For the past few weeks, politically-engaged Twitter users have either lamented or praised Elon Musk for his new ownership of Twitter, a microblogging social media website that has an outsized influence on much of American politics. Recently, he handed over a tremendous cache of internal files to a handful of free speech journalists like Bari Weiss and Matt Taibbi as part of his effort to clear the air surrounding the company’s opaque content management practices. 

While Musk is the hero, the villains are the ousted Vijaya Gadde and Yoel Roth, who seemed to have managed content in a highly ideological and partisan manner. They overwhelmingly favored left-wing causes and routinely fielded requests from Democratic campaigns and officials who wished to dictate what material Twitter would and would not allow. Recently, Weiss and Taibbi have revealed that Twitter freely engaged in “shadowbanning” or “deamplification,” allowing moderators to apply various methods for preventing conservative messengers from reaching their audiences. These could include preventing their tweets from “trending,” keeping their content from showing up in searches, or imposing suspensions with the thinnest of rationales.

These revelations follow years of new labeling of posts made to the site. Specifically, Twitter warned users that posts contained inaccurate information about the 2020 presidential election or COVID-19 policies. Moderators would “shadowban” or suspend accounts that violated what Twitter took to be the right kind of information to share on its platform. Those decisions were influenced by government agencies concerned about the propagation of bad ideas, such as the wrong-headed notion that the 2020 election was “stolen,” or the much more reasonable position that school lockdowns entailed greater costs than government officials admitted. 

The very topic of Twitter might cause some readers to roll their eyes, but the significance of these discoveries should not be overlooked merely because one finds Twitter unpalatable. What we are seeing in the so-called “Twitter Files” is what media scholar Zongyi Zhang calls “the infrastructural turn in platforms.” As a platform for discussion becomes more relevant to government interests, governments seek to control that platform, hoping to direct it toward their own ends rather than those either of the users or owners. 

Infrastructuralization: A Chinese Technique on American Shores

Zhang’s inquiry is not focused on Twitter; it examines the trajectory of TikTok as it is used in China. Zhang finds that the government initially treated TikTok in China as more or less a frivolous nuisance, given that most content was, as it is in America, based around dancing and lip-synching short videos. The founder, Bytedance CEO Zhang Yiming, said in 2018 that his platform was not just an “information aggregation and distribution platform” but a “technology company.” These terms indicated that Yiming had embraced a neutrality ethic, in which he was happy to develop the platform in a way that drew in new users, which in turn opened greater opportunities to draw from their metadata new sources of revenue. In essence, TikTok was no different from other social media platforms like early Twitter, Facebook, and the like, on which content moderation was kept to a minimum in order to encourage new growth in the user base. 

The Chinese government was not pleased and began that same year an effort to stymie the platform’s embrace of content neutrality. State media pointedly criticized Yiming’s approach, and TikTok users discovered they were unable to share their videos over other social media and messaging apps like WeChat. Tech giant Tencent throttled access. Yiming escaped digital purgatory by coordinating content moderation with the Chinese government. The Chinese government wanted TikTok to stress certain content in its launch of a “video encyclopedia” that stressed traditional Chinese culture and education over its previously popular content. Yiming complied and his company has since found itself freed from constraints and returning to its previous earnings. 

The above is a compressed narrative that showcases the “infrastructural turn.” Once TikTok was simply a neutral platform for sharing videos, but now it is an extension of the government’s broad social and cultural priorities. 

Something similar seems to have happened at Twitter. At first, Twitter was a relatively small social media site that had the character of a bulletin board for the subcultures that found a home there. Perhaps the earliest (forming around 2009) was “Black Twitter,” which generated much of the vernacular used on the site, from terms like “drag” (to mock someone justly) or “to have receipts” (to have evidence against a person). There was also the formation of “Weird Catholic Twitter” or “Weird Christian Twitter” that featured a hodgepodge of communists, integralists, and “normies” (people without very online, bespoke ideological modifications to their faiths). These subcultures are the bread and butter of social media networks.

The success of Twitter as a platform did not necessarily translate into profits. Twitter has always struggled with monetizing the platform, and it has branched out in such ventures like Vine (launched in 2013 and closed in 2017) and other short-lived (2020-1) Twitter features like “Fleets,” similar to the “stories” on Facebook and Instagram. Twitter did find success, however, in certain “superusers” whose content drove people to the platform in droves—chief among them being former President Donald J. Trump. Immediately prior to Musk’s acquisition of Twitter, some people noted that the number of superusers had declined, spelling possible trouble for the platform. That was part of a broader media downturn following Trump’s departure from the presidency, as well as the return to relative normalcy after most COVID restrictions came to an end.

As Matt Taibbi explains, the fall of 2020 through 2021 was the period of “erosion of [moderation] standards within the company.” This “erosion” is the same as Twitter’s “infrastructural turn.” In November of 2020, Chaya Raichik started a Twitter account dubbed “Libs of TikTok” that posted weird, sometimes unsettling progressive TikTok videos, along with occasional commentary on them. Throughout 2021, as the account gained traction, Twitter moderators put the account through multiple shadowbanning techniques and repeatedly imposed temporary suspensions. 2021 began with banning Trump from the platform for his encouragement of the January 6 riot. During the early fall of 2021, Twitter used shadowbanning techniques to suppress messages by Stanford University professor Jay Bhattacharya for his warnings against COVID-19 lockdowns. As the Twitter files have revealed, the Twitter moderation team greatly expanded its mandate to suppress “disinformation” and the tools to implement this mandate. As Bari Weiss and others at The Free Press conclude:

The bottom line was: If someone in [Site Integrity Policy, Policy Escalation Support team] didn’t like you, they could find a way to shut you down.

That was the subtext of this direct message from Yoel Roth to colleagues in early 2021, written in a nearly indecipherable Twitter-ese:

“A lot of times, SI [Twitter’s Site Integrity team] has used technicality spam enforcements as a way to solve a problem created by Safety [team at Twitter] under-enforcing their policies.”

In a follow-up message with a colleague, Roth said he was looking for ways to marginalize accounts that had fallen into disfavor without banning them outright. Banning a user, especially a prominent one with many followers, generated bad publicity. Possible workarounds included “disabling engagements” and “deamplification/visibility filtering.”

Roth claimed the sub rosa censorship amounted to a public service. “The hypothesis underlying much of what we’ve implemented,” he said, “is that if exposure to, e.g., misinformation directly causes harm, we should use remediations that reduce exposure.”

Without informing either users or the suppressed accounts, Twitter moderators would suppress the reach of certain users on their platform and then lie about doing so—even before Congress. 

The comparison between Chinese TikTok and Twitter is not perfect. The former is a case of imposition; the latter is one of collaboration. Leading figures within Twitter like Gadde and Roth seemed happy to work with the US government to assist in moderating against “stop the steal” conspiracy theorists, dissenters to the government’s COVID regime, and accounts lampooning “woke” Tik Tokers. This attitude reflects an elite monoculture shared between the upper echelons of Silicon Valley and the administrative state, though it is important to note that Roth and Gadde, in exercising moderation oversight, behaved as private actors in a private company.

This difference matters less, though, when it becomes clear that none of these relationships were disclosed, that moderation was heavily one-sided, and that these connections were denied under oath before Congress. In addition, the unintended consequence of these efforts was to provide fuel to the fire for conspiracy theorists of all kinds as the platform suppressed genuine scientific disagreement over how to manage the response to the virus. Perhaps worst of all is the co-mingling of these two issues, since skepticism of the 2020 election outcome was unwarranted while doubt concerning COVID lockdowns was perfectly valid. It is hard to persuade the public to discount conspiracy theories when, sometimes, the theories are true.

The Press and a Free People

Musk has referred to these matters as First Amendment violations, but this is not correct from the perspective of constitutional law. It is better to say that these issues touch upon the original reason for establishing freedom of the press in the first place. A “free press” contrasts with a “licensed press.” During the seventeenth century, religious and political dissenters in England objected to the requirement to procure a license to print publications, and the suppression of their efforts was partly responsible for the civil strife that engulfed the nation for the better part of that century. The issue remained relevant into the eighteenth century, prompting the once-famous Cato’s Letters to demand a press free of a licensing regime in the name of encouraging public discussion.

As Michael Kent Curtis has persuasively documented, Cato’s Letters, or at least the ideas expressed in them, had a strong influence on the Founders, who had fomented rebellion as well as a Constitution through the use of a free press. Far from licensing the press, the Founders agreed to subsidize newspapers with low postage rates that incurred losses covered by postage on letters. The legacy was a society in which newspapers flourished. Alexis de Tocqueville was not overly impressed with the result, but could not think of a superior arrangement. Newspapers were so prominent a part of the American culture he saw in 1831 that he remarked how the American frontiersman wandered into the wilderness with a Bible and a newspaper.

Expediting the spread of the newspaper were technological developments that rendered its production much cheaper at small and large scales. The Gordon jobber (which some still use today) is cheap and could produce small-scale publications for local civic and political associations with ease. The steam press could churn out newspapers by the thousands for regional or even national distribution. As a result, censoring these newspapers was nearly impossible. When Elijah Lovejoy, an abolitionist, ran afoul of Democrats in Missouri in 1835, he found his printing press destroyed. He quickly procured a new one and began printing across the border in Illinois until in 1837 a mob found his press in a barn, shot him to death, and burned the barn to the ground. While Woodrow Wilson found it relatively straightforward to use war powers to censor radio, newspapers were a constant nuisance. The reason was that the stakeholders in radio were small enough to organize; printers were in the thousands. He responded by using the Committee of Public Information to uncover and suppress seditious speech against the First World War and the government, but his efforts were limited and produced some of the first Supreme Court cases ruling in favor of press freedoms.

The rise of broadcast media was simultaneous with the rise of the administrative state, and the creation in 1927 of the Federal Radio Commission, now the Federal Communications Commission, brought a licensed press back to the United States for the first time. For radio and television stations to broadcast legally over airwaves, they were required to secure a permit from the FCC and comply with the terms of use. Like all early platforms, these terms were somewhat limited at first, but by the 1950s, the infamous “Blue Book” of recommended content creation for broadcast platforms began to structure what American radio and television should look like. By no means was the FCC licensing regime inevitable—it was an ideological project. As Ronald Coase wonderfully illustrated in his paper on the subject, the competition for use of airwaves could have been easily managed through private contract, and by extending common law solutions to land use rights.

Social media combines both newspaper and broadcast features. Like newspaper printing of old, social media has a low barrier to entry for users and producers. Like broadcast media, it has incredible reach because its material is easy to access. No longer does one have to wait for the paper to arrive on a doorstep; one merely opens an app. Low barriers to entry and user-generated content are what define social media. This creates a problem for print, however, because its increasing dependence on digital distribution over social media exposes print publications to the censorship that they successfully avoided for so long.

The decline of broadcast in favor of digital technologies has created new mandates for the FCC, but we are now no longer seeing these questions addressed either by legislative efforts from elected officials or by the adjudication of judges. Now they are the province of law enforcement and other administrative agencies, which directly coordinate with digital platforms to build government infrastructure into their platforms. This development is worse than licenses—at least they are publicly disseminated. Now, the de facto licenses are rooted in private communications between agencies and corporate leaders, and the full extent of these regulations is known to us only because of Musk’s revelation of them. These communications are different from the Chinese interventions in TikTok only by degree, not in kind.

This kind of government infrastructuralization is possible because there are relatively few social media platforms, and because there is considerable ideological overlap between the employees of government agencies and those running the major tech companies. I do not wish to dwell on the ideological overlap too long—many, many others are already doing so. Rather, I want to draw attention to what a major loss it is for Americans to have abandoned newspapers as a mode of communication. The digitization of print meant that it became necessary to funnel print stories through a very small number of digital networks in order to reach audiences. Thus, printers lost the advantage of being so large in number as to avoid government censorship. When the New York Post broke the Hunter Biden laptop story, their story was suppressed on Twitter, and Twitter moderates even temporarily banned their official account for sharing it. 

Return to Print?

The introduction of infrastructuralization into platforms is not so much a violation of free press in statute or case law—or at least, it is not only that. It is a direct violation of a foundational American principle, by which free people are permitted to consult whatever publications they wish for political information. The argument against a free people doing so has always come down to the same rationale: the information can be dangerous. The only thing that makes dangerous information more dangerous, however, is its suppression, as people begin to believe that the story is not just dangerous but damaging to government interests. People lose faith in a government that seeks to suppress political information against it, no matter how spurious; ironically, the suppression itself grants that spuriousness a veneer of truth.

Political solutions to this problem are difficult and not within the scope of this essay. Perhaps the best first step an individual can take is to bypass social media as a primary source for news and information and, instead, go directly to one’s preferred publications online or in print. Given that online stories can be taken down or edited without notice, print is arguably the best method to preserve free discussions. As streaming sites have begun to pull digital downloads and books, media consumers have relearned the value of having hard copies of their favorite music and movies. The same can be said about what we read.