By Corbin K. Barthold, Internet Policy Counsel and Director of Appellate Litigation at TechFreedom. Mr. Barthold previously served as WLF’s Senior Litigation Counsel.


In April, Justice Clarence Thomas, writing for himself in an otherwise unrelated case, speculated about whether large social media websites should be treated as common carriers. The following month, Florida Governor Ron DeSantis signed into law SB 7072, a sweeping set of restrictions on how the companies that run such websites shall moderate what is said on them. SB 7072 forces the likes of Facebook and Twitter to host various categories of speech against their will. Florida may do this, SB 7072 says, because “social media platforms” may “be treated similarly to common carriers.”

SB 7072 was bound to get challenged in court, and that litigation, in turn, was bound to test the common carriage theory put forth by Justice Thomas. So it has come to pass. Two groups of internet companies promptly sued, a judge issued an order preliminarily enjoining most of the law, and Florida appealed. Both the judicial opinion, written by federal District Judge Robert Hinkle, and Florida’s opening brief on appeal, filed earlier this month in the U.S. Court of Appeals for the Eleventh Circuit, address whether it makes sense to treat social media as common carriage.

What can be learned from these discussions of the common carrier theory? Judge Hinkle concludes that social media websites are somewhat like common carriers, but ultimately more like traditional speakers fully protected against government-compelled speech (hence the preliminary injunction). Florida, naturally, argues the common carrier theory to the hilt, relying heavily on Justice Thomas’s work along the way. Neither the judge nor the state depicts common carriage in a way that’s at once accurate, useful, and convincing. Identifying the holes in their thinking returns us to a conclusion that would, in a less anxious time, be obvious to all. Websites—even large ones that host the speech of others—are engaged in expressive conduct protected by the First Amendment.

Judge Hinkle’s Good (But Flawed) Opinion

Judge Hinkle reached the right conclusion—SB 7072 violates the First Amendment. Moreover, his opinion makes a number of astute, laudable, and impeccably correct points. Here are a few:

  • “The State has asserted it is on the side of the First Amendment; the [internet companies] are not. It is perhaps a nice sound bite. But the assertion is wholly at odds with accepted constitutional principles.”
  • “The internet provides a greater opportunity for individuals to publish their views … than existed before the internet arrived.”
  • “The [internet companies] assert, … with substantial factual support, that the actual motivation for this legislation was hostility to [the largest social media websites’] perceived liberal viewpoint.”
  • “Leveling the playing field—promoting speech on one side of an issue or restricting speech on the other—is not a legitimate state interest.”
  • SB 7072 “comes nowhere close” to passing First Amendment scrutiny.

When it came to common carriage, however, Judge Hinkle hedged. The parties had presented him five Supreme Court decisions to guide his analysis. Three of those decisions came from the internet companies:

  1. Miami Herald v. Tornillo, 418 U.S. 241 (1974), strikes down a Florida law that required a newspaper to print a political candidate’s reply to the newspaper’s unfavorable coverage.
  2. Hurley v. Irish-American Gay, Lesbian and Bisexual Group of Boston, 515 U.S. 557 (1995), holds that a private parade has a First Amendment right to exclude some groups from participating.
  3. Pacific Gas & Electric Co. v. Public Utilities Commission of California, 475 U.S. 1 (1986), blocks a state from compelling a public utility to include certain disclosures in its billing envelopes.

The upshot of these decisions is that (as Hurley puts it) “a speaker has the autonomy,” under the First Amendment, “to choose the content of his own message.” This is, at bottom, a right (in Miami Herald’s words) to “editorial control and judgment” over the speech one hosts.

The two decisions Florida raised are:

  1. Rumsfeld v. FAIR, 547 U.S. 47 (2006), which upholds a law requiring law schools, on pain of losing federal funding, to host military recruiters.
  2. PruneYard Shopping Center v. Robins, 447 U.S. 74 (1980), which requires a shopping center, in obedience to the California Constitution, to let students protest on its private property.

These cases show that one speaker can sometimes be required to host another speaker, if (in Rumsfeld’s words) doing so does not “interfer[e]” with the host speaker’s “desired message.”

After comparing, on the one side, Miami Herald, Hurley, and PG&E, and, on the other, Rumsfeld and PruneYard, Judge Hinkle concluded that social media websites fall “in the middle” between being “like any other speaker” and “like common carriers.” In reaching this conclusion, however, Judge Hinkle focused on whether such websites “use editorial judgment” in “the same way” as the entities at issue in those cases. That’s not the right question.

Similarity to the precise kind of curation or editing done by the entities addressed in Miami Herald, Hurley, and PG&E does not inform whether social media has a First Amendment right to editorial control. We already know that social media has that right. We know it because Reno v. ACLU, 521 U.S. 844 (1997), tells us so. “[O]ur cases,” Reno says, “provide no basis for qualifying the level of First Amendment scrutiny that should be applied” to the internet. As far as the First Amendment (and binding Supreme Court precedent) is concerned, edge providers on the internet are, in fact, “like any other speaker.”

Judge Hinkle concluded that, because social media websites at least act more like the entities in Miami Herald, Hurley, and PG&E than like the entities in Rumsfeld and PruneYard, SB 7072 is “subject to First Amendment scrutiny.” He then proceeded to enjoin most of SB 7072 for being blatantly content- and viewpoint-based and failing to overcome strict scrutiny. Judge Hinkle was right that SB 7072 is egregiously discriminatory, and he was right to enjoin the government from enforcing it. Even so, he missed an entire other avenue by which SB 7072 violates the First Amendment. What Miami Herald, Hurley, and PG&E establish is not simply that a law compelling social media companies to host certain speech is “subject to First Amendment scrutiny,” but that such a law presumptively violates the First Amendment by forcing those companies to “alter the expressive content” (as Hurley says) of their websites.

Judge Hinkle thought it important that much of the content on a social media website is supposedly “invisible to the provider.” Given that his entire exercise in comparing “editing” by social media with “editing” under Miami Herald, etc., was unnecessary, however, it should come as no surprise that his “visibility” distinction, raised as part of that unnecessary exercise, is irrelevant and illusory. Indeed, Judge Hinkle cut from whole cloth the proposition that content “visibility” affects an entity’s right to editorial control.

What’s more, the proposition is perverse. The more material a website blocks, it suggests, the stronger the site’s First Amendment protection becomes. The First Amendment contains no such “use it or lose it” trapdoor. “In spite of excluding some applicants,” the parade in Hurley was “rather lenient in admitting participants.” But it did not “forfeit constitutional protection simply by combining multifarious voices.”

Finally, the proposition is simply wrong. A large social media website’s first round of editorial control might be wielded via algorithm; the content at issue is no less “visible” to the website (nor the website’s editorial choices less deserving of First Amendment protection) for that. And content is certainly not “invisible” after it’s been posted. Material that, once published, is reported, and found to be objectionable, is regularly labeled, answered, de-amplified, downgraded, hidden, blocked, or deleted. Judge Hinkle never explained why, under the First Amendment, the timing of these varied displays of editorial control—their being ex post as opposed to ex ante—should matter. As anyone will understand after listening to a few hours of talk radio—in which the station lightly “screens” calls in advance, yet retains the (much needed) right to cut off callers at will—it does not.

Florida’s Bad Brief

If Judge Hinkle’s intellectual flirtation with common carriage is a flaw in an otherwise shining opinion, Florida’s treatment of the topic is a rotten egg in a nest of fallacies. For example:

  • Florida asserts that social media websites must present a “unified speech product” to enjoy First Amendment protection, a claim made in naked defiance of Hurley and its “multifarious” parade.
  • Florida seems to believe that advertisers, civil rights groups, the (old school) media, and the public at large will stop holding social media websites responsible for the speech they host if the sites simply “speak on their own behalf”—presumably more than they already do—and “make clear their own views.” This claim is naïve at best.
  • Florida says that “systematic examinations” reveal instances in which websites “apply their content standards differently” to “similarly situated,” but politically distinct, users. Note that it’s the “examination,” rather than the supposed bias, that claims to be “systematic.” In any event, one can only marvel at Florida’s sangfroid, as it announces that it has surmounted the numberless fine distinctions and shades of context that bedevil even basic content moderation.

This much can be said for Florida: whereas before the trial court, it pressed Judge Hinkle to consider common carriage through a lens of strained analogies—law schools (FAIR) and shopping malls (PruneYard), after all, are not literally common carriers—on appeal it turns to factors that are (for better or worse) widely considered traditional indicia of common carriage. There is no straightforward and widely accepted definition, in the courts or elsewhere, of what common carriage is. Regardless, tacking the discussion toward these tokens of common carriage brings social media websites no closer to qualifying as common carriers.

Common carriers tend, Florida correctly notes, to hold themselves out as “serv[ing] the public indiscriminately.” “The businesses regulated” by SB 7072, the state then adds—now going astray—“hold themselves out as platforms that all the world may join.” Although it might indeed be said that the websites welcome “all the world” to join, whether one gets to stay is contingent on one’s complying with the sites’ terms of service. Gov. DeSantis has claimed that social media websites “evade accountability” by “claiming they’re just neutral platforms.” Actually, these websites are by no means “neutral” about violence, harassment, and hate speech, all of which are widely banned.

Even if the websites did hold themselves out as serving the public indiscriminately (they don’t), the “holding out” theory of common carriage is conspicuously hollow. As Professor Christopher Yoo observes, a “holding out” standard is easy to evade. Say SB 7072 went into effect, and the websites responded by tightening their terms of service further, thereby making clear(er) that they do not serve the public at large. What then? Rather than admit how badly its law had backfired in its attempt to force the websites to host unwanted speech, Florida would probably declare that the websites are common carriers because the state has ordered them to serve the public at large. Such a declaration would confirm that the “holding out” theory is empty at best, and circular at worst.

Florida suggests that social media websites may be treated as common carriers because they are “clothed” with “a jus publicum.” Unsurprisingly, it doesn’t press the point. The Supreme Court has said that whether a business serves a “public interest” is “an unsatisfactory test of the constitutionality of legislation directed at [the business’s] practices or prices.” Even Justice Thomas concedes that a “public interest” test for common carriage “is hardly helpful,” given that “most things can be described as ‘of public interest.’”

More heavily does Florida lean on a claim that social media websites can be treated as common carriers because of their (purported) market power and (supposed) ability to control others’ speech. The first problem on this front is the brute legal fact that an entity does not forfeit its constitutional rights by succeeding in the market. The Supreme Court accepted that the Miami Herald enjoyed near-monopoly control over local news; yet the newspaper retained its First Amendment right to exercise editorial control and judgment as it saw fit.

This is not to say that media firms, social or otherwise, are above the antitrust laws. A newspaper that uses its market power to inflict economic pain on a rival—one that, say, convinces advertisers to boycott, and thereby bankrupt, a local radio station—is inviting antitrust liability for its business practices. It is to say, however, that the right to reject speech for expressive reasons travels with a company, like a shell on a turtle, wherever the company goes—even if the company, like Yertle, is king of the pond.

If that were all there is to say about social media, monopoly, and free speech, SB 7072’s supporters could be forgiven for some griping about the demise of their unconstitutional law (though fall it still would). But the reality is that the social media market is as lively as ever. It continues to offer a wide array of useful, differentiated, and rapidly evolving avenues of expression and communication. If you’re convinced that “Big Tech” is “out to get” Republicans, you can do your blogging on Substack, your posting on Parler or Gab, your messaging on Telegram or Discord, and your video watching and sharing on Rumble. And anyone who claims, as Florida does, that network effects will ultimately thwart this competition must grapple with the astonishing rise of TikTok.

As for the major players’ alleged “control” over speech, Facebook and Twitter are not, as Florida would have it, “like telegraph and telephone lines of the past.” The internet, Reno v. ACLU explains, is not “a ‘scarce’ expressive commodity. It provides relatively unlimited, low-cost capacity for communication of all kinds.” Even the largest social media websites are just a piece of that “relatively unlimited” world of “communication.” As a (conservative) commentator, Charles C.W. Cooke, recently put it, social media websites are “equivalent not to the telegraph line,” but to a few “of the telegraph line’s many customers.” They are just a handful of “website[s] among billions.”

The receipt of special privileges from the government can nudge a business toward common carrier status. Florida claims that Section 230 is such a privilege, but it is not. “Section 230 helped clear the path for the development of [social media],” Florida reasons, “as the government did generations ago when it used eminent domain to help establish railroads and telegraphs.” True enough, businesses that employed property acquired through eminent domain sometimes had to operate as common carriers. It does not follow that Section 230, which broadly protects all websites for hosting speech that originates with others, creates a similar quid pro quo obligation. There are several problems with the comparison:

  • Section 230 was not a gift to “Big Tech” (or any other select group). It applies to every internet website and service. If Section 230 doesn’t turn a blog (or Yelp, or the Wall Street Journal’s comments sections, or an individual social media account) into a common carrier, it’s unclear why it should turn Twitter or Facebook into one.
  • Section 230 simply ensures that the initial speaker is the one liable for speech that causes legally actionable harm. It is not a “privilege” akin to when the government hands a business real property for exclusive use as a railroad or a telegraph line.
  • Far from being a sign that the government wants social media websites to act as common carriers, Section 230 is a sign that it wants them to act as discerning editors. Section 230 ensures that a website can curate and edit content without (in most cases) worrying that doing so will trigger liability.

If the federally enacted Section 230 is the quid, by the way, why should a state government get to impose the quo? The history of common carriage in the United States, going back to the Interstate Commerce Act of 1887, is one of aiding interstate commerce by setting and enforcing national standards. Precisely because they were regulated as common carriers, telegraph companies were not subject to regulation by the states. Even if Section 230 could serve as the basis for common carriage rules, it couldn’t serve as the basis for common carriage rules imposed by Florida.


So what have we learned? We’ve seen that various arguments in favor of the common carrier theory don’t work. We’ve seen that the orthodox view, under which social media websites enjoy a First Amendment right of editorial control, remains sturdy and sound.

At the outset of his opinion, Judge Hinkle noted that SB 7072 “compels [social media] providers to host speech that violates their standards—speech they otherwise would not host.” We can be confident that this is, and will remain, a violation of the First Amendment.