How to regulate speech on social media platforms has been the subject of vigorous debate and recent Supreme Court opinions. In particular, government agencies communicating with the public via digital platforms must balance the need to moderate comments with users’ First Amendment rights. Recently, in People for the Ethical Treatment of Animals (PETA) v. Tabak, the D.C. Circuit held that keyword filters used by the National Institutes of Health (NIH) to hide “off-topic” comments on its social media posts constituted unreasonable speech restrictions. Although the court reached the correct outcome in this case, the D.C. Circuit failed to consider how speech regulations may function differently in digital spaces. Failures to analyze online regulations in light of their unique digital context may have negative consequences for free speech in the long term. Adopting a form of internet exceptionalism that balances technological characteristics with traditional First Amendment considerations may result in more uniform free speech protections online, while preserving interactivity between government agencies and the public.
NIH is the nation’s foremost federal agency conducting biomedical and other public health research. Like many government entities nowadays, NIH has verified social media accounts. On its Facebook and Instagram pages, NIH shares research and news updates with the public “to communicate and interact with citizens” about the agency’s work. Any member of the public may leave a comment on NIH’s posts. Indeed, the NIH Comment Guidelines “encourage[] [the public] to share [their] thoughts and ideas” in a “respectful and constructive dialogue,” while clarifying that “NIH blogs are not intended to serve as public forums.” The Guidelines also set etiquette standards around post engagement, prohibiting certain types of language and content in the comments, including “[o]ff-topic posts.” To monitor comments according to their Comment Guidelines, NIH takes advantage of Facebook’s and Instagram’s default filters, as well as custom keyword filters which automatically hide any comment containing a flagged keyword. Due to resource constraints, the keyword filters “in practice” comprise the entirety of NIH’s moderating efforts.
Among NIH’s frequent commenters are animal rights advocates, including PETA, a nonprofit organization that works to advance animal rights. PETA often commented on NIH’s social media posts to protest the agency’s use of animal testing in its research. Inundated with such comments, NIH implemented keyword filters to moderate the comment threads on their posts. The keyword filters, which included PETA, PETALatino, animal(s), cruel, cruelty, and hashtags #stopanimaltesting, #stoptesting, and #stoptestingonanimals, automatically hid many comments critical of NIH from public view. In response, PETA and two animal rights activists sued to challenge NIH’s use of keyword filters, arguing that the filters infringed on their First Amendment rights. After NIH removed several of the keyword filters from their social media platforms, both parties moved for summary judgment.
The district court granted summary judgment for NIH, holding that the keyword filters were viewpoint neutral and constituted a reasonable speech restriction. Looking to the publicly available NIH Comment Guidelines, the district court found that NIH intended its social media pages to be used to discuss only certain subject matters, as the Guidelines required comments to stay on topic. Given this clear and consistent policy, the district court held that NIH’s social media pages were limited public forums where the agency may “exclude speakers on the basis of . . . subject matter” if the exclusions are “viewpoint neutral and reasonable in light of the purpose served by the forum.” The district court found the policy to be reasonable because the “clutter of off-topic comments” was disruptive and distracted from NIH’s “purpose of fostering productive dialogue.” Additionally, the district court found that the exclusions did not discriminate against any particular viewpoint because the keyword filters screened out comments based on content, not perspective.
The D.C. Circuit reversed. Writing for a unanimous panel, Judge Garcia first conducted his own forum analysis and agreed with the district court that the comment threads on NIH’s posts were limited public forums. Applying factors from Cornelius v. NAACP Legal Defense & Educational Fund, Inc. to elucidate the government’s intent, Judge Garcia considered “objective indicia” like “the nature of the [government] property, its compatibility with expressive activity, and the consistent policy and practice of the government.” He noted that the “stated purpose” and “categorical[] . . . subject matter regulations” in NIH’s Comment Guidelines and keyword filters clearly indicated that NIH aimed to limit discussion on its social media pages to topics relevant to the agency’s research. That NIH inconsistently enforced its policy was immaterial to the question of intent because the agency did not “affirmatively” adopt a stance of nonenforcement. Rather, Judge Garcia relied on the existence of the keyword filters as proof that NIH did not abdicate its subject matter policy. Pointing to United States v. American Library Ass’n, he asserted that underinclusive censorship is not determinative where the government makes some effort to restrict certain subject matters. While conceding that “social media is inherently compatible with expressive activity,” he concluded that the Cornelius factors taken together weighed toward classifying the comment threads as limited public forums.
Judge Garcia’s reasoning, though, parted ways with the district court’s when assessing whether NIH’s keyword filters constituted a reasonable speech restriction. Again relying on the Cornelius factors, he noted that because the forum’s primary purpose was “compatible with expressive activity” — that is, encouraging dialogue with citizens — “a more demanding test” for reasonableness was required than if the forum’s purpose had been “less compatible with expressive activity.” In his analysis, Judge Garcia identified three problems with NIH’s off-topic policy: First, the keyword filters were too vague and overbroad. NIH did not employ “objective, workable standards” to define the distinction between on- and off-topic subjects. Nor could NIH “articulate some sensible basis” for why some comments were blocked by the keyword filters. This overinclusion, Judge Garcia asserted, “d[id] not ‘ring[] of common-sense,’” as many of NIH’s posts directly related to research involving animal testing. For example, the filters were set up to block “a comment like ‘animal testing on zebrafish is cruel’” on a post about zebrafish research. NIH also failed to define “off-topic,” leaving its social media moderators without clear guidance on what was or wasn’t an acceptable comment. Second, the keyword filters were “inflexible and unresponsive to context.” Unlike the website-blocking filters at issue in American Library, the keyword filters applied automatically without regard to context, and NIH did not follow up with any manual review. The lack of recourse available to blocked commenters and the undiscerning nature of the off-topic policy’s implementation “reinforce[d]” the keyword filters’ “unreasonableness.” Lastly, Judge Garcia expressed skepticism toward NIH’s off-topic policy because the keyword filters “skew[ed] sharply against the appellants’ viewpoint.” Such targeted censorship was unreasonable because of its “potential to distort public discourse” by disparately impacting a relevant, but critical, perspective. As such, Judge Garcia held that the keyword filters were unconstitutional.
Although the D.C. Circuit reached the right outcome, its analysis ignored the differences between virtual speech online and spoken speech offline. Rather, the court’s approach in PETA v. Tabak is emblematic of a general trend across the judiciary to analyze virtual forums as simply another type of public forum, even though public forums have historically been physical spaces. This preference for doctrinal consistency may, in fact, come at the cost of consistent speech protection, as what may be reasonable in a physical forum may not exist or be possible online. A balancing inquiry that takes into account the unique characteristics of virtual forums would both better ensure uniform First Amendment protections across forums of vastly different characters and help preserve interactivity between the government and the public on social media.
Advocating that virtual forums may sometimes necessitate different treatment constitutes a type of exceptionalism. Others have discussed internet exceptionalism generally. Professor Mark Tushnet, for example, has noted how — because of the internet’s distinctive qualities — a standardized balancing test can produce exceptional outcomes when indiscriminately applied online. Instead of a generalized approach, this comment proposes adopting an ad hoc balancing test, termed functional internet exceptionalism. Functional internet exceptionalism would look at only one particular online regulation and corresponding speech problem at a time to determine the regulation’s effect on free speech. Building exceptional characteristics into the balancing analysis on the front end would paint a clearer picture of the regulation’s proportionality. Then, based on how the regulation operates in the virtual forum, one can better assess whether any exceptional treatment is warranted. This approach could be incorporated into the existing doctrinal framework.
Exceptionalism in the First Amendment context is not a new concept. In a concurrence, Justice Robert Jackson once extolled that “[t]he moving picture screen, the radio, the newspaper, the handbill, the sound truck and the street corner orator have differing natures, values, abuses and dangers” such that “[e]ach . . . [may be] a law unto itself.” Moreover, there is some precedent that different media may require exceptions to certain general rules. In Red Lion Broadcasting Co. v. FCC, the Supreme Court “strongly suggested” that the limited access to broadcasting channels warranted exceptional regulation of radio broadcasters to ensure that power imbalances did not impinge the rights of listeners and speakers. Similarly, the Court upheld federal must-carry regulations for cable television providers in Turner Broadcasting System, Inc. v. FCC even though it had struck down parallel regulations for the newspaper industry. These exceptions for broadcasting systems and cable television were justified, in part, by how those technologies made it possible for a limited number of providers to control the dissemination of information over the airwaves or cable network. To ensure proper protections, the Court treated these technologies differently than other media or forums.
So far, the Court has rejected the notion of internet exceptionalism, at least in principle. The principle of internet exceptionalism is “the idea that the network has its own sovereignty . . . or is an exception to law,” an idea that has likely died out, particularly in First Amendment law. Instead, lower courts have largely taken a nonexceptionalist approach, interpreting dicta in Supreme Court opinions like Packingham v. North Carolina and Reno v. ACLU to presume that virtual forums should be treated the same as physical forums. But although the Supreme Court has “unequivocally rejected” the principle of internet exceptionalism, the question lingers whether some other form of internet exceptionalism — such as functional exceptionalism — may still be viable.
Last Term, the Court heard five social media cases but avoided resolving any of the questions raised. In Lindke v. Freed, the case perhaps most analogous to PETA v. Tabak as it involved comment blocking, the Court announced a straightforward test for determining when state action occurs on private social media accounts. Yet the Court declined to apply the test or find that the government official in Lindke acted unconstitutionally. By remanding the case instead, the Court “left significant questions open” about how online speech issues should be analyzed.
These questions are all the more pertinent because of the unique challenges posed by virtual forums. The internet lowers the cost of information dissemination while providing amplification and anonymity to any member of the general public. Social media can also be “interactive and dialogic, rather than monologic, as traditional media ha[s] been.” These technological features can result in a plethora of postings that drown out important information and diverse voices. Given the volume of comments possible on virtual forums, efforts to capture problematic material will undoubtedly be both overbroad and vague, at least under traditional conceptions of those terms. Applying functional internet exceptionalism, the court could account for the uniquely large volume of speech made possible by virtual forums. The court could balance the volume of comments possible with the overbreadth and vagueness of a restriction to determine the restriction’s reasonableness.
The spatial differences between virtual and physical forums should also be taken into account. PETA v. Tabak is illustrative on this point. The D.C. Circuit distinguished NIH’s keyword filters from the reasonable filtering software in American Library “because library patrons could easily disable the filtering software by asking a librarian to unblock” sites that were “erroneously blocked.” This reasoning poses obstacles when applied to cyberspace. The sheer volume of comments possible online renders American Library-esque manual, on-the-spot corrections infeasible. Moreover, online commentators may enter and post on digital forums from any location, at any time, outside the moderators’ presence (unless the moderator is a 24/7 filter). Unbounded by physical constraints, digital platforms possess a fundamentally different spatial quality than physical forums: Time and space don’t pose the same kind of constraints online that they do in the physical world. As with comment volume, courts employing functional internet exceptionalism could weigh spatial dissimilarities when applying traditional First Amendment analyses to digital forums. This balancing act would produce greater uniformity by acknowledging the distinct characteristics of online forums to achieve well-calibrated free speech protections — even if, at first glance, they may seem inconsistent across different mediums of communication.
Because of cyberspace’s practical shortcomings, failure to adopt functional internet exceptionalism could increase difficulties in moderating online forums and frustrate interactivity between government actors and the public. Ever since federal agencies flocked to social media during the Obama administration, interaction between the citizenry and government has increased. But how to moderate disruptive comments remains an intractable problem. Curtailing government use of moderation tools like keyword filters may lead agencies to reduce interactivity on their social media pages, resulting in less overall expressive liberty. If faced with a flood of disruptive comments degrading discourse, agencies may choose to disable commenting, rather than expend limited resources on manual review or grapple with a First Amendment headache.
Although at first blush a victory against unreasonable censorship, PETA v. Tabak should give free speech proponents pause. The court applied traditional vagueness and overbreadth doctrines to reach the correct outcome, yet disregarded problems posed by social media’s distinct technological characteristics. A continued failure to consider the distinct challenges posed by virtual forums could result in less free speech, rather than more. In order to consistently protect the rights of speakers and listeners on social media while preserving interactivity between the public and federal agencies, courts should embrace functional internet exceptionalism — and nonexceptionalism should be the exception, not the norm.
The post <em>People for the Ethical Treatment of Animals v. Tabak</em>, 109 F.4th 627 (D.C. Cir. 2024) appeared first on Harvard Law Review.