On Treating Like Cases Alike.

Idealism, intelligence, and importance are a rare combination. Politics by Principle, Not Interest, the book on "non-discriminatory democracy" by James Buchanan and Roger Congleton, achieves the feat of combining them, though the occasional strain shows that the three are not natural bedfellows. The book is enormously ambitious: it sets out to prove that by enshrining the principle of treating like cases alike ("the generality principle") in the constitution, the body politic could be purged of much sordidness and conflict. There would be no winning and losing coalitions, for coalitions could no longer take wealth from one another. Yet with "politics as taking" barred, the deadweight costs of redistribution and rent seeking would be saved, and all sides would be not only more fairly treated but also better off materially. Imposing the generality principle on politics would be a Pareto-improving move. Politics by Principle pleads for the move to be made.

The book is remarkable in at least two respects. The first is its courage in exposing the sacred cow of what it calls "majoritarian" democracy (is any "real-existing" democracy not majoritarian?) in all its immorality and inefficiency. Admirable, too, is the painstaking application of the analytical schema to such fields as externalities, public services, fiscal policy, interpersonal and intergenerational redistribution, and social insurance, implying that if the generality principle works, it does so right across all fields usually regarded as the proper domain of collective choices.

Three theses are put forward. The first, best called Contract, argues the possible legitimacy of coercion by collective choice and establishes a link to the contractarian tradition that now constitutes the mainstream of political thought. The second, Cycle, argues that because collective choices dealing with distributions of benefits and burdens are unstable and every distribution is dominated by another, winners cannot win permanently and would in fact be better off under a constitution that, by imposing generality, abolished the rotation in which winners become losers and vice versa. The third and main thesis is Generality, a formal rather than a substantive principle that, if applied, puts distributional conflict out of bounds by equal treatment, the treating alike of like cases.

Here is a massive job for the Devil's Advocate. The present essay traces the line of attack he could be expected to follow. The author of the essay wishes to add, for the record, that he would not be displeased if, on some judgment day, the plea of the Devil's Advocate were found to have failed and that of Buchanan and Congleton to have prevailed.

Contract

Pivotal to the argument is the lesson to be drawn from David Hume's parable about draining a swampy meadow by what is called "collective action" in modern jargon. In his Treatise of Human Nature, Hume tells us that some neighbors agree to dig a drainage ditch, to the benefit of their adjacent farms. "Agreement" here means more than the shared opinion that it would be nice to have a drainage ditch. It is a reciprocal promise by the neighbors to dig it. It is, in effect, a contract. Hume goes on to say that such promises are the means by which men achieve shared ends, from the digging of ditches and the building of bridges, harbors, ramparts, canals, and armies to the formation of governments. He is quite explicit in giving contract causal priority over the state, which most other theorists say is needed before a contract becomes a binding commitment. The contract here binds the parties because it is self-enforcing. In modern parlance, we would say that the parties see this interaction as a single link (a "node") in a chain of interactions stretching indefinitely into the future, with a probabilistic and ever-receding endpoint. The parties ("players") choose to perform according to their contract, as well as future ones to be entered into, because in comparing the relative merits ("payoffs") of performing as promised or defaulting, they consider the present value of two streams of expected payoffs along the whole chain. Performing prolongs the chain; defaulting (taking the "free rider" strategy) damages the chain or breaks it off altogether, so that the defaulter gets few or no chances to default again. Hume, in fact, implicitly employs the Folk Theorem of modern game theory, which shows the cooperative strategy in repeated prisoners' dilemmas under certain conditions to be a Nash equilibrium.

Buchanan and Congleton, taking the drainage ditch as an isolated collective enterprise, present the usual, and perfectly correct, solution that for each farmer, not digging is always a better strategy than digging, whatever the other farmers do; hence there will be no ditch, nor any other collective action. The farmers cannot credibly bind themselves by contract, and for their own good they need to be coerced by a contract enforcer, the state.

By this reasoning, contractarian philosophy explains the individual's willing subjection to a collective choice mechanism, backed by a monopolistic coercive agency, as a "social contract" that could be agreed to by all rational persons. The necessity for coercion breeds consent that, in turn, lends legitimacy to the state.

Buchanan and Congleton's restatement of the doctrine is more explicit and precise than most. It presents a clearer target for the Devil's Advocate than do more fudged and elusive versions. It has, or so he would argue, three easily identified chinks in its logical armor.

The first has long been recognized by critics of all forms of contractarianism, but to the best knowledge of the Devil's Advocate, no effective defense of it has ever been offered. If man can no more bind himself by contract than he can jump over his own shadow, how can he jump over his own shadow and bind himself in a social contract? He cannot be both incapable of collective action and capable of it when creating the coercive agency needed to enforce his commitment. One can, without resorting to a bootstrap theory, accept the idea of an exogenous coercive agent, a conqueror whose regime is better than anything the conquered people could organize for themselves. Consenting to such an accomplished fact, however, can hardly be represented as entering into a contract, complete with a contract's ethical implications of an act of free will.

Because prior to the construction of the collective choice mechanism the purported agreement must be unanimous (for, remember, there is as yet no rule for reaching non-unanimous decisions), no party to the social contract must be asked to accept a worse position, taking good periods with bad, than he would occupy without the contract. Here is the second chink in the doctrine's armor. Buchanan and Congleton stress that contractarianism "rules out payoffs that are sensed to be negative by any participant ... over the whole set of anticipated political choices, the base point being defined by positions achieved when the majority rule, collective action game is not played at all" (p. 18). However, life without any collective action is difficult to imagine except on desert islands. If it is conceivable at all, it entails unmitigated misery. Because the authors believe categorically that collective action presupposes a coercive authority, and because almost anything is a better payoff than that available at the "base point," it follows that almost any terms, no matter how harsh or lopsided, must be accepted by the rational player. In terms of the economist's familiar device, the situation is like a very, very long Edgeworth box with a very, very long contract curve running inside it, with all points on it from one end to the other being bargaining solutions, better for both (or all) players than no contract. The actual solution may be outrageously good for one player and outrageously bad for the other; we do not know how the bargain will turn out, except of course that it will be Pareto optimal. But if the alternatives are rigged in such a way that the terms of mutually acceptable bargains can vary from one gross extreme to the other, from princely privilege to abject submission, the condition of universally positive payoffs and unanimity is trivial.

The third weak spot reveals itself in the attempt to remedy the second. Contractarianism uses a veil of some kind to obtain certain behavioral results. In the original Buchanan and Tullock version, the social contract was concluded behind a veil of uncertainty; the lack of visibility concerned the future. People did not seek social contract terms favoring one situation over another, because finding themselves in one kind of situation seemed no more probable than finding themselves in another. John Rawls, of course, operated with a very different device, the veil of ignorance, behind which people were unaware of their personal endowments and indeed of any differences that distinguished them from one another. Behind the veil of ignorance, contracting parties became, in the strictest sense, each other's clones. Somewhat drastically, Buchanan and Congleton resort to a double veil they call "the veil of uncertainty and/or ignorance," presumably obliterating knowledge of the future, or one's self-knowledge, or both, to suit all occasions. They argue, plausibly enough, that the veil or veils will powerfully narrow the bargaining problem, because parties unaware of their own interests cannot have conflicts over terms and should find it easy to agree on what seems fair and impartial to both.

It is no doubt as easy to reach unanimous agreement on innocuous terms (and perhaps on any terms whatever) under these conditions as it is to agree with oneself. The result is still trivial, no less so than the one reached without the help of veils. But it is a little disappointing to find perhaps the foremost champion of methodological individualism of our time producing this result by resorting to a device that effectively obliterates individuals and opens the back door for some holistic entity to take the place they can no longer usefully occupy.

Cycle

Politics by Principle recalls the well-known, and somewhat overworked, proposition of formal social-choice theory that if "society's choice" is taken as the largest number of individual preferences attracted by one alternative state of affairs, certain preference configurations will produce an incoherent result under which, given pairwise choices by at least three persons (or groups) among at least three alternatives, any "socially chosen" alternative will be dominated by another. Consequently, the actual choice must go round and round the available alternatives in a cycle. It is widely accepted that such cyclicality, in which any distribution of income or wealth is overturned by another that is "socially preferred" to it, should obtain in a society where majority coalitions are freely formed and able to expropriate the minority. For Buchanan and Congleton, the rotating nature of distributions, with today's gainers becoming tomorrow's losers, is the saving grace of majoritarian democracy, ensuring that no group in society will be a permanent loser from redistribution.

The Devil's Advocate will contend that this conclusion would undergo radical revision if due distinction were drawn between a pure and a non-pure distribution "game" and also between distributional and electoral rotation. He will consider these distinctions in turn.

Consider a pure three-person Neumann and Morgenstern distribution game, where a coalition of two players can decide how a fixed stock of goods, say, one hundred golden eggs, shall be divided between them and a third player. With players A, B, and C entitled to form one two-person coalition and take one distributive decision per period, and with an exogenously given bargaining solution to tell us how the members of the coalition will share the golden eggs between them and how many they will leave for the third player, the following table shows how the solution will rotate over a full cycle of three periods. In each period, the payoffs in italics indicate which two players will form a coalition in the next period.

 Player

Period A B C

 1 50 40 10
 2 10 50 40
 3 40 10 50

Whatever the bargaining solution (provided it does not vary from period to period), over any full cycle the three players come out even, with an average share of thirty-three golden eggs. The present values of the three streams of payoffs differ a little because of the timing of high and low payoffs, but this difference is a minor matter if the discount rate is not exorbitant. Any coalition can lift its average payoff over the three periods by sticking together. For example, once A and B have secured their ninety golden eggs in the first period, if B sticks with A, both will keep having the permanently higher payoff of ninety between them, whereas C will be permanently exploited.

Myopia and bargaining difficulties may, for all we know, weaken the dominance of the stable over the cyclical solution. Consider, however, the incomparably stronger likelihood of the stable, noncyclical pattern of exploitation as we move from a pure to a non-pure three-person distribution game. There is now both production and consumption in each period. When, in period 1, player A has the most golden eggs, the redistributive payoff available to the coalition of B and C dominates that which any coalition that includes A can gain; hence the coalition BC will form to exploit A in period 2. B and C consume what they take from A. Player A, however, is a goose laying golden eggs. She produces, within the period, the same number of eggs that B and C take from her. Under this condition, at the end of period 2 the distribution is the same as at the end of period 1, and forming the coalition BC is still more profitable than forming any coalition that would include A--and this remains true at the end of every subsequent period. The "socially preferred" solution is perfectly stable.

Empirical evidence clearly demonstrates that in modern democracies redistribution is, broadly speaking, always from the richer half of the electorate toward the middle (which harbors within it the median voter) and the poorer half, with no rotation, which is what we should expect from looking at the distribution game with production and consumption, as well as on grounds of common sense. The richer half is not made poor and the poorer half is not made rich, by turns, as the Cycle thesis would predict.

However, Buchanan and Congleton invoke a different set of historical evidence to the effect that democracies tend to display an irregular but inevitable electoral rotation; sooner or later the team of political entrepreneurs in power is voted out and another team is voted in. Here, the authors seem to equate an electoral with a distributional outcome, instead of distinguishing between the two. One team of politicians will normally demarcate itself from the other by adopting different slogans, professing different values, and proposing different policies whose impact on the pattern of redistribution is small or opaque, hence difficult to perceive. It is, however, almost inconceivable that the first team should offer redistribution from rich to poor and the second team from poor to rich, which is what would be needed if electoral rotation were to amount to distributional cycling. The second team's offer would be strongly dominated by that of the first in terms of transfers and "targeted" public goods and services. We can assume that the larger sums in the first team's offer would swamp any effect that might be exerted by the intangible elements of the two teams' offers. To have a chance of winning the bid for the votes of one-half of the electorate plus a margin to make sure of the median voter, both teams have to formulate rich-to-poor redistributive offers. One would expect to capture the poorer half's allegiance, the other that of the richer half, with the latter voting for the lesser of two evils. Schematically, this is how democracies must function, and do in fact function.

However, although neither the rotating solution of pure distribution games nor electoral rotation is relevant to real-life redistribution under majoritarian democracy and there is no cycle, still it is always the same half of society that it is exploited by the other half. Stopping this exploitation by inserting an appropriate rule in the constitution would not simply put an end to futile and costly back-and-forth sloshing of wealth, as Buchanan and Congleton appear to believe. If generality were such a rule and had the effect in question, it would permanently hurt the exploiters and permanently help the exploited. It could hardly be Pareto superior. If it were, we should have to wonder why society has never groped toward adopting it.

Generality

Politics by Principle proposes that constitutions should prohibit the makers of collective choices--whether democratic majorities, dictatorships, or anything in between-from resorting to other than general rules in determining distributions of benefits or burdens.

Such reform may or may not be feasible. Assuming for argument's sake that it is, principle would replace interest, and the reformed constitution would be going against the grain, forbidding the very things the decisive force--for example, the majority coalition--most wishes to do. What entitles us tacitly to suppose that such a constitution could be durably enforced? Following the public choice tradition of treating constitutional rules as being on a different plane and being determined by different interests from the "in period" choices such rules permit, Buchanan and Congleton do not treat this problem. Even if such difficulties were not prohibitive, but could be overcome by paying whatever enforcement costs were required, it is legitimate to ask whether such costs would not offset the savings that the greater political efficiency of general rules promises to bring.

However, this objection is but a quibble compared with the major and fundamental one that concerns the very concept of generality. Indeed, the Devil's Advocate submits that it is logically impossible to formulate constitutional clauses whose strict respect would constrain the lawmaker from passing any rule on the sole ground that the rule would violate the generality principle. Generality cannot be sufficiently defined to allow us to tell rules that are general from rules that are not.

Statute laws, judicial precedents, conventions, bylaws, customs, taboos, decrees, regulations, and administrative directives are all rules of varying degrees of stringency and precision. Every rule resembles every other in at least one respect: each rule is an invariant relation between a class of defined circumstances (here denominated "cases") and a set of consequences (sanctions for the breach of behavior norms; distributions of benefits and burdens) that must result if the rule in question is applied to the case within the class in question. The same cases must bring forth the same consequences if the rule is in fact applied.

Buchanan and Congleton plead, in the name of fairness and what they call "political efficiency," for rules in all domains, and notably in the political domain, to conform to the generality principle. If rules applied in politics were general, some of the worst features of majoritarian democracy would be purged or at least greatly mitigated. Distributions would be stabilized; they would become better in the normative sense of being less dependent on the will of the politically stronger battalions. The socially wasteful expenditure on promoting distributional interests would cease. The outcome would be Pareto superior; no coalition under majoritarian democracy could expect to do better, taking good years with bad.

What meaning is being imputed to such a promising principle? When the authors speak of "treating like cases alike" or "equal treatment under the law," they seem to be saying, uncontroversially, that a rule must be applied to all the cases to which it is applicable, that there must be no positive or negative deviations in treatment between such cases according to arbitrary whim, fear, or favor unprovided for in the rule itself. In short, they state that a rule is a rule--an analytic statement that conveys no additional information about generality, for in this sense every rule is a perfectly general rule and conforms to the principle postulated.

More controversially, they use "generality" in a different sense when they say that general rules do not discriminate among persons. Two objections arise. One is a shade pedantic but not wholly superfluous, for it adds some clarity to what might easily degenerate into confused rhetoric about equal "rights"--though, to their great credit, Buchanan and Congleton never resort to such vocabulary. The objection is that rules do not directly target persons, much less discriminate among them, but are directed at cases in which persons may play various roles. Breaches of contract, for example, are cases in a class to which a rule applies and provides for certain consequences. In such cases, particular roles are played by plaintiffs and defendants. The rule discriminates between them. For example, it places the burden of proof on one party rather than on both of them. Likewise, a rule about lawmaking may attribute certain powers to representatives, others to senators. There is discrimination between them, but the rule about lawmaking is a general one for all that--if the term generality is indeed more than a tautology.

The more serious objection provoked by the authors' recurrent characterization of generality as non-discrimination is that this requirement contradicts the very logic of rules. Their function is to separate the cases that require a particular uniform treatment from the rest of the universe of cases that do not require the same treatment. Non-discrimination is inconsistent with treating like cases alike and different ones differently, except in the logical limit where from some infinitely distant celestial perspective, every case is like every other.

The more nearly we approach this logical limit, the more curious is the resulting "generality," understood as non-discrimination. Suppose a rule requires that the destitute be given food and shelter by some of those who have these things. Generality is doubly violated, both in the entitlement to alms and in the selective nature of the obligation to give. To satisfy generality, either nobody must give nor get or all must give and get. There are a few sound reasons for the first alternative, though most people would think the balance of arguments is tilted against it. But there are no good reasons speaking for the second version, whose very generality raises the suspicion that satisfying the generality principle, even if it is interpreted as more than a tautology, may not in itself make for better rules.

By way of demonstration, consider alternative rules for distributing punishments. One rule might lay down one punishment for all felons ("off with their heads") and another for minor offenders (a suspended jail sentence). This is obviously not a general rule, discriminating as it does between felonies and misdemeanors--though it is vastly more general than any real-life criminal code with its fine gradations of offenses and punishments. A large step toward greater generality would abolish the discrimination between types of crimes. It would require that all criminals have their heads chopped off or all get a light suspended sentence, but not that some should be beheaded and others not. Buchanan and Congleton, incidentally, make the case that general rules are politically efficient in that they reduce the cost of administering the system as well as the wasteful expense people incur to avoid the negative or to benefit from the positive discrimination available under non-general rules. This cost-reducing effect is clearly visible as we move, in the hypothesized case, from a more to a less discriminating rule of distributing punishments. In the true logical limit of the generality principle, however, where every case is like every other and is treated as such, it is the final discrimination, that between criminals and noncriminals, that falls to the ground; everybody's head is chopped off or everybody gets off scot-free.

Although the "treat like cases alike" meaning of generality is no more than a tautology for "apply the rule," the "non-discrimination" meaning clearly does not bear pushing anywhere near its logical limit--a telling sign that it has some defect that comes to light when the meaning comes to be stretched a little. In fact, Buchanan and Congleton are far too intelligent not to sense this problem. They are fairly diffident about providing a working definition that would tell us, in every case, what rule would pass for general.

In one basic case, though, they are perfectly confident about what rule would be the truly general one. This case is the two-person (or two-coalition), two-strategy game of benefit- or burden-sharing, as exemplified by Hume's farmers digging a drainage ditch. A two-by-two matrix describes four alternative allocations of the workload. Along what the authors call the diagonal, both farmers dig for two days or neither digs. Along the off-diagonal, either one digs for three days and the other for one, or the other way round. Which of the two off-diagonal, asymmetrical alternatives is the actual solution depends on which "farmer" is enabled by the political choice mechanism to coerce the other. A rule applied to this type of case is general if it outlaws the off-diagonal solution, so that only the symmetrical, equal-sharing solutions remain available. Both farmers work the same length of time. Obviously, the Pareto-optimal solution among all the symmetrical ones is that they should both work as long as it takes to complete the ditch; but this is not the point. The point is that generality has been found to reside in symmetry.

From here, a promising avenue seems to lead toward a more developed form of generality. The simple version of the rule would say that the farmers, when placed in the circumstances described, should dig the same number of days, preserving one kind of equality, albeit a rather rudimentary one.

However, there is no compelling reason why equality of days worked should be regarded as the best, let alone the sole valid criterion of symmetry. If one farmer is frail, old, or arthritic, or if one has a higher opportunity cost because when he digs he cannot attend to the calving or lambing, or if his end of the ditch has a nasty, sticky, clayey patch, then a rule laying down equality of labor time might well be held to produce asymmetrical shares of pain or cost. Likewise, it might also be argued that symmetry calls for labor time to vary inversely with productivity or directly with the benefit each farmer will derive from the drained meadow. Symmetry in the relevant variable must prevail, but why is labor time the relevant variable rather than pain, productivity, opportunity cost, benefit, or something else?

The "equal treatment" schema of a distributive rule is "to/from each, according to...." The blank space is filled in by some function f(x), where x is the reference variable common to a class of cases, deemed relevant for holding that those cases ought to be treated alike, and differently from other cases. The treatment is coded in f(x). If x is a constant, the rule becomes "to/from each that is x, the same" (that is, absolute rather than proportionate equality).

What this formula fails to decide is whether the relevant variable is x, y, w, z, or none of them. Which, if any, meets the generality principle? Don't they all do so despite being mutually inconsistent?

One of the most famous dialogues in Through the Looking-Glass could well have been written about symmetry, non-discrimination, and the generality principle: "When I use a word," Humpty Dumpty said in rather a scornful tone, "it means just what I choose it to mean--neither more nor less." Alice is doubtful about this declaration: "The question is," said Alice, "whether you can make words mean different things." But because it is a matter of opinions, feelings, and "values," not of "facts," which variable is the relevant one that makes one case like another and unlike a third one, and because different people may legitimately hold different views, generality can mean neither more nor less than what each chooses it to mean. Alice is wrong, and Humpty is right.

Before like cases can be treated alike, it must be decided which case is like which other case. Any one case is described by an indefinitely large number of variables, many of which would be dismissed as unimportant by most reasonable people, leaving a number that are important for at least some people. Every case is like every other, because all are "cases." A rule incorporating a reference variable is applicable to all cases that are relevantly described by that variable. These are, then, the "like cases" that the rule treats alike. Many things can be said about the chosen variable: that it is not as relevant as some other, that it is insufficiently inclusive (hence discriminating) or too inclusive (hence indiscriminate, perhaps absurdly so). Ultimately, however, all such observations are intrinsically subjective and can be reduced to my say-so against your say-so. My general rule is your special one; my non-discriminating rule is your indiscriminate one. How does a constitutional court tell my generality from yours?

The inconclusive search for generality is inseparable from arbitrariness and discretion. Between the two extremes "every case is like every other" and "no case is like any other," certain cases are like certain others if one variable is chosen as relevant for rule-making, and other cases are like yet others if relevance is judged differently. The champions of each rule will claim that theirs is the truly general one, but such claims are irrefutable and function as value judgments.

What would be the general rules of public finance? Buchanan and Congleton consider that to ensure generality, the relevant variable for the rule governing transfers of all kinds should in fact be a constant, namely, "human being"; whether pauper or millionaire, each should get the same poll grant. However, to ensure generality on the revenue side, the relevant variable should be "income," and the rule should be that all incomes are taxed at the same flat rate (making some people pay more tax than others). Rival "political entrepreneurs" might well appeal to some constitutional court against this rule, arguing that a rule that does not discriminate between incomes but taxes them all at the same rate does manifestly discriminate among income recipients, by treating bachelors and the rich more favorably than heads of large families and the poor, or by failing to compensate for positive and negative externalities associated with the earning and spending of certain incomes, and so forth. Who can decide whether generality requires, permits, or excludes taking account of such matters? And if on the expenditure side a poll grant satisfies generality, why does an income tax (whether flat or progressive) and not a poll tax satisfy it on the revenue side?

The answer seems to be that Buchanan and Congleton seek rules that combine a measure of simplicity and inclusiveness with a curb on redistributive infighting without being hopelessly impractical (as combining the poll tax with the poll grant would be). Once they have selected rules likely to fill this bill, they call them "general," or more general than alternative ones. But this approach involves arguing from desirable effects to generality, rather than from generality to its desirable effects. Their rules are good rules, perhaps the best one can think of if one shares their political predilections. The Devil's Advocate does. Yet he would not be doing his job if he did not insist that although the tautological sense of generality ("a rule is a rule") is unhelpful, no meaningful sense of generality is as yet in sight, to be proposed as a formal constitutional constraint on admissible rules. Disappointingly, legislating for generality seems to be about as effective, and as constraining, as legislating for the common good.

Suppose, however, that the defense of generality were to admit that ultimately it is a value judgment to say that one rule applicable to one class of cases is general, another applicable to another is not, or is less so; but then to point out that everybody's value judgment rejects as a basis for classifying the cases certain notorious "reference variables," such as the race, sex, color, or religion of the persons playing certain roles in a class of cases. Would this approach not serve to ward off the charge of subjectivity and special pleading, and provide an unassailable, firm foundation for generality to rest on? Such a universally rejected set of variables can tell the constitutional court that rules differentiating between cases according to these variables are not general, if indeed generality is no more and no less than "general approval," that is, unanimity in value judgments. However, the rejected set cannot help determine which of the countless remaining, actual or potential, rules differentiating between cases according to the countless remaining unrejected variables is general and which is not. Moreover, whatever the foundation of the taboo variables, it is not a firm but a shifting one, contingent on cultural change. Currently taboo variables are liable to be eroded or swept away by the next wave of political correctness, whose reigning value judgments might well seem outrageous, and outrageously un-general, to our own generation.

Conclusion

Here, the Devil's Advocate rests his case. Buchanan and Congleton have had their day in court. Additional submissions may come, from them or from sympathizers or adversaries, believers or agnostics, before the jury can retire. An interim summing up, however, may be in order at this stage.

Idealism or intellectual convictions persuade Buchanan and Congleton that constitutions determine politics, rather than the other way round. If so, it may be possible to purify politics by constitutional reform, making both the winners and the losers of the old, sordid game of political tug-of-war better off. Enshrining the right clause, hitherto missing, in the constitution would force people willy-nilly to give up the politics of interest. Generality is said to be the clause that would achieve this result. Only rules that satisfy generality in governing distributions of benefits and burdens, rewards and punishments should be constitutionally admissible.

In treating like cases alike, all rules satisfy the generality condition; it is, in this sense, a tautology. Is there another, meaningful sense that would permit rules that are general to be distinguished from those that are not? Does generality require us to treat every case like every other? Or does it permit us to treat each case on its merits as unlike every other? If the answer to both questions is "no," where between these extremes does generality lie? Inclusiveness and non-discrimination, besides their logical and semantic defects, raise the difficulty that equal treatment of cases according to one variable will normally entail their unequal treatment according to other variables--offering free public education to all children treats families with children better than the childless, and so on through endless examples. What is deemed to satisfy, and what to violate, the generality principle does not appear to rest on any intersubjective criterion but ultimately on somebody's value judgment. The question to be decided, then, is whether a proposition whose meaning is contingent and discretionary--like the value judgment that lies behind it, with no stronger claim to validity and truth than its rivals--can function as a major constitutional constraint on the domain of collective choice.

Anthony de Jasay is an independent scholar and author living in Paluel, France.

Article Details

Printer friendly Cite/link Email Feedback
Author:	DE JASAY, ANTHONY
Publication:	Independent Review
Geographic Code:	1USA
Date:	Jun 22, 1999
Words:	5868
Previous Article:	Doomsday Every Day.
Next Article:	Race Policy in Three American Cities.
Topics:	Discrimination in justice administration Analysis Justice discrimination Analysis

Related Articles

Do the courts treat everyone equally?

live reviews: Mostly Autumn Liquid Room, Edinburgh, May 20 **; singles and albums.