Facebook and Politicians’ Speech

70 Am. U. L. Rev. F. 203 (2021).

* Professor of Law, Washington and Lee University School of Law. The Author thanks Josh Fairfield and Shaun Shaughnessy for their helpful comments.

Abstract

In his Article Facebook’s Speech Code and Policies: How They Suppress Speech and Distort Democratic Deliberation, Professor Joseph Thai argues that Facebook skewed public debate with a policy that exempted politicians from its content-based rules. This Response updates the reader on Facebook’s retreat from this policy and identifies some preliminary lessons from it. Between May 2020 and January 2021, Facebook moved away from its “light touch” regulation of politicians’ speech by employing strategies like labeling and down-ranking—and, eventually, removal of content. After the January 6, 2021 insurrection at the U.S. Capitol, Facebook de-platformed President Trump altogether, putting a final end to the “hands off” policy and ushering in a new era in which, apparently, Facebook will more openly regulate politicians’ speech using curation strategies. The Response concludes that down-ranking is the next major front in the regulation of politicians’ speech.

Introduction

With his contribution to the American University Law Review’s May 2020 symposium Issue on social media and democracy, Professor Joseph Thai joined the growing ranks of legal experts who have called attention to intersections between Facebook’s speech regulation and its business interests.¹ In Facebook’s Speech Code and Policies: How They Suppress Speech and Distort Democratic Deliberation, Professor Thai argued that Facebook’s speech code—its Community Standards—shapes political discourse in the United States and warrants sustained scholarly study.² He focused in particular on Facebook’s policy of excluding “politicians” from its content-based rules.³ This approach, Professor Thai argued, “skews public debate by amplifying the expressive power of already dominant speakers in our society.”⁴

However, shortly after Professor Thai published his Article, circumstances changed. Facebook retreated from its policy of exempting politicians from content-based regulation; it added warning labels to politicians’ posts, employed curation strategies to limit their reach, and eventually removed some posts by politicians altogether.⁵ A main target of these changes was President Trump,⁶ whose posts between May 2020 and January 2021 continually tested Facebook’s resolve to maintain a “light touch.” After a mob seized the U.S. Capitol on January 6, 2021,⁷ Facebook first suspended President Trump’s account,⁸ then shut it down indefinitely.⁹ As of this writing, President Trump is prohibited from posting on Facebook, a complete reversal of the company’s original insistence that it would “protect political speech” by exempting politicians from its content-based rules.¹⁰ Facebook has referred its decision to de-platform President Trump to its Oversight Board, which operates as a sort of private “Supreme Court” that reviews the company’s speech-regulating decisions.¹¹ Some commentators believe the Oversight Board is likely to overturn the ban and restore the former president’s account¹²; in its first five rulings, in January 2021, the Board overturned four of the company’s decisions to take down content.¹³

Building on Professor Thai’s thesis that a pressing need exists for further study of Facebook’s speech regulation,¹⁴ this Response extends his account of Facebook’s treatment of politicians’ speech before and after the 2020 election. It argues that scholars should pay attention to the company’s rapid abandonment of its professed commitment to exempt politicians’ speech from content-based regulation. Though the retreat could evidence incompetence inside the company, I argue that it is better understood as a story about a corporate actor behaving the way that we should expect a corporate actor to behave: articulating “values” when doing so has public relations value, and moving away from those values when the gravitational pull of the company’s profit motive reorients. Facebook’s original choice to publicly announce an exemption for politicians protected its access to users’ data and worked to appease the elected officials who held power over it.¹⁵ That choice may even have protected the company from charges that its curation policies produced the equivalent of in-kind campaign contributions to one candidate or another in an election.

Facebook’s choice to backtrack from the policy, and to employ curation techniques that obscured this choice early on, was also driven by its profit motive.¹⁶ The sensitivity of the company to President Trump’s ebbing political power, and to public opinion, are evident as the story progresses. Do for-profit companies possess free speech values? Does Facebook? Facebook’s rapid tack away from its professed values suggests the answer.

I. Facebook Walks Back Its Hands-Off Approach

When Professor Thai published his Article in May 2020, Facebook was still asserting that it would not apply its content-based speech rules to politicians, citing free speech grounds.¹⁷ Although the company specifically advanced this position with reference to the upcoming November 2020 U.S. elections, it would soon pivot. Closer to the election, in the middle of a pandemic and an upsurge in misinformation about public health and election fraud, Facebook ended its “hands-off” approach to politicians’ speech.¹⁸ After President Trump and others used social media to foment political violence¹⁹ and an insurrection at the U.S. Capitol ended in five deaths,²⁰ Facebook shut down President Trump’s account, ending his ability to speak on its platform.²¹ A brief sketch of this chronology follows.

Facebook’s first step back from its policy exempting politicians from content rules came in March 2020 and concerned a politician outside the United States: Jair Bolsonaro, the president of Brazil.²² That month, Bolsonaro attended a dinner with President Trump at Mar-a-Lago and returned home with a box of hydroxychloroquine.²³ Before the month was over, Facebook had removed content posted by Bolsonaro in which he touted hydroxychloroquine as a COVID-19 cure.²⁴ Twitter also deleted a tweet promoting hydroxychloroquine by Rudolph Giuliani, a Trump political surrogate, around the same time.²⁵ Roughly a week before these first take-downs, an Arizona couple had been sickened—and the husband had died—after ingesting chloroquine phosphate, a chemical used to clean fish tanks.²⁶ The pair had taken the chemical after watching a press briefing in which President Trump touted hydroxychloroquine as a COVID-19 cure.²⁷

In May, Facebook took steps to limit the reach of a paid political ad by an anti-Trump group of Republicans, the Lincoln Project.²⁸ It labeled the ad “partly false,” and downranked its circulation when users shared it.²⁹ Yet a Facebook spokesperson told Wired magazine that Facebook would have treated the ad differently if it had come from a politician rather than an outside group.³⁰ Later that month, Facebook refused calls to take down a post by President Trump that referenced the Black Lives Matter protests and warned, “when the looting starts, the shooting starts.”³¹ President Trump had posted identical language on Twitter, which—like Facebook—allowed him to violate its content rules.³² Twitter left the post up but affixed a warning label that covered the offending language.³³ Mark Zuckerberg, Facebook’s CEO, stated that Facebook had chosen to leave the post up without a label because “the National Guard references meant we read it as a warning about state action, and we think people need to know if the government is planning to deploy force.”³⁴ Thus, Facebook’s reasoning seemed to turn on President Trump’s position as an elected official, rather than his inclusion in the broader category of “politician.”

The following month, in June, Facebook took down posts and ads from the Trump campaign that showed an upside-down red triangle, a symbol used by Nazis—the first instance in which Facebook removed speech from the Trump campaign.³⁵ The ads were paid speech, which might have justified different treatment.³⁶ However, at the end of the month, Facebook’s executives were still claiming that the platform regulated President Trump’s speech, if at all, with a “light touch.”³⁷ “There is an election coming in November[,]” Facebook asserted, “and we will protect political speech, even when we strongly disagree with it.”³⁸

In July, in an extension of its both-sides approach, Facebook began adding labels to posts by both President Trump and presidential candidate Joseph Biden Jr., including one added to a President Trump post alleging a connection between mail-in voting and “#RIGGEDELECTION.”³⁹ The label linked to a U.S. government website about voting.⁴⁰ The move, apparently meant to redirect users to additional information, renewed an unsuccessful approach that Facebook used in 2017 and 2018, in which the company had pushed “Related Articles” to readers to counter false information.⁴¹

At the end of July, Breitbart posted a forty-three-minute video to Facebook, Twitter, and YouTube that made false claims about the effectiveness of hydroxychloroquine as a treatment for COVID-19.⁴² President Trump and his son, Donald Trump Jr., both amplified the video on social media by retweeting it and sharing links to it.⁴³ Within days, all three platforms had removed the video.⁴⁴ However, before they pulled it down, the video was viewed more than 17 million times on Facebook alone.⁴⁵ Just a few days later, Facebook took down a post by the Trump campaign that embedded a video claiming children were “virtually immune” to coronavirus.⁴⁶

In September, Facebook took down advertisements from the Trump campaign claiming that an influx of refugees to the United States would exacerbate the COVID-19 pandemic.⁴⁷ It also announced new limits on politicians’ speech related to the election.⁴⁸ It said it would ban election ads from the campaigns for a week leading up to the November 3 election, and that it would “reject[] political ads that claim victory before the results of the 2020 election have been declared.”⁴⁹ Although it may have appeared at this point that Facebook was singling out politicians’ paid speech (campaign ads) for regulation, the company intensified its regulation of President Trump’s speech the following month, in October, when it removed a post by President Trump that falsely stated that seasonal flu was more dangerous to most people than coronavirus.⁵⁰

The U.S. presidential election took place on November 3, 2020; Democrat Joseph Biden Jr. won the popular vote by more than seven million votes.⁵¹ Immediately after the election, President Trump began posting content that suggested fraud was tainting the counting of votes. Facebook began labeling posts like the one authored by President Trump below.

⁵²

Journalist Charlotte Klein argued in Vanity Fair that such warning labels “are basically doing nothing to slow the spread of false content” online and decried the company’s “incompetence” at handling post-election disinformation.⁵³

On December 2, 2020, a day after Attorney General William P. Barr announced that the Justice Department had found no significant voting fraud in the 2020 election,⁵⁴ President Trump used social media to distribute a “rambling” video that he had recorded in the White House.⁵⁵ The video captured a long speech that President Trump described at the beginning as “possibly the most important speech I’ve ever made,” and in which he asserted, “[t]his election was rigged. Everybody knows it.”⁵⁶ The New York Times described the video as “one falsehood after another about voting irregularities in swing states, attacks on state officials and signature verifications, and false accusations against Democrats.”⁵⁷ Notably, although the video showed the President of the United States speaking in what appeared to be his official capacity—standing behind a lectern in the White House—its production values undercut its authenticity. No audience was seen or heard, he did not take any questions, and cuts in the video suggested some edits.⁵⁸

President Trump posted the full video on Facebook and distributed a two-minute excerpt on Twitter.⁵⁹ Both of the companies left the video available for viewing but both attached disclaimers to it.⁶⁰ The Facebook disclaimer stated: “Both voting by mail and voting in person have a long history of trustworthiness in the US. Voter fraud is extremely rare across voting methods. Source: Bipartisan Policy Center.”⁶¹ Twitter’s disclaimer refuted the video’s content more directly. It stated: “This claim about election fraud is disputed.”⁶²

If Facebook down-ranked President Trump’s video, it did not disclose this publicly. Some of the tools that Facebook could have used to limit the reach of the video include reducing its appearance in users’ News Feeds⁶³ and redirecting searches for terms related to the video to other sources.⁶⁴

As of December 17, 2020, the full video had been viewed on Facebook more than 14 million times; a short excerpt circulated on Twitter had been viewed more than 3.5 million times.⁶⁵ This is millions of fewer views (after more than two weeks) than the Breitbart video touting hydroxychloroquine had received in a matter of days in July.⁶⁶ This suggests that platform curation strategies may have played a role in limiting the reach of—and therefore the number of Americans who watched—the video.

Though Facebook left the video up and gave it only a weak label, the institutional press—including major newspapers—declined to link to the video online and downplayed the speech in their news coverage.⁶⁷ CNN, the cable network, refused to air the speech.⁶⁸ Rather than debunk content from the speech, news outlets simply refused to give it coverage or to engage with the ideas it expressed. Filled with falsehoods, the 46-minute diatribe was one of the most extraordinarily anti-democratic speeches ever given by a modern U.S. President.⁶⁹ For this reason, the video was certainly “newsworthy,” and the press’s refusal to give it much coverage is especially interesting.

On January 6, a mob of pro-Trump insurgents attempted to take over the U.S. Capitol to stop the certification of electoral college votes establishing Joseph Biden Jr., as the winner of the presidential election.⁷⁰ In the preceding weeks, President Trump himself and pro-Trump activists had used social media to spread word of upcoming action on that date.⁷¹ One analysis found that the phrase “Storm the Capitol” was used 100,000 times on social media in the preceding month.⁷² Watchdog groups noticed the surge in social media content promoting a January 6 assault on the Capitol and issued warnings, but neither Facebook nor Twitter removed “Storm the Capitol” content.⁷³

On the afternoon of January 6, 2021 insurgents stormed the Capitol building, leading to the deaths of five people, including a Capitol Police officer.⁷⁴ President Trump produced a short video and posted it to Facebook, Twitter, and YouTube.⁷⁵ The video did two things: it asked rioters to stand down and reaffirmed the central claim of President Trump’s post-election rhetoric, that the election had been stolen from him.⁷⁶ Twitter gave the tweet that embedded the video this label: “This claim of election fraud is disputed, and this tweet can’t be replied to, Retweeted, or liked due to a risk of violence.”⁷⁷ Facebook and YouTube removed the video,⁷⁸ followed by Twitter, which stated that “on balance we believe it contributes to rather than diminishes the risk of ongoing violence.”⁷⁹ Around 7 p.m., Twitter “required the removal” of three tweets by President Trump;⁸⁰ it later suspended President Trump’s @realDonaldTrump account.⁸¹ Facebook suspended President Trump from posting for twenty-four hours.⁸² According to the New York Times, “[a]fter Twitter locked Mr. Trump’s account late Wednesday, Mr. Zuckerberg approved removing two posts from the president’s Facebook page . . . . By that evening, Mr. Zuckerberg had decided to restrict Mr. Trump’s Facebook account for the rest of his term—and perhaps indefinitely.”⁸³

II. Lessons from the Walk-Back

Professor Thai argued that Facebook’s choice to treat politicians’ speech differently from the speech of non-politicians raised free-speech concerns.⁸⁴ As a private company controlling a major online platform, Facebook shaped information flows to millions of Americans before and after the 2020 election. There can be little doubt that Facebook’s decisions about how to regulate speech on its platform implicated “free speech” as the concept is understood by ordinary Americans, millions of whom used Facebook on a daily basis in the months before the election.⁸⁵

In the period from May to December 2020, Facebook became increasingly hands-on about regulating politicians’ speech.⁸⁶ Ironically, when Facebook finally silenced President Trump with an indefinite ban, its decision may have been motivated partly by the concern that his political notoriety—his success as a politician—was the problem, because it amplified the effects of his speech. The political impact of President Trump’s speech made it a liability for Facebook. In the end, despite the company’s early pronouncements, election to the highest office in the United States did not protect President Trump from Facebook’s content-based axe. Or, perhaps it was the loss of the 2020 election that cost him that protection, even while he finished out his first term of office.

Facebook has not disclosed much about the reasoning behind its policy change, especially considering how strongly it once celebrated its hands-off policy. Since Facebook is not a state actor, it is exempt from the sort of judicial scrutiny that would require it to publicly identify the interests served by its rules. As a result, we can only guess about the company’s motives.

Was there ever any bona fide reason to treat politicians’ speech differently from the speech of other users? Doing so might let voters receive candidates’ speech free from third-party filters, perhaps improving the quality of information they can use to make voting decisions, which is good for democracy, while simultaneously increasing trust in the platform companies that turn their filters off, which is good for those companies’ business. (Of course, as I discuss below, it might not be true that transmitting an unfiltered stream of candidates’ speech improves voters’ information).

Separately, citizens might benefit when platforms take a hands-off approach to the speech of elected officials. The speech of a public officeholder might shed light on the workings of government itself. Mark Zuckerberg suggested as much when he defended Facebook’s treatment of President Trump’s statements about the Black Lives Matter protests in May 2020. And, as Professor Thai pointed out, the elected officials “whom Facebook frees from the constraints of its speech code and fact checking, and to whom it sells the keys to its ad targeting kingdom,” hold power over the company and “could wield that authority in ways that threaten its existence.”⁸⁷ At the time that Professor Thai published his Article, President Trump was Facebook’s top spender on political advertisements.⁸⁸ Post-election results show that this trend continued; between January 1, 2019 and Election Day 2020, President Trump, the losing candidate, spent more on Facebook ($115 million) than his opponent, Joseph Biden, the winner, did ($106 million).⁸⁹

Professor Thai’s suggestion that Facebook designed its speech rules to give special protection to people wielding power over it appears particularly incisive in light of subsequent events. As President Trump’s electoral prospects dimmed in the lead-in to the November election, the company increased its regulation of his speech. And Facebook only de-platformed him after he had lost the election, when his political power was weak.

One of the big take-aways of the events of January 2021—the narrative likely to make it into the history books—is that Facebook’s lax controls made it possible for President Trump to use Facebook’s platform to cause harm and to foment real-world political violence. President Trump proved willing to post harmful content: false information about public health, thinly-veiled racist statements, manipulative and self-serving claims about mail-in voting, and, after the election, blatant lies designed to undercut the legitimacy of his successor’s presidency.⁹⁰ He was even willing to use Facebook to drum up support for a violent insurrection.⁹¹ When faced with a politician willing to make irresponsible and false claims, Facebook’s speech code not only failed as ground rules for healthy discourse, but also contributed to and amplified anti-democratic rhetoric and incitement, thereby inflicting harm. Facebook underestimated the potential for a politician to use its platform to undermine the public interest and to swiftly escalate posts into real-world physical violence.

Professor Thai wrote that Facebook’s hands-off approach to politicians’ speech could not withstand First Amendment scrutiny.⁹² He described Facebook as “labeling and throttling” content in connection with its fact-checking processes and argued that this, on top of the company’s microtargeting strategies and basic Community Standards, amounted to censorship of the news.⁹³ While I do not agree that such strategies amount to censorship, which I would define as the complete removal of speech, I take him to mean that these acts suppress speech in ways that virtually assure that some people will never see it.

Professor Thai raised the concern that Facebook’s exemption of politicians’ speech from its content-based rules “creates a two-tier speech platform that . . . ‘treats people who aren’t politicians as second-class citizens,’”⁹⁴ amplifying the “powerful” speech advantages already enjoyed by national politicians. Is popular sovereignty meaningful in a community in which political leaders comprise a peerage with enhanced rights? Professor Thai’s insight is important: any bright-line distinction between “politicians” and the rest of us is anti-democratic. It habituates us to political discourse in which the interests of those with less political power are functionally subordinated to those with more of it. Since one’s political stature can be enhanced by wealth or corporate political support—without electoral success—the two tiers reinforce some of the most unpopular and plutocratic aspects of our existing political system. If content-based rules restrict your speech on Facebook, but not the next person’s, why bother speaking on a controversial topic at all? Let the politician speak, and “like” or amplify that person’s voice. Instead of expressing yourself, which takes thought, initiative, and risk, you can play a different, more passive role in discourse, up-ranking or down-ranking the speech of other, more important actors. It is hardly democracy in action, but it is commerce in action. After all, Facebook and other social media companies derive value from data about your behavior—how long you spend reading a post, for example—even if you never pen a sentence.

Professor Thai also argued that the two-tier approach “promote[d] a race to the bottom in which willing [politicians] may take the low road of spreading lies that . . . [are] politically advantageous.”⁹⁵ Along this line, Steven Levy, of Wired magazine, has argued that a policy that expressly allows politicians—and no one else—to lie may reinforce longstanding mistrust of politicians as liars.⁹⁶ In other words, the “two-tier” approach itself might undermine American’s faith in their democratically elected representatives. These are powerful arguments and, if they are right, the informational benefits of the two-tier approach to voters or citizens may be offset by the drawbacks.

Facebook’s policy reinforced the idea that politicians’ speech has special importance—that it is worthier of users’ attention than ordinary people’s speech. The institutional press has always “covered” politicians’ speech, as well as the speech of business and civic leaders, with an intensity that it does not give to the expressions of ordinary citizens.⁹⁷ “Vox pop” or “man-on-the-street” coverage used to be a staple of journalism, but increasingly has been replaced with embedded Tweets.⁹⁸ (Ironically, a 2012 NPR poll found that many Americans felt that “vox pop” features “take time and space away from more valuable analysis and fact-checking”).⁹⁹ But press coverage of politicians’ speech has always offered context, interpretation, and often pushback to the speech of candidates and elected officials. Unlike social media platforms, which function as speech conduits, the institutional press has utilized a more active approach in transmitting the speech and ideas of politicians to the public.

Professor RonNell Andersen Jones has argued that the President of the United States has an obligation to engage with journalists, particularly over contested matters, as part of a “constitutional system of dialogue between the press and the executive.”¹⁰⁰ Instead of publishing the President’s statements word-for-word, journalists interact with the President, framing and asking questions of him or her, in a process that serves both to challenge the President’s positions and to clarify and present the President’s ideas to the public. We hear often about how politicians can use social media to speak “directly” to the people, without the intermediation of the press, and how this is a good thing.¹⁰¹ Less attention has focused on the Trump administration’s shift away from dialogue with the press in favor of unilateral statements delivered via social media, a shift that served to insulate the President from direct interaction with a corps of experienced White House reporters. The decision by Facebook and other companies to start regulating politicians’ speech represents a new phase in the evolution of social media—one in which these companies will potentially step into a new, mediating role alongside the institutional press.¹⁰² Importantly, of course, social media companies are governed by different ethics and incentives than press companies. For example, they are incentivized to preserve their access to users’ data on a near-constant basis.

Down-ranking, which has become a favored tactic of social media companies to combat false and divisive content, has not been the subject of sustained scholarly analysis. Down-ranking is unique to social media because of the platforms’ ability to covertly amplify or throttle users’ speech; it has no real antecedent in the history of speech regulation. It is a relatively new strategy on social media, and probably does not date back much further than 2015.¹⁰³

When it comes to democratic deliberation, down-ranking may be the most dangerous tool that social media companies wield. Thus, it is particularly interesting that, in 2020, Facebook gave up its previously generous treatment of politicians’ speech in favor of down-ranking strategies that lacked user transparency. Free speech advocates will focus on burdens to speakers and listeners—how a speaker cannot know whether posts are being down-ranked and how a listener might miss down-ranked content that is informative. But there is also an interesting story here about democratic dialogue—how Facebook’s choice to curate politicians’ speech, rather than simply remove it, shapes political discourse. Until recently, our democratic system had relied upon a robust institutional press with a constitutionally-contemplated role in publicly dialoguing with and challenging the President’s speech. Social media platforms initially seemed to offer politicians a way to circumvent the press and speak, without a filter, directly to the People. But that short-lived regime is over.¹⁰⁴ Even if Facebook’s Oversight Board restores President Trump’s account, that account will be subject to curation strategies that Facebook applies to politicians’ speech. Down-ranking is the next major front in platform regulation of politicians’ speech.

Conclusion

Professor Thai’s Article, Facebook’s Speech Code and Policies: How They Suppress Speech and Distort Democratic Deliberation, highlights the need to think more cogently about how the surveillance capitalism business model distorts democratic deliberation. This Response has shown that although Facebook pivoted after the publication of Professor Thai’s Article and changed its approach to politicians’ speech, the concerns Professor Thai raised in his Article remain. Facebook’s decision to treat politicians’ speech differently from the speech of ordinary Americans proved unworkable in the Trump era and Facebook ultimately walked back its much-trumpeted “hands off” approach. We are entering a new era in which we must grapple head-on with the reality that platform companies regulate politicians’ speech, in obvious ways and through less-obvious methods of curation.

Share this post

Contact Us

Join Our Mailing List