms prioritize content
encouraging eating disorders or self-harm in adolescents’ feeds; any active
such choices would discriminate among user content and so would be protected by
the First Amendment. Under the
neutrality triangulation approach states can regulate the former as their
democratically elected or empowered policymakers think best, but any regulation
of latter must be supported by a sufficient (and sufficiently-tailored) state
interest.
Frischman and Benesch helpfully analogize
neutrality triangulation to the distinction between “content-based” and “time,
place, and manner” restrictions in traditional free speech law. It has long been understood that states may
more readily regulate the “time, place, and manner” of speech directly than
they may its content. Neutrality
triangulation draws a similar line along the free speech triangle’s other vertex:
It permits states to regulate platforms’ choices about the “time, place, and
manner” of user speech more readily than they may regulate platforms’ choices to
discriminate among user speech based on its content or viewpoint.
To be sure, much remains unresolved. The majority’s endorsement of the neutrality
triangulation approach in Moody is technically dicta because its holding
was premised on the lower courts’ failure to assess the appropriateness of the
platforms’ facial challenges in the case.
Of course, I am here offering my own understanding of the Court’s ruling
in Moody; triangles do not actually appear in Justice Kagan’s opinion
(though, in light of Moody, we may well see more triangles in future
opinions). Moreover, the majority made
clear that states may well have interests sufficient to regulate even platform
design and operation choices that are covered by the First Amendment. As Justice Kagan put it, “[m]any possible
interests relating to social media” could justify regulation even of protected
speech. Op. at 4. She hinted that public health concerns around
youth mental health may be one potential source of such interests. Op. at 19 (“Today’s social media pose dangers
not seen earlier. No one ever feared the
effects of newspaper opinion pages on adolescents’ mental health.”). There is
also the question, teed up by Justice Barrett, whether platform choices must be
actual choices made by an actual human being who is entitled to the First
Amendment’s protections in order to trigger coverage. Op. at 22 n. 5.
While much remains unresolved, the Supreme Court’s
endorsement of the neutrality triangulation approach provides guidance that
legislators can consider in crafting laws regulating social media, that courts
can consider in adjudicating challenges to such laws, and that researchers can
consider in developing the evidence base addressing the benefits and costs of
such regulation. After Moody, all
these groups should remain mindful of the potentially-determinative question
whether regulated platform design and operation choices discriminate among user
expression based on its content.
Matthew B. Lawrence is Associate Dean of Faculty and Associate Professor at Emory University School of Law. You can reach him by e-mail at matthew.lawrence@emory.edu.