The Volokh Conspiracy
Mostly law professors | Sometimes contrarian | Often libertarian | Always independent
First, congratulations, everybody on Obergefell v. Hodges! Eight out of nine Justices implicitly or explicitly encourage you to "celebrate today's decision". O.K., come back when you're done, and read this nitpicking of the Court's rationale.
The Court had two rationales here: substantive due process and equal protection.
Let's peek at the second rationale first: equal protection. The nature of an equal protection claim is that people who are similarly situated should be treated the same. For example, consider Brown v. Board of Education: it didn't hold that the government has to run public schools, but since it has public schools, access to them should be non-discriminatory. In general, you can resolve an equal protection problem by leveling everyone up (everyone gets the preferred treatment or benefit) or by leveling everyone down (no one gets it).
O.K., now on to the first rationale. The Court held that "the right to marry is protected by the Constitution" under the Due Process Clause. It gives four reasons for this: (1) marriage is "inherent in personal autonomy", (2) marriage is a "union unlike any other in its importance to" the parties, (3) the right to marry connects to the right to make decisions about family, procreation, and childrearing, and (4) marriage is a "keystone of our social order".
But the problem with this is that when something is a substantive due process right, then it can't be taken away—even if it's taken away from everyone evenhandedly. What if a state decided to abolish the (legal) institution of marriage entirely, i.e., just get out of the marriage business? It doesn't seem to me like that should be unconstitutional—and maybe even the majority here isn't intending to rule it out. But it seems like the plain implication of this section of the opinion is that states must provide marriage to all couples.
In light of this, Part II of Justice Thomas's dissent seems right to me. There's no "liberty" to get married: you can already have intimate relations and live with whoever you want. Whatever legal benefit gay couples lack, you can imagine constitutional holdings mandating those without going the next step and saying there must be a recognition under the name of "marriage". Here's what Thomas writes in Part II: marriage isn't a liberty at all, it's not "locomotion" or "freedom from governmental action more broadly"; the parties here are claiming that the states should "grant them government entitlements"—which the Due Process Clause generally doesn't provide.
Now wait a minute, though, doesn't Loving v. Virginia, the case that struck down state bans on interracial marriage, already hold that marriage is a substantive due process right? Justice Thomas distinguishes Loving (and some similar cases) on the ground that that case didn't just involve a failure to recognize—it involved criminal penalties for interracial couples who married. I would go a bit further than Thomas here and suggest that that part of Loving is incorrect. Here's the complete text of the substantive due process portion of Loving:
These statutes also deprive the Lovings of liberty without due process of law in violation of the Due Process Clause of the Fourteenth Amendment. The freedom to marry has long been recognized as one of the vital personal rights essential to the orderly pursuit of happiness by free men.
Marriage is one of the "basic civil rights of man," fundamental to our very existence and survival. Skinner v. Oklahoma, 316 U.S. 535, 541 (1942). See also Maynard v. Hill, 125 U.S. 190 (1888). To deny this fundamental freedom on so unsupportable a basis as the racial classifications embodied in these statutes, classifications so directly subversive of the principle of equality at the heart of the Fourteenth Amendment, is surely to deprive all the State's citizens of liberty without due process of law. The Fourteenth Amendment requires that the freedom of choice to marry not be restricted by invidious racial discriminations. Under our Constitution, the freedom to marry, or not marry, a person of another race resides with the individual and cannot be infringed by the State.
Loving does clearly say, in the first paragraph quoted above, that there's a substantive due process right to marry. (This is after a whole previous discussion about why the ban on interracial marriage violated equal protection.) But then in the second paragraph, it slips right back into equal protection rationales. Skinner doesn't hold anything about due process at all—it's an equal protection decision based on "fundamental interests". Maynard has a general discussion about marriage but isn't about due process either.
So to the extent Loving holds that bans on interracial marriage violate anything more than equal protection, (1) its reasoning is extremely sloppy and, indeed, almost non-existent, and (2) its conclusion seems incorrect to me.
So if this is a case about access to government entitlements—which I think is right—then it seems to me like only the equal protection portion of the opinion (potentially) makes sense. You can do equal protection in two ways: (1) the classification implicates a fundamental interest, or (2) the classification burdens a class. Within (2), you can win in two basic ways: (2a) the class burdened is a protected class, which gets some heightened degree of scrutiny, or (2b) the class burdened isn't protected, so the law survives if there's a rational basis, but this law is so irrational that it fails rational basis.
Recent gay-rights cases written by Kennedy—Romer v. Evans and United States v. Windsor—have followed method 2b: without making any holding that gays are a protected class, they've ruled out any reasonable justification for the classification, leaving only "animus" or a "bare desire to harm". I find those decisions very unconvincing because I think the majority gives very short shrift to non-hateful justifications for the laws which, even if wrong, are non-hateful. Once it rules out everything except animus, O.K., animus isn't a rational basis, but it shouldn't have rejected some other rationales. The Obergefell opinion flirts with this—with its references to "long history of disapproval", "grave and continuing harm", and "disrespect and subordinate"—but doesn't say it explicitly.
The court has never used method 2a for gays, which would require arguing that gays are a protected class and get heightened scrutiny. I'm partial to the idea of intermediate scrutiny myself, so that everything related to gender, including later cases involving trans issues, can all be considered under the same rubric. It also means that it's not important to decide whether this is discrimination based on sexual orientation or (as Ilya Somin has argued) based on gender. But this approach doesn't have Supreme Court support at this point.
It seems like this opinion is most defensible as an instance of method 1, an equal protection opinion based on fundamental-interests. (Something can be a "fundamental interest" for equal protection purposes without being a "fundamental right" for substantive due process purposes. Aside from marriage, I'd point to the interest in not being sterilized (compare Buck v. Bell with Skinner v. Oklahoma) and the interest in voting for President (compare Article II, § 1, cl. 2 with Bush v. Gore).) The opinion also cites cases like Zablocki v. Redhail and Skinner that use this method.
An equal protection holding based on fundamental interests means that if marriage exists, there can't be unreasonable discrimination, but you could still abolish marriage altogether—which seems correct to me.
It would help, though, if the opinion had been more explicit about this rationale—one might, for instance, expect some discussion of the level of scrutiny here (i.e., how well should the government defend why its discrimination isn't unreasonable?), which you might think would be strict scrutiny. But instead, the "disrespect and subordinate"-type language suggests that something else is being used, maybe the animus idea? Anyway, this part could have been a lot better reasoned.