Ultimately, I think Carpenter is right: One could easily craft an anti-SSM decision that would pass the rational basis test. New York just didn't do it.

Notably, they never address the question of legitimacy of aim -- confining themselves only to the aim's rationality, which deviates from the test itself. I still fail to see why no one finds this important: While the rational basis test may be utterly lenient, surely they could at least employ the proper wording, right?

But in any case, I'd justify hetero-only marriage as follows: Marriage benefits the children of two heterosexual parents. To benefit the children of two heterosexual parents is a legitimate state interest, and extending marriage to these children's parents (or their potential parents) is rational. Nothing about the extension of marriage to heterosexual couples (the legitimate aim) requires the extension of marriage to homosexual couples. End of decision.

Nothing about the rational basis test asks us to consider questions of inclusion or exclusion. Under rational basis review, these may be as invidious as one likes.

Marriage benefits the children of two heterosexual parents. To benefit the children of two heterosexual parents is a legitimate state interest, and extending marriage to these children's parents (or their potential parents) is rational. Nothing about the extension of marriage to heterosexual couples (the legitimate aim) requires the extension of marriage to homosexual couples. End of decision.

However, nothing about extending the benefits of marriage to the children of heterosexual couples requires the extension of marriage to childless couples, couples with one or more sterilized partner, post-menopausal women, otherwise infertile couples . . . . Your test makes it possible to discriminate between fertile heterosexual couples and all other couples, but not between heterosexual couples tout court and homosexual couples.

Also, nothing about "benefit[ting] the children of two heterosexual parents" - as distinct from any other type of parents - is a rational state end. There is no rational, non-discriminatory reason the state should benefit children with a certain type of parents and not otherwise identical children whose parents happen to be different. Only irrational animus explains that position, which is almost the only grounds that will not pass the "rational interest" test. (The state's interest cannot be justified by the claim that heterosexual marriages are more beneficial than homosexual ones for their children. Not only is that claim proven false, but it is the result of the policy you propose, whereby the state explicitly aims to provide more benefits to such children. The state cannot cite its own policy as justification for its policy.)

So, your basic "state interest" is indefensible, and, more importantly, your defense of that interest cannot make the distinctions you attempt to derive from it - in fact, it appears to invalidate large numbers of currently-legal heterosexual marriages.

Your test makes it possible to discriminate between fertile heterosexual couples and all other couples, but not between heterosexual couples tout court and homosexual couples.

This isn't correct. The rational basis test never requires that there be a perfect fit between the ends and the means. It requires only that there be some fit, almost regardless of how imperfect it might be. Errors of inclusion and exclusion are seldom a problem.

Whether one agrees with this is another question, but such is the standard of review.

Also, Kevin, nothing about the rational basis test demands that the state also pursue some other rational end. We might wish it to do this, and it might be nice if it did, but the test doesn't ask that. Rational basis review asks only whether the chosen aim of the legislature is legitimate, and whether a rational person might think that the means will further the end.

Thus, the aim of supporting heterosexual families is rational (I support them too), and the means -- offering marriage -- is rational. End of story.

(And yes, I agree that rational basis jurisprudence is deeply flawed... But let's not kid ourselves about the way ti works.)

It's beyond dispute that the end must be legitimate and not just rational. Insofar as the end offends the Constitution it will be illegitimate and the law stricken. Likewise, as in Romer, if a court finds that the end is pretext for some animus (or illegitimate end) it will also be stricken - regardless of the rationality of the relationship between the end and the law.

Here's a hypothetical along similar lines.

Assuming that strict and intermediate scrutiny did not exist, could we argue then that

1. It is legitimate to want Christians to form stable families,

and

2. It is rational to extend marriage to Christians as a means of attaining this end?

Jews, Muslims, and others need not apply.

If I understand it correctly, rational basis review does not inquire into the completeness of the legislature's job.

(Keep in mind, strict scrutiny doesn't exist in this hypothetical, so we can't say "Aha! Discrimination on the basis of religious affiliation!" and invoke that doctrine.)

I submit that both 1) and 2) would satisfy rational basis review, which never asks about the distinctions being drawn, or about what groups might be left out of the process. It asks only if the goal at hand -- chosen by the legislature as narrowly as it likes -- is, in near isolation from the rest of the world, a legitimate goal. It then asks whether a rational person might find a relationship between the means and the end.

So, legal experts, what's the decision?