[UPDATE: For more on the Oregon order itself, see this post (about the liability for the bakers’ past statements) and this post (about the lack of liability for public reaction to the bakers’ publicizing the complaint).]
The Oregon baker decision — which required a bakery to pay $135,000 to a same-sex couple for refusing to prepare a cake for their same-sex commitment ceremony — also ordered the bakers to:
“…cease and desist from publishing, circulating, issuing or displaying, or causing to be published, circulated, issued or displayed, any communication, notice, advertisement or sign of any kind to the effect that any of the accommodations, advantages, facilities, services or privileges of a place of public accommodation will be refused, withheld from or denied to, or that any discrimination will be made against, any person on account of sexual orientation.”
Statutes prohibiting similar communications (including as to race, religion, and sex, and as to employment and housing as well as public accommodations) are common, and generally thought to be constitutional. But why? Here’s what I think is the right answer, though I agree that courts haven’t been clear on it.
Assume that it is indeed against the law to refuse to serve someone based on race, religion, sexual orientation, and so on — and, in particular, to refuse to provide a cake for a same-sex commitment ceremony. Then, saying we “will … refuse” to provide a cake is essentially a true threat of illegal conduct.
To be sure, it is not a threat of violence, or even a threat to commit a crime, but it is a threat to act illegally (by violating the anti-discrimination statute). And it is a threat that would have much the same effect as an outright refusal to provide a cake to someone who shows up and asks for it, because it tells people that it’s futile to even ask.
Indeed, I think we’d see the same in lots of other situations where speech is properly treated as civilly actionable. If a landlord (or anyone else) locks a tenant out of the property that the tenant has leased, he is committing a tort.
If the person instead tells the tenant, “If you come onto the property, you will find that you are locked out,” he is committing precisely the same tort, even if the tenant doesn’t bother coming to the property and checking whether his key still works. (Say this is a vacation home and the tenant is out of town, so trying to go on the property and finding himself locked out will be expensive.) Threatening the tenant with unlawful exclusion from the property is unprotected speech, because it is a threat of tortious conduct.
Note, by the way, that simply saying “we disapprove of the Oregon decision,” or “we disapprove of same-sex marriages,” is not a constitutionally unprotected threat of denial of service — which is why the text of the order, properly read, does not prohibit such statements. But saying “we won’t do same-sex marriage, same-sex wedding cakes” (or, in the typical context, “we don’t do same-sex wedding cakes”) is such a threat.
Now some court decisions — most prominently, the Supreme Court’s decision in Rumsfeld v. FAIR (2006) — articulate the prohibition on statements that service will be refused by saying that the statement is “conduct” rather than “speech”:
Congress, for example, can prohibit employers from discriminating in hiring on the basis of race. The fact that this will require an employer to take down a sign reading “White Applicants Only” hardly means that the law should be analyzed as one regulating the employer’s speech rather than conduct.
But why is posting a sign, a quintessential means of speaking, “conduct” rather than “speech”? It’s not because the law that bans the posting of the sign also bans conduct, or even (just) because the posting of the sign has the same effect as conduct. Indeed, in Holder v. Humanitarian Law Project (2010), the Court warned of the limits of such “conduct not speech” arguments:
[T]he Government … claim[s] that the only thing truly at issue in this litigation is conduct, not speech. Section 2339B [which bars the provision of assistance to designated Foreign Terrorist Organizations, including the provision of “training" and “expert advice"] is directed at the fact of plaintiffs’ interaction with the PKK and LTTE [two terrorist groups], the Government contends, and only incidentally burdens their expression….
[But t]he Government is wrong that the only thing actually at issue in this litigation is conduct …. [Section] 2339B regulates speech on the basis of its content. Plaintiffs want to speak to the PKK and the LTTE, and whether they may do so under § 2339B depends on what they say. If plaintiffs’ speech to those groups imparts a “specific skill” or communicates advice derived from “specialized knowledge” — for example, training on the use of international law or advice on petitioning the United Nations — then it is barred. On the other hand, plaintiffs’ speech is not barred if it imparts only general or unspecialized knowledge.
The Government argues that § 2339B should nonetheless receive intermediate scrutiny because it generally functions as a regulation of conduct. That argument runs headlong into a number of our precedents, most prominently Cohen v. California (1971). Cohen also involved a generally applicable regulation of conduct, barring breaches of the peace. But when Cohen was convicted for wearing a jacket bearing an epithet, we did not apply [the standard applicable to content-neutral regulations of conduct]. Instead, we recognized that the generally applicable law was directed at Cohen because of what his speech communicated — he violated the breach of the peace statute because of the offensive content of his particular message. We accordingly applied more rigorous scrutiny and reversed his conviction.
This suit falls into the same category. The law here may be described as directed at conduct, as the law in Cohen was directed at breaches of the peace, but as applied to plaintiffs the conduct triggering coverage under the statute consists of communicating a message.
(The Court ultimately upheld the restriction on the speech in Humanitarian Law Project, but only after viewing it as a speech restriction, and evaluating it under strict scrutiny.)
The “regulates conduct rather than speech” argument from Rumsfeld v. FAIR thus has to rest on something special about the relationship of “White Applicants Only” to anti-discrimination law — something absent in the relationship of training and expert advice to the assistance-to-terrorist-groups ban. That something, I think, is that “White Applicants Only” falls within an unprotected category of speech that is closely related to punishable conduct: a true threat of illegal conduct.
The speech prohibited in the cake case, like the “White Applicants Only” sign, is thus quite similar to solicitation of illegal conduct, or offers of illegal conduct, which the Court has held are constitutionally unprotected (see United States v. Williams (2008)), and that are often themselves labeled a form of “conduct” (again, here shorthand for unprotected speech closely related to particular forbidden conduct). Threats of illegal conduct are no more protected than such solicitation or offers.
Now this doesn’t resolve the question whether the cake case properly interpreted Oregon public accommodation law; whether a $135,000 damages award is sound; whether all the reasoning behind the Oregon administrative agency’s decision that the bakers’ public statements indeed violated Oregon law is correct; or whether there should be such laws, or religious exemptions to such laws.
But if refusing to sell cakes for same-sex commitment ceremonies (which today would be same-sex legal marriage ceremonies, since same-sex marriage is now legally recognized) indeed violates Oregon law, an order prohibiting such refusals is constitutionally valid.