On Dec. 20, Vermont's supreme court held that the state is required by its constitution (not the federal one) to provide homosexual couples with the same benefits that are available to married heterosexual couples, either through actual marriage or through some sort of partnership program (the legislature can choose). "Here we go again," sighed many conservatives. "Another social revolution imposed by activist judicial fiat."

In 1973 the U.S. Supreme Court ruled that the government may not restrict a woman's right to an abortion during the first trimester, and may restrict abortions only within certain limits after that. When Roe v. Wade was handed down, states were gradually moving toward legalization of abortion. The Roe decision smashed that slow but discernible progress toward consensus by leaving no place in the sun for those who dissented. Out of Roe sprang a deeply aggrieved pro-life movement that challenged not just abortion but the very legitimacy of the courts. The cost to the country's social fabric has been immense.

Is Baker v. State of Vermont simply Roe all over again? Not really. This time we are getting it right.

I should say that I firmly favor gay marriage, both on humanitarian grounds and because I think it is good social policy. If gay people exist--that is, if we are not just neurotic heterosexuals who need to get our act together--then surely we ought to be encouraged to marry and settle down. It has never been clear to me why discouraging stable gay relationships in favor of sex in parks and porn shops is good for the American family, or anyone else.

Nonetheless, gay marriage is a deeply polarizing issue, to put the case mildly. To impose it judicially on a predominantly hostile country would beg for a backlash--against gays, against the courts, against government broadly. Not long ago, it seemed likely that Hawaii's courts might rule in favor of gay marriage there, and that other states would be required to recognize Hawaiian marriages. That would indeed have been Roe redux.

The grenade, however, has been disarmed. In the 1996 Defense of Marriage Act, Congress decreed that no state need recognize another state's homosexual marriages. A majority of states have subsequently passed similar laws. The result is that, for the time being, the issue of gay marriage has been de-nationalized. In California, voters will decide an anti-gay-marriage initiative in March, and they may also face a pro-gay-marriage initiative in November. I expect to live long enough to see at least one state legalize gay marriage through the conventional legislative process.

Vermont, of course, has not used the conventional legislative process. It has used judicial fiat. Doesn't the court's decree undercut the democratic process?

Yes: of course it does. In my own ideal world, the Vermont legislature would approve gay marriage and that would be that. In many traditionalists' ideal world, the legislature would disapprove gay marriage and that would be that. Neither side, however, lives in an ideal world. Both must settle for second best. In the case of gay marriage, as in the case of abortion, real-world second-best means localism.

Judicial minimalists argue that courts--federal or state--should simply absent themselves from policy disputes when the law does not give them clear and explicit grounds to intervene. Vermont's constitution includes a so-called "common benefits" clause, which says that the government is "instituted for the common benefit, protection and security of the people, nation or community, and not for the particular emolument or advantage of any single person, family or set of persons who are a part only of that community." To read this clause as offering clear or explicit grounds to mandate marriage or domestic-partner benefits for homosexuals is a stretch, to say the least.

On the other hand, gay marriage, of its nature, implicates a core constitutional issue, namely equal protection of the law. Vermont's court held that the exclusion of homosexuals from marriage "treats persons who are similarly situated for purposes of the law, differently." Legislatures cannot withhold the right to drive or vote, say, from blacks, at least not without offering the courts a constitutionally compelling rationale. Can they, then, withhold marriage from homosexuals, or withhold the state benefits that flow from marriage, without facing judicial scrutiny?

Here the comparison with Roe is instructive. If abortion is indeed the killing of babies, pure and simple, then the courts can no more bow before legislatures' desire to legalize it than they can agree to let a legislature legalize any other form of murder. The great mistake in Roe was not that the courts became involved; judicial scrutiny was inevitable. The mistake, rather, was that the federal courts became involved, long before there was anything like a national consensus.

On gay marriage, most state supreme courts probably would not have ruled as Vermont's did. But then, most states are not Vermont. It is entirely conceivable that some other state eventually will approve gay marriage or partnership benefits legislatively, only to see the arrangement knocked down by an activist conservative court. C'est la vie. The beauty of a federalist arrangement is that it does not require that the states all get either the process or the answer right; it requires merely that they all get it different.

Gay marriage's "Roe" moment will arrive if the U.S. Supreme Court decides that states cannot treat marriage differently. Both opponents and friends of homosexual marriage ought to hope that that day is long in coming. Whatever you think of the outcome in Vermont--or in Hawaii, whose court this month finally threw out the gay-marriage claim--federalism is working.

Jonathan Rauch is a senior writer at National Journal.