Experimenting in Same-Sex Marriage and Other Matters

We live in a golden age for empiricism. Pythagoras said the world is made up of numbers and that is the slogan of empiricists. But until this time of big data and exponential computer power, empiricism had an Achilles heel, because it lacked substantial processing capacity. Fortunately, the relentless increase in computational power and the mining of big data has eroded this constraint.

As a result, empiricism is on the rise throughout the social sciences. Because more can be done with data, more sophisticated tools are being invented to help distinguish correlation from causation. Empiricists now form a community that applies quite rigorous standards to assess the validity of their colleagues’ work.

The rise of empiricism has important implications for the shape of our politics. As I argue in Accelerating Democracy, the government should transform itself into a better tool for policy experiments, because more learning is possible. Information markets would also make empiricism more valuable because they make it easier to use past analysis to predict the future. Life is understood backwards but lived forwards, as Kierkegaard famously said.

The promise of better empiricism provides another reason for government decentralization, because sophisticated empiricism allows us to compare the results of different policies even in different jurisdictions. This renewed promise provides yet another reason for the Supreme Court, other things being equal, to enforce the enumerated powers against the federal government, permitting space for state experimentation.

Cutting back on substantive due process and other mechanisms for loose interpretation of the Fourteenth Amendment would also permit states to experiment with different bundles of rights. In an age where we can better gauge the effects of newly claimed rights, it becomes less justifiable for the Court to rely on its own intuitions to establish the content of rights outside those specified with a measure of clarity in the Constitution.

This general observation has relevance to Hollingsworth v. Perry. If the Court eschews a federal right for same-sex marriage, we can then measure the effects of the different marriage regimes in different states, such as effects, if any, on divorce rates of both same-sex and opposite sex couples. As Justice Kennedy said, we currently have only five years of information on the consequences of society of same-sex marriage. While my inclination is to favor legalizing same-sex marriage at the state level, I also believe more information would be valuable, particularly because new data can be more systematically evaluated than ever before.

Randomization of policy can at times be an even more effective way to determine the effects of different policies, because it avoids confounding factors such as demography that distort jurisdictional comparisons. In some circumstances, it may be appropriate for agencies or Congress to impose at random different regulations on different corporations or people to better assess results.

So long as government has power to make the regulations at issue, it can as a general matter constitutionally engage in such trials for the time they are needed to generate data. Indeed, even without randomization, politicians are already making citizens experimental subjects, because they are effectively trying out policies on citizens today in the hope of better social results.

Rules and institutions that test policies change our political culture. They create a more humble and less divisive politics, helping us concentrate on what we have in common—the facts of the world—rather than the often unsupported intuitions that divide us. Openness to new evidence is an important part of what Judge Hand called the “spirit of liberty that is not too sure it is right.”