The Volokh Conspiracy
Mostly law professors | Sometimes contrarian | Often libertarian | Always independent
Journal of Free Speech Law: "Platform Transparency and the First Amendment," by Daphne Keller
The article, which is directly relevant to the Netchoice cases that the Court will consider this Term, is here; here's the Introduction:
Major U.S. platforms can powerfully influence public discourse by removing, promoting, and otherwise moderating users' online speech. Better information about their content moderation would help policymakers enact better laws. It would also serve Internet users' interests as readers, speakers, and participants in democracy.
A recent wave of "platform transparency laws" around the world now require such disclosures. The EU, for example, adopted transparency laws following extensive consultation with experts, and continues to take public comments on implementation issues. The U.S. approach has instead followed an increasingly familiar and depressing pattern: State lawmakers enact hastily-drafted laws, and platforms try to get those laws struck down as First Amendment violations. The Supreme Court recently agreed to hear cases about two such state laws, from Texas and Florida. Its review will encompass only portions of those states' sweeping transparency mandates: the provisions concerning notice and appeal for individual users affected by content moderation. Its ruling on those questions, however, may effectively determine the constitutionality of the states' other transparency mandates.
Judicial analysis of Texas's and Florida's transparency mandates has, to date, been quite superficial. We should hope for better from the Supreme Court. Precedent does not provide clear answers to important constitutional questions about platform transparency, though. Platforms' and states' legal arguments both rely on flawed analogies: Platforms compare themselves to newspapers, and states compare them to food vendors or hospitals. Both analogies give short shrift to the rights and interests of Internet users.
Advocates and courts in platform transparency cases can find plenty of loose analogies, and few or no precise ones. They can also find precedent to support almost any standard of First Amendment review, which leaves them free to be as outcome-oriented as they wish in advancing their preferred policies. With the Supreme Court having granted review in the NetChoice cases, now is the time to think much harder about what those preferred policies should actually be, and what doctrinal framework will best achieve it.
This Article discusses First Amendment concerns with platform transparency laws generally, and the Texas and Florida laws in particular. I will argue that the laws have major problems that were scarcely addressed in the rulings so far—and that there are arguments and framings in favor of transparency that have also been insufficiently considered. I will also identify concrete ways in which the laws might be improved.
[* * *]
The Texas and Florida transparency laws are part of broader "must-carry" legislation enacted by Texas and Florida in 2021. Platforms challenged the laws in two cases, NetChoice v. Moody and NetChoice v. Paxton (collectively called "NetChoice" here). To date, the parties' and courts' attention has mostly focused on the states' must-carry rules, which compel platforms to change their editorial policies. Florida's law, for example, would require platforms to carry all speech—even hate speech or disinformation—as long as it was posted by a political candidate. Texas's law would require platforms' policies on hate speech, disinformation, and other topics to be neutral as to speakers' viewpoint.
The major problem with the Texas and Florida transparency laws is that they will powerfully incentivize platforms to change their editorial rules for online speech. Both platforms and their users will suffer First Amendment harms as a result. Transparency laws—including the provisions under review by the Supreme Court—can be expected to change platforms' speech rules in two very predictable ways. The first is by imposing new documentation burdens, which can affect every step of platforms' industrial-scale editorial operations. Platforms can reduce those burdens by changing their policies. They may take down fewer offensive or harmful posts, or apply blunter rules like simply prohibiting all discussions of racism or prohibiting all nudity regardless of artistic or medical context. They may also simply shut down speech-supporting features like comments on videos.
Even the biggest platforms may have sound economic reasons to forfeit editorial control in these ways. But this state-imposed burden will pose an even bigger problem for the mid-sized platforms caught up in Texas's and Florida's laws, and held to standards designed for incumbent giants like YouTube. Platforms with fewer resources will have more reason to change their editorial policies or even cease competing with incumbents in offering particular features. Texas and Florida lawmakers say that their laws are intended to curb the biggest platforms' concentrated power over online speech. But their laws' poorly-calibrated burdens are likelier to do the opposite—reducing the diversity of forums for online speech, and increasing state influence on the platforms that remain.
The second way that transparency laws will cause platforms to change their speech policies is through state coercion. The NetChoice transparency laws will give the Texas and Florida Attorneys General (AGs) powerful new tools to influence platforms' speech policies. This problem is easy to foresee because it is happening already. Even with their current, more limited authority to investigate "deceptive" representations to consumers, AGs have pressured platforms to align their speech rules with enforcers' political preferences.
Texas AG Ken Paxton brought such an investigation against Twitter in express retaliation for its ouster of former President Trump, for example. He demanded that the platform turn over what one expert called "every document regarding every editorial decision that Twitter has ever prepared"—an expense the company could presumably have avoided by quietly adopting the AG's preferred speech rules. Indiana's AG similarly investigated whether Twitter's posted rules violated consumer protection laws when the company removed his jocular tweet questioning the outcome of the 2020 Presidential election. That inquiry has since led him to subpoena civil rights leaders, including the head of the NAACP, about their conversations with platforms.
An Office of the AG (OAG) investigation in Washington, D.C. provides an example with a different political valence. Because it led to litigation over the OAG's subpoena to Facebook, it also provides an unusual amount of public detail about the real-world dynamic between platforms and enforcers. That case arose from the OAG's concerns about anti-Covid-vaccine posts—speech that may endanger public health, but is also often lawful. The OAG told reporters that its goal was to "make sure Facebook is truly taking all steps possible to minimize vaccine misinformation on its site," presumably by taking down users' posts. The OAG's subpoena demanded that Facebook disclose the identities of users who had posted misinformation. Learning the names of individual speakers was necessary, the OAG said, in order to assess whether the platform had adequately penalized people who repeatedly violated the platform's rules.
Laws like the ones in Texas and Florida will require a massive number of new disclosures, and empower AGs to investigate the truth of each compelled statement. AGs enforcing the laws could easily believe they are acting within their statutory authority, while effectively pressuring platforms to change their speech policies on culture war flashpoint issues ranging from LGBTQ+ rights to hate speech. Rational platforms will likely choose to appease those AGs or other enforcers at least some of the time. Affected Internet users may never know about concessions negotiated by platforms and state enforcers, or be aware of government actors' roles in shaping the information they can see and share online.
Transparency laws don't have to work this way. It is possible to expand platform transparency without simultaneously expanding states' influence over online speech—or, at least without doing so to the dramatic degree that the Texas and Florida have. In this Article, I list many alternative approaches, as well as ways in which the Texas and Florida laws themselves could be amended to reduce their most obvious threats to online speech. None of these improvements will happen, though, if courts do not scrutinize transparency laws more closely. The basic questions of First Amendment analysis—what state interests the laws advance, at what cost to speech, and with what possibilities for better tailoring by legislative drafters—deserve careful attention in the transparency context.
Such attention was sorely lacking in the lower court NetChoice litigation. Both the Fifth and Eleventh Circuits applied what the latter called "relatively permissive" review under Zauderer v. Office of Disciplinary Counsel, a 1985 Supreme Court case about deceptive attorney advertising. Their rulings did not meaningfully examine the ways in which mandatory disclosures about speech and editorial policies might vary from disclosures about the tobacco, sugary beverages, or hospital prices. The Fifth Circuit upheld all of Texas's transparency mandates. The Eleventh Circuit upheld most of Florida's, though it rejected as unconstitutional the user notification provisions at issue before the Supreme Court.
The courts' cursory analysis is understandable given the scant briefing to date—the platforms' brief to the Eleventh Circuit, for example, spent just one of its 67 pages on transparency. But the transparency issues in NetChoice are complex and weighty. Like other questions in NetChoice, they are also truly novel. As one Fifth Circuit judge put it, "[t]hese activities native to the digital age have no clear ancestral home within our First Amendment precedent."
The job of the parties' lawyers, of course, is to make the questions look easy, like something resolved long ago in another context. Platforms attempt to do this by arguing that they are basically like newspapers. They compare transparency mandates to laws requiring the Wall Street Journal to publicly explain every detail of its editorial policies and publication decisions. Such mandates, they argue, are obviously counter to the First Amendment, and prohibited by a case about litigation discovery against newspapers, Herbert v. Lando. Texas and Florida, on the other hand, insist that their rules are basic consumer protection measures, to be reviewed under Zauderer and lower court cases applying its standard of review. Their arguments frame compelled speech about editorial policies as constitutionally indistinguishable from labels on food or warnings in advertisements for commercial services. Both the "platforms are newspapers" and "platforms are ordinary sellers of goods and services" arguments usefully illuminate some aspects of platform transparency laws. But both are also incomplete, whether as analogies for the function of today's platforms or as pointers to relevant case law.
Following this Introduction, Part II of this Article will describe the specific transparency mandates at issue in the NetChoice cases, and briefly outline major relevant precedent. Part III will then describe the concrete ways in which poorly-tailored transparency mandates may cause platforms to change the editorial policies that they apply to Internet users' speech. This burden on speech is fundamentally different from the burdens created by transparency mandates for commercial offerings in areas like food safety. Case law addressing analogous speech-related problems exists, but is scant.
Part IV will explore the potential state interests and First Amendment framings for platform transparency laws, and delve more deeply into the case law. Subpart IV.A begins with the consumer protection interests advanced by the states in NetChoice, and critiques Zauderer as a basis for upholding the Texas and Florida laws. Subpart IV.B then considers precedent involving more complex disclosures from regulated industries like banking or pharmaceutical production. It flags what I believe is a major lurking issue in NetChoice: A ruling on platform transparency issues will likely be relevant for future cases in which businesses seek to "weaponize" the First Amendment as a legal tool against the regulatory state. That issue is far bigger than platform regulation. Its gravitational pull will likely shape the choices of advocates and Justices in NetChoice.
Finally, in Subpart IV.C, I discuss a fundamentally different basis for transparency mandates, as a tool to advance democratic self-governance goals. This foundation is underexplored in case law and academic literature about platform transparency. But it is, I will argue, profoundly important as a basis for future, better transparency laws. It also provides a potential constitutional framing that avoids many of the pitfalls of Zauderer.
Like many First Amendment cases, the outcome of this one will likely turn on the standard of review. Unlike in many First Amendment cases, the right standard of review is highly indeterminate. Advocates and judges can effectively pick the outcome they want, and find an argument to support it. Platforms' arguments in NetChoice would lead to strict scrutiny, effectively killing most possible transparency laws. The states' arguments have led to review so lax as to disregard major constitutional issues.
A better standard of review would lie in between. In principle, it might be found in intermediate scrutiny. Or the exacting scrutiny standard from election law cases might be a fit, in recognition of states' democratic interests in platform transparency laws. That would be a stretch from existing precedent. But so is applying Zauderer, or much of anything else. Perhaps even Zauderer itself could be interpreted as the source of a sufficiently robust standard—though given the analysis in the Fifth and Eleventh Circuits so far, I am skeptical.
Whatever analysis the Supreme Court adopts, it should insist on a clear connection between states' goals and the laws' consequences, and require meaningful tailoring. The First Amendment problems with badly-drafted platform transparency mandates are far too real for hasty analysis and unduly permissive standards of review.
