That’s right, I said three conservatives dissented, not four. Alito, Thomas, and Gorsuch. The fourth vote in favor of letting the law take effect was … Elena Kagan.

Don’t ask me how we ended up with that configuration, as Kagan didn’t write an opinion explaining her reasoning and didn’t join the three conservatives in theirs. But that leaves us with an unorthodox majority of Roberts, Breyer, Sotomayor, Kavanaugh, and Barrett.

Which is reassuring for those of us who favor the traditional understanding of the First Amendment, that private actors like social media companies get to regulate their platforms without government interference. A 5-4 majority on this question *probably* means there are at least five votes in favor of striking down Texas’s law when that matter comes squarely before the Court. But the fact that three of the six Republican appointees are evidently prepared to have a rethink on the ol’ 1A in the age of Facebook means the reassurance is no more than mild.

“[S]o the answer to the question of how close conservatives are to completely inverting the meaning of the First Amendment and using it to bludgeon their political enemies is ‘pretty close,'” said one Twitter pal about today’s ruling. I don’t know about “pretty close,” but too close.

Here’s the decision, which consists almost entirely of Alito’s scant five-page dissent. Texas’s law, H.B. 20, allows a social media user to sue the largest platforms if he believes they discriminated against him based on viewpoint in moderating his content. A federal district court issued a preliminary injunction blocking the law from taking effect on grounds that it will likely be held to violate the First Amendment. The Fifth Circuit issued a stay of that injunction. SCOTUS has now reversed that decision, restoring the injunction — a hint that five justices agree that there’s a First Amendment problem here, although not definitive confirmation. The case will go back to the Fifth Circuit now for a more considered judgment about whether the law should be enjoined while the constitutional litigation plays out.

Kagan disagreed with the majority although the fact that she declined to specify why means she may have an issue with the procedural posture, not the underlying First Amendment question. If and when the law comes back before the Court, it could be 6-3 to strike it down. Or it could be 9-0, as Alito made a point in his dissent of saying that he hasn’t fully formulated a view of how amenable social media companies should be constitutionally to regulation. But he’s … interested in the idea, as are Thomas and Gorsuch.

The crux of their disagreement with the majority appears to be whether, under existing precedent, social media companies are likely to win on the underlying First Amendment question. The three conservative dissenters aren’t sure, viewing this area of law as unsettled. Should an entity like Facebook be treated like a traditional print publication, which gets to set its own rules? Or should it be treated as more like a shopping mall, which is technically private property but which the Court has said is close enough to a public space in its nature that people should enjoys some free-speech rights there?

The five members of the majority evidently feel more sure that there’s a First Amendment violation, presumably because they think the print publication analogy is more apt. Like I say: Reassuring, although five votes is the least reassurance one could hope for.

The most noteworthy part of his opinion comes in a footnote, which touches on a point that will be familiar to readers. Section 230 of the Communications Decency Act immunizes social media companies from liability for comments posted by users; it’s not at issue in this case but it’s a hobbyhorse for some righties who insist that a company like Facebook is either a “platform” that simply neutrally hosts content or a “publisher” that edits it by moderating what its users post. There’s no such distinction under the law but Alito took care to flag the “publisher” issue in his footnote, again signaling that he might be willing to get creative in deciding what social media companies can and can’t properly do:

In other words, according to Texas, Facebook shouldn’t be allowed to claim that it’s a publisher (i.e. print publication) for First Amendment purposes yet merely a passive platform (i.e. a shopping mall) for purposes of Section 230. But Section 230 doesn’t require social media companies to be passive fora for whatever their users want to post, with no right to moderate their content. Just the opposite:

No provider or user of an interactive computer service shall be held liable on account of—

(A) any action voluntarily taken in good faith to restrict access to or availability of material that the provider or user considers to be obscene, lewd, lascivious, filthy, excessively violent, harassing, or otherwise objectionable, whether or not such material is constitutionally protected

The CDA encourages them to moderate “otherwise objectionable” speech even if it’s constitutionally protected, a broad catch-all. There’s no conflict, in other words, between how social media companies view themselves under the First Amendment and under Section 230: In both cases they’re private actors who are entitled to moderate what their users post. They’re “neutral” inasmuch as no one is forbidden from using their services for belonging to a particular political movement or faith. (For instance, 16 months after his father’s ban, Donald Trump Jr continues to sh*tpost with abandon on Twitter to an audience of millions.) Typically you have to earn your ban, the way someone handing out leaflets in a shopping mall might end up being escorted out if he started shouting obscenities and sieg-heil-ing. The real question here is, or should be, whether there might properly be some legal recourse against companies for applying their own rules *capriciously,* which they often do. And not always with “objectionable” speech.

Alito, Thomas, and Gorsuch are willing to entertain the possibility. Five justices, and maybe six, are less willing.