Since the death of John Rawls last week at the age of 82, many eulogistic pieces have appeared in the press hailing him as the greatest political philosopher of his time. Virtually all of them cited Rawls’ two criteria for a just society, as derived in his 1971 book A Theory of Justice: 1) Each person should have the most extensive array of liberties compatible with a like liberty for all; and 2) economic inequalities are justified only if they benefit the worst-off members of society.
The first criterion, the one about basic liberties, is pretty uncontroversial. However, the second one, which Rawls called the “difference principle,” has raised hackles, and not just among right-wingers. It holds that differences in wealth, status, etc., can be defended only if they create a system of market forces and capital accumulation whose productivity makes the lowliest members of society better off than they would be under a more egalitarian system.
In effect, the difference principle says that if you want to assess how economically just a society is, you don’t need to consider its overall level of wealth, or the details of how that wealth is distributed among the various classes. All you have to do is look at how its worst-off members are doing. This is not a very flattering standard for the United States. True, we are on average the wealthiest nation in the world. But we also have the most inequality: Our Gini coefficient, which measures how skewed the distribution of wealth is, overtook Britain’s sometime during the Reagan years and has never looked back. And apparently this inequality exceeds what is needed to lift the well-being of the most unfortunate. America’s poor children happen to be poorer than those of any other industrialized nation, save Ireland and Israel. By the lights of the difference principle, we have a less just society than the Scandinavian countries, where generous welfare states have all but abolished poverty. Is the problem with Rawls’ theory or with us?
Personally, I have always found the difference principle congenial, a nice compromise between rigid socialist egalitarianism and devil-take-the-hindmost capitalism. But when you go back and look at how it was deduced, it looks a little like a put-up job. What Rawls set out to do in A Theory of Justice was derive the nature of a just society from two basic notions: fairness and rationality. He proceeded by way of a thought experiment. Imagine that we are in an “original position” trying to agree on a scheme for a just society. To make our deliberations fair and equal, Rawls stipulated that they take place behind a “veil of ignorance”: None of us is to know what position we will occupy in the society we choose—what natural talents, wealth, or conception of the good life we will end up with. Under these conditions, he argued, a rational person would choose the society that would make his worst-possible life—that is, the one he would have on the most pessimistic assumptions about what his natural advantages/disadvantages would turn out to be once the veil of ignorance is lifted—as good as possible. In other words, he would choose to maximize his minimum well-being—by opting for a society that embodied the difference principle.
Is that really so? The “maximin” strategy may be a good one to use in a zero-sum game, where you are facing an opponent who is trying to maximize his gain and therefore your loss. But it seems unduly conservative for an agent in Rawls’ original position—unless that agent assumes that nature is somehow his adversary, intent on fitting him out with the worst possible endowments in whatever society he plumps for. Normally, when making decisions under uncertainty, a rational person will use a very different strategy. He will seek to maximize his expected utility—gauging the probabilities of the various outcomes and using those probabilities to weight the corresponding utilities. Suppose you are in the original position, contemplating a choice between a Scandinavian-style society and an American one, without knowing what economic stratum you would occupy in either. Instead of just looking at the lowest stratum in each society, as Rawls would have you do, you would consider the proportion of the population in each stratum, take that as the probability that you will end up there, and multiply it by level of well-being at that stratum. You might then find that, by the expected-utility criterion, the American-style society trumps the Scandinavian-style alternative.
This anti-Rawlsian view of the original position was championed by the late John Harsanyi, a game theorist who shared the 1994 Nobel Prize in economics with John (Beautiful Mind) Nash and Reinhard Selten. To sharpen the relevant moral intuitions about the difference principle, Harsanyi offered the following example:
Consider a society consisting of two individuals. Both of them have their material needs properly taken care of, but society still has a surplus of resources left over. This surplus can be used to provide education in higher mathematics for individual A, who has a truly exceptional mathematical ability, and has an all-consuming interest in receiving instruction in higher mathematics. Or, it could be used to provide remedial training for individual B, who is a severely retarded person. Such training could achieve only trivial improvements in B’s condition (e.g., he could perhaps learn to tie his shoe laces), but presumably it would give him some minor satisfaction. Finally, suppose it is not possible to divide up the surplus resources between the two individuals.
I find Harsanyi’s example vexing. I also find Rawls’ reasons for rejecting the expected-utility principle in favor of his own more egalitarian one unpersuasive. (Rawls himself might have felt the same way—he certainly spent little time defending them in recent years.) He seemed to rig his thought experiment so that it could only yield the difference principle. The agents in his original position are denied any knowledge of probabilities—which makes it impossible for them to figure expected utilities. And they are utterly averse to risk, unwilling to take the slightest gamble on their life prospects. Are they more reasonable than their diametrical opposite: agents who would choose a society that maximizes the well-being of the best-off class? (That, by the way, seems to be the sort of society we are drifting toward.) Perhaps the expected-utility strategy strikes the rational balance between the risk-nothing maximin and the risk-all maximax.
Rawls’ great thought experiment may be logically flawed, but it at least points up the conceptual link between our moral attitude toward inequality and our rational attitude toward risk. Of course, if your attitude toward inequality is formed after the veil of ignorance is lifted, it doesn’t really count as moral, does it?