Monday, July 5, 2010

Rawls' A THEORY OF JUSTICE First Installment

Part IV Applications -- Rawls

    We come now to what might plausibly be considered the real payoff for all the technical thrashing about we have been engaged in: an extended analysis of the core argument in John Rawls' famous book, A Theory of Justice. Rawls' hauptwerk is widely considered the most important contribution to English language political theory of the past century, and is arguably the most influential work of philosophy written in the English language during that time. It is worth our while, therefore, to take the time to look at his central argument carefully and in detail. Thirty-three years ago I wrote a book-length examination of A Theory of Justice, called Understanding Rawls, published by Princeton University Press. Much of what I say here overlaps with what I said in that book, but my focus here is more narrowly on Rawls' attempt to apply Bargaining Theory to his subject. Those interested in a somewhat broader discussion are invited to hunt up my book and take a look. I am going to assume that everyone reading these words has some familiarity with Rawls' theory.

    The core of Rawls' work is a simple and rather lovely idea. In the middle of the twentieth century, Anglo-American ethical theory was stuck in what Kant, two centuries earlier, had called an antinomy. Utilitarians and intuitionists were locked in a death struggle, with each side more than capable of exposing the weaknesses of the other, but each unable to defend itself against the other's crushing arguments. Rawls had the idea that the conflict might be resolved by combining an old tradition of political philosophy -- social contract theory -- with a brand new field of mathematics and economics, Game Theory. Early in the development of the theory that eventually found its full-scale exposition in A Theory of Justice, Rawls claimed that he could prove a theorem in Bargaining Theory, and that the proof of that theorem would constitute a justification for the pair of principles which, he said, were or ought to be the foundation of a just society.

    This was a very bold claim, and had Rawls been able to fulfill its promise, it would have been a monumental achievement. As we shall see, Rawls very early recognized that the original version of the theorem was unprovable, and indeed false. In response to this realization, he made sweeping changes to his theory, resulting in the distinctive form that the theory takes in A Theory of Justice, but unfortunately, the revised theory is not more defensible than the original. Rawls himself seems to have realized this fact, for while repeating the language of "theorem" and "proof," he very considerably backs away from the strong claims that he made in the earliest published version of his theory.

    Before we begin the detailed examination of the argument, let me take just a moment to explain why I believe it is appropriate to bring the tools and insights of Game Theory to bear on A Theory of Justice. That is, after all, not the customary manner in which we engage with the arguments of Hobbes, Locke, Rousseau, Mill, Kant, or any of the other great figures of the Western tradition of democratic political theory. Quite simply, the reason is this: A great part of the plausibility of Rawls' theses derives from his claim that they can be grounded in a formal argument of Bargaining Theory. Absent that claim, the reader is left simply to contemplate Rawls' political theory and consider whether he or she likes it. The argument for the theory is, when all is said and done, the claim that the two principles would be chosen by rationally self-interested individuals situated in what Rawls eventually came to call the Original Position. If that is simply not true, then it is hard to see what other justification Rawls has for his theory.

    It is actually rather difficult to figure out exactly what Rawls' Two Principles mean, and the only way I can see to grapple with them is to take Rawls at his word that they are the solution to a bargaining game, and then see how we might so construe them. In this case, as we shall see, the formal machinery of Game Theory is quite helpful in guiding us to turn Rawls' non-technical language into something precise enough to be subjected to analysis.

    It will be useful for our purposes to begin with the earliest statement of Rawls' theory, as it appeared in an article entitled "Justice as Fairness," published in 1962 in an important collective volume of essays called Philosophy, Politics, and Society, Second Series, edited by Peter Laslett and W. G. Runciman. Two passages from the essay will set things up for the first stage of my analysis.

    "The conception of justice which I want to develop," Rawls writes, "may be stated in the form of two principles as follows: first, each person participating in a practice, or affected by it, has an equal right to the most extensive liberty compatible with a like liberty for all; and second, inequalities are arbitrary unless it is reasonable to expect that they will work out for everyone's advantage, and provided the positions and offices to which they attach, or from which they may be gained, are open to all."

    These principles, Rawls says, would be agreed upon unanimously in a deliberation that he characterizes roughly in the way that "state of nature" political theorists describe the agreement on the Social Contract that constitutes a nation. Although he acknowledges that his remarks "are not offered as a rigorous proof" that persons engaged in this deliberation would agree on the two principles, such a proof requiring "a more elaborate and formal argument," nevertheless, he goes on to say:

    "[T]he proposition I seek to establish is a necessary one, that is, it is intended as a theorem: namely, that when mutually self-interested and rational persons confront one another in typical circumstances of justice, and when they are required by a procedure expressing the constraints of having a morality to jointly acknowledge principles by which their claims on the design of their common practice are to be judged, they will settle on these two principles as restrictions governing the assignment of rights and duties, and thereby accept them as limiting their rights against one another."

    I think it is patently clear that Rawls in these passages is laying claim to the rigor and demonstrative power of Game Theory, at least in its somewhat looser form as Bargaining Theory. He seeks to show that his proposition is a necessary one; he asserts that it is intended as a theorem. Throughout the long evolution and transformation of his theory, Rawls never gave up this claim, for all that he also never came close to providing an argument for it. It is, I believe, the heart and soul of his entire enterprise. Without it, he has nothing but a rather affecting, albeit extremely murky, expression of his personal preferences in social organization.

    What can Game Theory tell us about Rawls' claim? There are two questions that we must try to answer: What do his two principles mean? and Is his assertion a theorem that can be proved with necessity?

    First, a problem: In the original statement of the principles, Rawls states his first principle thus: "each person participating in a practice, or affected by it, has an equal right to the most extensive liberty compatible with a like liberty for all. " In A Theory of Justice, however, the reference to practices is omitted, and instead we get "Each person is to have an equal right to the most extensive basic liberty compatible with a similar liberty for others." [p. 60] Eventually, this is tweaked a bit, and becomes "Each person is to have an equal right to the most extensive total system of equal basic liberties compatible with a similar system of liberty for all." [p. 302] The original formulation is thus intended to apply to practices, such as marriage, or capitalism, or the military, or the judicial system. The final formulation applies to nothing less than the total organization of a society. The shift, as we shall see, makes it forbiddingly difficult to figure out what the principle actually means.

    The first principle, in all of its variants, uses the phrase "the most extensive." That implies that one can rank alternative arrangements of a practice, or, alternative sets of fundamental or constitutional arrangements, in order of the degree of liberty that they embody or promise or make possible or guarantee. But as the term "liberty" is ordinarily used in the context of debates about political systems, it refers to a wide variety of institutional arrangements or guarantees that vary along multiple dimensions. The right to trial by a jury of one's peers, we may suppose, is a form of liberty. So is the right of all adults to vote periodically in elections to select the members of the government. Is a system of government with the first but not the second a more extensive or a less extensive liberty than a system of government with the second but not the first? One might reply, it does not matter, because a system of government with both is superior to either. Hence it would be Pareto preferred to either. But suppose there are liberties that in some of their forms are incompatible, in the sense that guaranteeing one interferes with guaranteeing the other.

    For example [if I may be a trifle facetious], imagine a society consisting solely of authors and literary critics [these days, what with the internet, it does seem as though everyone is either an author or a critic, and sometimes both]. Now authors wish to be free of what they consider unfounded attacks on their writings, and critics wish to be free of what they consider unjustified limitations on their critiques. So the critics will prefer the American system of libel law, which gives very broad latitude to critics, and authors will prefer the British system, which favors those supposedly libeled [unless, of course, the authors wish, in their writings, to say nasty things about the critics, in which case the situation is more complex.] There is no way in this situation simultaneously to maximize the liberty of both groups, since each extension of the liberty of one will be experienced as a loss of liberty by the other. Consequently, as the authors and critics gather to bargain on the founding principles of liberty in their society, they will find it impossible to achieve unanimity on Rawls' first principle, because every attempt to spell out what it means will have them at loggerheads. This problem is not at all trivial, for all that my example may make it seem so. It is referred to in economics as the Indexing Problem, and it will crop again in a different guise in the final version of Rawls' theory.


  1. Do you think there is no justification of the Original Position than aesthetic preference then? Or do you think it has some weight apart from Rawls' arguments?

  2. As I see it, the notion of the Original Position is a striking and suggestive way of capturing what other authors have described as impartiality or disinterestedness or The Moral Point of View. One of the things it does rather forcefully is to make us think through what that really involves. Thus, if being disinterested [or neutral or "objective"] means setting aside our knowledge of who we are and what we want as individuals, then clearly as disinterested persons we can have no desires or aims or intentions. The comparison with judges merely adjudicating cases beought before them captures this.

    Rawls' problem, as I indicate, is that disinterested persons can have no motive for bargaining, unless you reintroduce some purposes or preferences or goals into their mental machinery.

    This, by the way, is why Kant in the GROUNDWORK was forced to introduce the notion of Objective Ends or Ends in Themselves, having ruled out all heterogeneous willing.

  3. I find your authors/critics example rather puzzling since, from the way that you described it, the quantity of liberty for each is (in the simple case) strictly opposed. Therefore, shouldn't they, when the come to bargain, each be willing to give up some of their liberty? Rather than a counter-example, this seems to me to be exactly the sort of situation Rawls has in mind. The real problem would come in if there were truly incommensurable amounts of liberty, which I take to be the case most of the time.

  4. Hmm. You may be right. It may indeed be a bad example. But then again, maybe not. Rawls is, I think, trying to capture the notion of a Pareto partial ordering, with all parties preferring a greater liberty [compatible with a like liberty for all] rather than a lesser liberty. But faced with the choice between the British and American systems of libel laws, there could be no unanimous agreement that either was preferable to the other, since each system advantages one group to the disadvantage of the other. So we are left with a situation that cannot be handled by a unanimous agreement. Am I missing something?

  5. I see; I didn't understand your example. I guess I was assuming that there was some spectrum of possible libel systems in which Britain and America represented the extremes. If this were the case, then you could bargain for an intermediate position. But the situation as you described it is one where there are two discreet choices, one that favors one group and the other the other. Yes, I see how that would be a real problem for Rawls and something that would probably happen often.

  6. Problems of the author/critic type don't just bother Rawls -- Gauthier runs face-first into them as well, for similar reasons. Think of these cases as two criticisms of political philosophical systems for the price of one.

    It should be said that there are many virtues to Rawls' book and project -- the method of reflective equillibrium, some important observations about the structure of political theories and how they should interact with personal morality -- which don't feature in the bit of his book where he harnesses some formal methods. Rawls is by no means the only philosopher with aspirations for formal proofs where the brunt of the work gets done in the margins (give me half a chance and I'll tell you about Gibbard's meta-ethics and how his use of decision theory is using a little bit of butter for a whole lot of bread).

    Prof Wolff, you're running through your material at quite some speed -- any thought to continuing on with more material after you've finished your original run?

  7. Right now, I feel like the Roadrunner, after he has raced fifty feet beyond the cliff edge and suddenly notices there is nothing underneath him. I am also writing and posting my Memoirs at a mad pace [450 pages since the beginning of April!] All of this is exhilarating but exhausting. In about two or three weeks, I am going to have to pause and catch my breath. At that point, I will welcome suggestions and discussion [as I do now, of course]. This is a bit like my first teaching job, at Harvard, after getting my doctorate in philosophy, when I found myself teaching the history of Europe from Caesar to napoleon, despite never having taken a history course of any sort beyond my junior year in high school. I stayed a century ahead of the students and prayed.
    A propos the method of reflective equilibrium, I am less enamored of it than some people, for reasons I discuss at length in my book [this entire blog, I now realize, is a shameless excuse to shill for my books! Oh well, live and learn.]