In September of last year, Texas passed a law declaring that large social media platforms are “common carriers” subject to onerous regulations over who and what they can host. The law prohibits platforms from removing, demonetizing, or blocking a user or a piece of content based on the viewpoint expressed. Platforms found to violate this requirement face liability for each piece of content they remove.

NetChoice, a trade association of online businesses, sued to block the law on First Amendment grounds, among other claims. A federal district court in Texas ruled in favor of NetChoice and preliminarily enjoined the law. Texas appealed to the Fifth Circuit, where Cato filed an amicus brief supporting NetChoice. After the Fifth Circuit let the law go into effect without a written opinion, NetChoice filed an emergency petition to the Supreme Court (which Cato again supported), and the high Court put the stay back in place by a 5–4 vote.

Last Friday, however, the Fifth Circuit finally issued a written opinion on the merits and lifted the stay once again, holding (by a 2–1 vote on the key issue of the speech-hosting requirements) that the law complies with the First Amendment.

The Fifth Circuit’s opinion is notable and startling for its complete rejection of a First Amendment right to editorial freedom, a right that the Eleventh Circuit recently affirmed in a decision blocking a similar Florida social media law. Indeed, on this crucial issue, the Fifth Circuit got exactly wrong what the Eleventh Circuit got right, and the implications could ripple far beyond just social media.

The key Supreme Court case that has been understood to establish a right to editorial freedom is Miami Herald v. Tornillo (1974), which struck down a Florida law mandating that newspapers print responses to negative editorials about political candidates. The Tornillo opinion concluded with this crucial passage, which summed up its holding:

The choice of material to go into a newspaper, and the decisions made as to limitations on the size and content of the paper, and treatment of public issues and public officials—whether fair or unfair—constitute the exercise of editorial control and judgment. It has yet to be demonstrated how governmental regulation of this crucial process can be exercised consistent with First Amendment guarantees of a free press as they have evolved to this time.

This holding, along with that of other cases which built on Tornillo, led the Eleventh Circuit to find that “a private entity’s decisions about whether, to what extent, and in what manner to disseminate third-party-created content to the public are editorial judgments protected by the First Amendment.”

But the Fifth Circuit has now interpreted Tornillo much more narrowly than the Eleventh, holding that editorial discretion is not its own right but merely “one relevant consideration when deciding whether a challenged regulation impermissibly compels or restricts protected speech.” On that view, forcing a newspaper or a website to host content is not a First Amendment violation per se; it only rises to the level of a First Amendment violation if such forced hosting compels the host itself to speak (or interferes with its speech).

Under the Fifth Circuit’s approach, whether compelled hosting rises to the level of compelled speech depends on whether the platform at issue implicitly endorses the value of the speech it chooses to host: “[W]hen a newspaper affirmatively chooses to publish something, it says that particular speech—at the very least—should be heard and discussed. So forcing a newspaper to run this or that column is tantamount to forcing the newspaper to speak.”

The Fifth Circuit concluded that social media platforms do not give similar endorsements to the speech on their sites because they “exercise virtually no editorial control or judgment” over the posts that they host. For that reason, the Fifth Circuit held that forcing a social media site to host content does not raise a compelled speech concern.

There are several problems with this reasoning. First, it seems to endorse what then‑D.C. Circuit Judge Brett Kavanaugh once called a “use it or lose it” theory of the First Amendment, namely the theory that platforms which exercise too little editorial discretion can lose their right to do so in the future. As Judge Kavanaugh explained, this theory makes little sense even on its own terms, since a “‘carry all comers’ decision itself is an exercise of editorial discretion.”

In addition, this test will raise difficult and unanswered line-drawing problems with enormous First Amendment implications. How many submissions does a newspaper have to reject to be understood as implicitly endorsing the worthiness of everything it runs (and thus adopting it as its own speech)? At what point might a social media company’s algorithm become aggressive enough in rejecting posts to reach the threshold necessary for compelled hosting to become compelled speech? And most relevant to the social media platforms at issue in this case, can the decision (whether by algorithm or human) to prioritize or deprioritize certain posts on social media ever qualify as protected speech? These questions are left unanswered.

But more fundamentally, the Fifth Circuit’s test is asking the wrong questions because its conclusion rests on two false premises. First, editorial freedom is a distinct form of free expression protected by the First Amendment; it is the right to curate and present a unique mix of speech in a distinct arrangement and order. Social media platforms do engage in such editorial expression, both when they bar some posts as against their rules (even when those rules are lenient) and when they present each user with their own “feed” arranging others’ posts in a particular order. Websites do not need to pass an “appearance of endorsement” test to have this right.

Second, the Fifth Circuit raises too high a bar for finding a compelled speech violation, which also does not require an appearance of endorsement. After all, the Supreme Court held in Wooley v. Maynard (1977) that a state may not force a driver to carry a motto on his license plate, even though no one would assume that the driver endorsed the message. And the Supreme Court has repeatedly held that compelled funding of others’ speech is a form of compelled speech, even though, once again, such funding would not be perceived by others as endorsement.

To be sure, the Supreme Court has not been as clear and consistent on the issue of compelled speech as it could have been. In Pruneyard Shopping Center v. Robins (1980) (a case heavily relied on by the Fifth Circuit), the Supreme Court held that a state may force a private outdoor shopping center to host political pamphleteers. In my view, Pruneyard was wrongly decided and is inconsistent with Maynard and other compelled-speech cases. But even though it remains binding precedent, Pruneyard has never been extended by the Supreme Court to apply to a media company that hosts and presents speech as its very reason for existing, as opposed to a shopping center or other piece of real estate open to the public that primarily exists for some other purpose.

There are many other questionable conclusions and leaps of logic in the Fifth Circuit’s opinion, too many to list here. But given the fundamental incompatibility between the holdings of the Fifth and Eleventh Circuit, one thing seems certain. As Fifth Circuit Judge Leslie Southwick notes near the end of his partial concurrence and partial dissent, “The Supreme Court will, as always, have the final word.”