The Dormant Commerce Clause, Social Media Platforms, and Restrictions on Political Discrimination

[Jack Goldsmith and I will have an article out about the Dormant Commerce Clause, geolocation, and state regulations of Internet transactions in the Texas Law Review early next year, and I'm serializing it here. There is still plenty of time for editing, so we'd love to hear any recommendations you folks might have; in the meantime, you can read the entire PDF of the latest draft (though with some formatting glitches stemming from the editing process) here.]

The same basic approach we laid out in the last several days, we think, should apply to bans on political discrimination. (Some California cases suggest that California law already bans such discrimination, and various cities, counties, and territories have explicit rules along those lines.[1]) If, say, Iowa law bans social media platforms from blocking Iowan-to-Iowan speech based on its viewpoint, whether religious, moral, or political, Iowa courts could, consistent with the Dormant Commerce Clause, apply that law to HitchedIn and, for that matter, to Facebook and Twitter—again, so long as those platforms could geolocate the communicating parties as being in Iowa. Likewise for common carrier statutes, which would ban social media platforms from blocking such communication more generally (perhaps with a few viewpoint-neutral exceptions, such as for spam or sexually themed material). This question was raised in the challenges to the Florida and Texas laws that banned social media platforms from discriminating based on political viewpoint, but those courts didn't reach it, because they struck those laws down on other grounds.[2]

To be sure, one element of the Pike balancing test is to ask whether "the burden imposed on such commerce is clearly excessive in relation to the putative local benefits,"[3] so courts will have to consider the local benefits. But those benefits are quite important.

In our HitchedIn and SafeBook hypos, the local benefits were the benefits of protecting local residents from discrimination based on sexual orientation, religion, and criminal history. And in the political discrimination ban, they are the benefits that the Court viewed as important in Turner, albeit as to the First Amendment—"assuring that the public has access to a multiplicity of information sources," "a governmental purpose of the highest order, for it promotes values central to the First Amendment."[4] "It has long been a basic tenet of national communications policy that the widest possible dissemination of information from diverse and antagonistic sources is essential to the welfare of the public."[5]

None of this tells us, of course, that applying such public accommodation laws or common carrier laws to social media platforms is a good idea. The only point is that the Dormant Commerce Clause doesn't categorically preclude these sorts of experiments.

Because this topic is so much in the news—with Florida and Texas enacting such statutes, and other states considering them—let's lay this out in some more detail, and in particular cover four possible categories of hypothetical Iowa statutes, and not just Options 1 and 2.

1. Forbidding viewpoint discrimination by platforms when Iowans read material posted by Iowans. This is the analog of Option 1 for the other statutes discussed in previous sections. It has the narrowest extraterritorial effect, because it doesn't materially affect the service the platform offers outside the state.

Indeed, such a nondiscrimination law would be similar to a normal public accommodation law that bans brick-and-mortar public accommodations—such as bars or stadiums—in a particular state or city from excluding people based on their "political ideology," including political speech.[6] Such a law may require multistate chains to develop different rules for different states in which they operate. It may lead to some interstate travelers being upset, for instance if they are used to the chain's restaurants forbidding (say) patrons from wearing Confederate-flag garb but have to endure seeing it in a jurisdiction that bans ideological discrimination. And it may have various other extraterritorial effects.

Still, it's clear that a state can indeed impose such rules on businesses within it. Likewise, a state can impose similar rules with regard to communications that are sent and received from that state.

2. Forbidding viewpoint discrimination by platforms when anyone reads material posted by Iowans. This is what we've also labeled Option 2 in the examples above, and it's similar to the coverage of Florida's social media law,[7] though that law focuses—improperly, we think—on "reside[nce]" or "domicile," legal questions that platforms might not be able to easily answer, rather than on place of posting (or place from which the user created the account), which is a geographical question that platforms can answer more reliably.[8]

As we have noted above, this is a harder case to resolve with certainty, because any such law would require the platform to provide out-of-state users with a different experience than they would otherwise provide. Here, though, is a potentially helpful analogy: Imagine a multistate chain of stores that take wedding invitations supplied by the couples who are marrying, and—in a display of conspicuous consumption—hand-deliver them to recipients throughout the country. And imagine the company declines to do this for invitations to same-sex weddings. Iowa antidiscrimination law may well forbid such discrimination by Iowa branches of the chain that are serving Iowans, even as to invitations that are to be delivered to other states.[9] (As usual, let's set aside any First Amendment objections to the law, and focus solely on the Dormant Commerce Clause.[10])

Iowa should be entitled to protect Iowans against discrimination based on sexual orientation,[11] even as to material that's shipped from Iowa to other states. The same logic argues for the constitutionality, under the Dormant Commerce Clause, of the Iowa law that forbids a platform from discriminating based on viewpoint when anyone, including an out-of-stater, reads materials posted by Iowans.

3. Forbidding viewpoint discrimination by platforms when Iowans read material posted by anyone. This is close to the California closed-captioning requirement upheld by the Ninth Circuit in the CNN case. California can require that CNN contents transmitted into California include closed captioning (even if CNN would otherwise prefer not to include it, and doesn't include it for viewers in other states). Iowa can likewise require that social media contents transmitted into Iowa include material that the platform would have preferred to delete.[12] A state generally has the power to require that products made available in that state have certain features, even if that covers businesses that would create those features outside the state.

This would mean that the platform has to keep posts, regardless of viewpoint, on its computers. But it seems likely that the platforms could conceal those posts from everyone except Iowans (and people in states with similar laws), so the Iowa law wouldn't affect what will be visible to people in other states. And while this would involve some extra coding and work for the platform, the same was true of CNN's obligation to provide closed captioning.

If these extra costs proved to be exorbitant, the outcome of a Pike balancing analysis might become difficult to predict. But as we have noted several times, platforms pervasively use geographical identification and filtering technology to serve their business ends, and so we suspect that it can done at a reasonable cost here as well.

4. Forbidding viewpoint discrimination by platforms when Iowans read material posted by anyone and when anyone reads material posted by Iowans. This appears to be similar to the approach of Texas's social media law,[13] and of proposals in other states, such as Georgia and Michigan. The same set of considerations that would govern approaches 2 and 3 above would apply here. These type of laws are difficult to generalize about and their constitutionality will likely turn on a fine-grained Pike analysis.

[1]. See Eugene Volokh, Bans on Political Discrimination in Places of Public Accommodation, 15 NYU J. L. & Liberty 709 (2021). Some of the laws ban only discrimination based on party affiliation, but others ban discrimination based on broader political beliefs as well.

[2]. See NetChoice, LLC v. Att'y Gen., __ F.4th __ (11th Cir. 2022); NetChoice, LLC v. Paxton, No. 1:21-CV-840-RP, 2021 WL 5755120, *6 n.1 (W.D. Tex. Dec. 1, 2021), appeal pending.

[3]. 397 U.‌S. 137, 142 (1970); see Greater L.‌A. Agency on Deafness, 742 F.‌3d at 433 (applying the Pike test); Goldsmith & Sykes, supra note 11, at 806 (discussing a possible Pike-based analysis in extraterritoriality cases).‌

[4]. Turner, 512 U.‌S. at 663.‌

[5]. Id. (cleaned up).‌

[6]. See, e.g., Seattle (Wash.) Mun. Code §§ 14.06.020–.030, .08.020–.030 (banning discrimination by public accommodations based on "any idea or belief, or coordinated body of ideas or beliefs, relating to the purpose, conduct, organization, function or basis of government and related institutions and activities," "includ[ing] conduct, reasonably related to political ideology, which does not cause substantial and material disruption of the property rights of the provider of a place of public accommodation"); Ann Arbor (Mich.) Code of Ordinances §§ 9:.151-.153 (banning discrimination by public accommodations based on "opinion, whether or not manifested in speech or association, concerning the social, economic, and governmental structure of society and its institutions").

[7]. Fla. Stat. § 501.2041(1)(h). Florida's law is substantively narrower than the one we hypothesize, because its scope is narrower than a total ban on viewpoint discrimination; but the Dormant Commerce Clause analysis shouldn't be affected by that.

[8]. People often reside in one place even when they're spending weeks or months accessing the internet from another place. Even if a platform asks for information about where users live when they first sign up (and many platforms won't), users often change their residence. And domicile of course turns on questions such as whether the users have "a certain state of mind concerning [their] intent to remain" in the place where they are physically present, Miss. Band of Choctaw Indians v. Holyfield, 490 U.S. 30, 48 (1989), something that the platforms have no way of knowing.

[9]. Iowa is indeed one of the many states that ban sexual orientation discrimination

[10]. Brush & Nib Studio, LC v. City of Phoenix, 448 P.3d 890, 908 (Ariz. 2019), concluded that a calligrapher has a First Amendment right to refuse to design "custom wedding invitations" that "contain[] their hand-drawn words, images, and calligraphy, as well as their hand-painted images and original artwork"; a similar question is now before the Court in 303 Creative, Inc. v. Elenis, But even Brush & Nib suggested that the result could be different if the store offered the same services to all couples, without such personalization; imagine, then, that our invitation delivery company doesn't hand-paint or hand-draw anything, but simply physically delivers it.

[11]. That's true even if the discrimination wouldn't be the customer's own sexual orientation, but rather the sexual orientation of the parties to the wedding—perhaps the customer's child and the child's prospective spouse.

[12]. To be sure, the social media company may want to delete the material for its own ideological reasons, while CNN's not putting up closed captioning likely stemmed mostly from a concern about cost and risk of error. But while that might conceivably make a difference in the First Amendment analysis of the two laws, it shouldn't affect the Dormant Commerce Clause analysis.

[13]. Tex. Stat. § 143A.002(a) bans platforms from censoring "a user's expression, or a user's ability to receive the expression of another person based on" the user's or another person's viewpoint. Tex. Stat. § 143A.004(a) provides that this applies "to a user who," among other things, "shares or receives expression in this state." and Tex. Stat. § 143A.001(6) defines "User" to include "a person who posts, uploads, transmits, shares, or otherwise publishes or receives expression, through a social media platform."

The Texas law also covers any user who "resides" in Texas or "does business" there, but that might be too hard for social media platforms to determine, for reasons given supra in note 165.