The Online Library of Liberty

A project of Liberty Fund, Inc.

• Foreword
• Preface
• The Reason of Rules
• 1.: The Constitutional Imperative
• I.: Introduction
• II.: Reasons For Rules
• III.: Rules of Games
• IV.: Rules of the Road
• V.: Rules of the Market Order
• VI.: Rules of Political Order
• VII.: The Importance of Rules
• 2.: The Contractarian Vision
• I.: Introduction
• II.: Noncontractarian Constitutionalism
• III.: Individuals As Sources of Value
• IV.: Contract and Exchange
• V.: Politics In the Exchange Perspective
• VI.: Unanimity As the Contractual Ideal
• VII.: Agreement On Rules and the Veils of Ignorance and Uncertainty
• VIII.: Conclusions
• 3.: The Myth of Benevolence *
• I.: Introduction
• II.: Private Good and Public Good
• III.: Science, Truth, and Politics
• IV.: The Authoritarian Imperative
• V.: Majoritarian Democracy In the Noncontractarian Paradigm
• VI.: The Aim of Politics
• 4.: Modeling the Individual For Constitutional Analysis
• I.: Introduction
• II.: Homo Economicus In Politics: the Argument For Symmetry
• III.: Science and the Empiricist Defense
• IV.: A Methodological Defense of the Differential Interest Model of Behavior
• V.: Social Evaluation and Quasi-risk Aversion
• VI.: Gresham’s Law In Politics
• VII.: Summary
• 5.: Time, Temptation, and the Constrained Future *
• Preface
• Part 1.: Individual Private Choice
• I.: Introduction
• II.: The Ultimate Z ’s
• III.: Preferences For Preferences
• IV.: Past, Present, and Future
• Part 2.: Individual Public Choice
• I.: Introduction
• II.: Society With a History
• III.: Temporal Interdependence
• IV.: An Illustration
• V.: Moral Rules And/or Constitutional Commitment
• 6.: Politics Without Rules, I: Time and Nonconstrained Collective Action
• I.: Introduction
• II.: The Social Discount Rate
• III.: The High-tax Trap
• IV.: The Inflation Trap
• V.: The Public-debt Trap
• VI.: Other Examples
• VII.: Conclusions
• 7.: Rules and Justice
• I.: Introduction
• II.: Just Conduct and the Notion of Desert
• III.: Justice and Promise Keeping
• IV.: Justice Among Rules
• V.: Just Rules, Agreed-on Rules, and Just Conduct
• VI.: Conclusions
• 8.: Politics Without Rules, II: Distributive Justice and Distributive Politics
• I.: Introduction
• II.: Distributive Justice: the Conventional View
• III.: The Constitutional Perspective and Institutional Incidence
• IV.: The Incidence of Unrestricted Majoritarianism
• V.: Tax Rules and Distribution Under Majority Rule
• VI.: Direct Constitutionalism and Distributive Justice
• VII.: Summary
• 9.: Is Constitutional Revolution Possible In Democracy?
• I.: Introduction
• II.: Pareto-superior Change and Wicksellian Unanimity
• III.: Distributional Limits and Prospective Rules
• IV.: Status Quo Entitlements and Distributional Envy
• V.: Constitutional Change and Free Riders
• VI.: The Role of Norms
• VII.: Toward a Civic Religion

IV.

Rules of the Road

Rules of the road, another familiar usage of the terms here, are not designed and/or did not evolve on the basis of any specification of the objectives of persons who are road users. Road users have widely varying purposes—business, pleasure, or some combination—which dictate many varieties of route, speed, and type of vehicle. Rules of the road serve the function of allowing persons to pursue their separate and independent courses, which may conflict in the absence of such rules. These rules do not imply that the objectives of users be reduced to a single counter, analogous to “winning” in ordinary games.

Road rules draw another feature to our attention. The efficacy of a set of rules does not depend on any matching of skill levels among those who use the facility. A set of rules may be preferred because it tolerates the coexistence of good and bad drivers on the road, a feature that does not apply to ordinary games. Road rules have a social function, which is to facilitate the achievement of the purposes of all persons who use the facility, regardless of what these purposes might be. And the rules are adjudged in accordance with their ability to satisfy this criterion.

In much the same way, the rules that constrain sociopolitical interactions—the economic and political relationships among persons—must be evaluated ultimately in terms of their capacity to promote the separate purposes of all persons in the polity. Do these rules permit individuals to pursue their private ends, in a context where securing these ends involves interdependence, in such a way that each person secures maximal attainment of his goals consistent with the equal liberty of others to do the same?

Concentration on the road rules example allows us to isolate another feature that is often overlooked. Rules provide to each actor predictability about the behavior of others. This predictability takes the form of information or informational boundaries about the actions of those involved in the interaction.

For example, suppose that in a small, developing country, automobiles are new and few in number. There has been both French and British influence in the country, so that early road users include both right-hand and left-hand drivers. As the number of automobiles increases, the absence of an established rule creates problems. Independent adjustment by each driver when two vehicles meet, with neither driver knowing how the other will react, produces a pattern of outcomes that is analogous to life in the Hobbesian jungle. All parties will be better off if they adopt a rule, any rule.

Matrix 1.2 illustrates this case. The game here is basically a coordination game, in that the rule adopted serves an informational purpose. Each of the two parties is given the ability to predict what the other will do. And it does not matter, by supposition, whether the rule adopted involves right-hand or left-hand driving, so long as the rule generates symmetric behavior. In such a case, there may be a role for government in announcing a rule. History, however, may do as well or better—for social conventions often serve to establish the relevant rules of conduct.

The interaction depicted in Matrix 1.2 differs from the more general dilemma game in Matrix 1.1 in the relative importance of the predictability content of the rule and the subsequent problem of enforceability. Matrix 1.2 depicts a game that is basically one of coordination; the major gains are secured on the adoption of a rule, any rule, and there is relatively little advantage to any player to be gained from defection. As shown in the matrix, however, there is some gain from defection and, hence, some enforcement problem. If A knows, for example, that B will always follow an agreed-on rule, then A will occasionally find it advantageous to depart individually from the adopted rule. But the temptation to violate the rule, once adopted, is not omnipresent as in the more general prisoners’ dilemma setting.

A pure coordination game (not depicted in matrix form here) would be one in which the advantage of individual defection from adopted or conventional rules would be wholly absent, one for which there would be no enforceability problem at all. Some such interactions surely exist. Language might be thought about in these terms. All persons in a social community have an incentive to use words that others understand. There is a natural force generating a common vocabulary and grammatical rules. The same description might be applied to the language of manners and etiquette whereby the apparent object of behavior is to convey meaning of some sort to others.

Other important characteristics of either the basic prisoners’ dilemma interaction or the information-coordination interaction are obscured, however, in the Matrices 1.1 and 1.2. Both illustrations are directed toward the ultimate choice between a rule and no rule. A second choice may involve a choice between rules, once the more inclusive game has been played, that is, once the need for a rule has been accepted by all parties. Consider, then, the case in which there is a difference between possible rules, even if we retain the assumption that there is symmetry in payoffs between the players. The “game” described here is a really a “subgame” of that illustrated in Matrix 1.2.

Consider Matrix 1.3, which takes place “within” cell I of Matrix 1.2. The options of the two parties in this case are not those of adopting a rule and adjusting behavior without a rule. The choices are those confronted in the set of alternative rules. As depicted in Matrix 1.3, the rule “Drive right” dominates the rule “Drive left.” It is important here to have a rule (shown in Matrix 1.2), but the question of which rule is also important. And because of the symmetry in payoffs between the players, both will, when given the constitutional choice, select the same rule.

Two points are worth making about the interaction depicted here. First, social conventions that emerge historically and take on the status of “unwritten rules” do not necessarily produce the best conceivable pattern of outcomes. Some modern social analysts (notably Hayek and his followers) display an apparent faith in the forces of social and cultural “evolution” to generate efficient rules. There seems to be no reason to predict that these forces will always ensure the selection of the best rules. In our example, the “Drive left” rule might well emerge and prevail—particularly if exogenous changes alter the relative payoffs to different rules over time. There may then be little or no evolutionary pressure toward the emergence of superior rules. This prospect alerts us to the need, periodically, to review alternative sets of rules and to regard rules themselves as objects of choice, to be changed and redesigned according to the patterns of social states they generate. The prospect also alerts us to a possible role for “government” in the collectivity, that of facilitating a shift from old to new rules. “Government” in this context can be variously construed—as a consensually appointed assembly, the entire set of relevant players, or, at the other extreme, some random dictator-king. Since, in this example, the gains are symmetric, there are no particular advantages to being the rule chooser, but it may be important to have some person, group, or process that is empowered to choose among rules.

Second, the move from “Drive left” to “Drive right” may not be desirable, despite the dominance of the latter in the matrix. If rules are viewed as providing information to enable the players to predict each other’s actions, it follows that any change in the rules destroys information. If the rule (“Drive right” or Drive left”) is determined afresh each morning by the toss of a coin, there is no rule at all. In order to function, rules require stability. If rules are continually subject to change, the information they provide becomes negligible. Each player can no longer take it as given that others will abide by the rule in existence, even if he knows it himself, because he cannot know that others will know that he knows it. And when others may be playing by “outdated” rules, each has less incentive to play by new ones.

This argument suggests that there is a natural predilection toward conservatism in the constitutional perspective. The mere demonstration that state A would be “better” than the status quo, once state A were achieved, is not sufficient to demonstrate that a move from the status quo is justified. A “local” maximum may turn out to be a global maximum once the local maximum has been achieved.

Recognition of this fact reveals a crucial distinction between constitutional design and constitutional reform. In constitutional design, where there are no effective preexisting rules, all that is relevant is the choice between the rule that generates one set of outcomes and the rule that generates an alternative set. The rule that gives rise to the preferred set of outcomes is to be preferred. But when there is the question of changing an existing rule, as is the case in constitutional reform, the rule that generates the most preferred set of outcomes carte blanche is not necessarily dominant.

The argument here lends some force to the social evolutionist’s antipathy to constructivist zeal. To the extent that stable and tolerable rules exist, a community may well be better off not to attempt change. Recognizing this claim does not, however, commit us to the view that the explicit reform of existing rules will never be desirable. The argument merely alerts us to the need for rules concerning the procedures by which existing rules might be changed, and in particular for ensuring that rule changes do not take place too often and without proper recognition of the transitional costs.

The basic coordination games depicted in Matrices 1.2 and 1.3 are simplistic in another important dimension. Quite apart from the ubiquitous conflict between individual and “social” interest, which creates the enforceability problem, is the disagreement among individuals over the choice of rules themselves. This conflict potential has been deliberately suppressed in the coordination games discussed to this point. There is no difference between the two players in the ordinal ranking of the cells in the matrices.

Consider, however, a different example, still within the more general rules of the road category. For reasons already noted, it is clearly advantageous to have some rule; the setting is identical to that described in the “Drive right”-“Drive left” example. But suppose that there are two possible rules for behavior at intersections, only one of which can be chosen. One rule is “Give way to the right”; the other is “Give way to the left.” Matrix 1.4 illustrates this interaction. Note that the ordinal ranking of the two relevant cells differs between A and B, with A much preferring the first rule, “Give way to the right,” and B much preferring the second rule, “Give way to the left.” Such divergent rankings may occur if, for example, A predicts that he will, on most mornings, approach an intersection from the right of B’s approach.

The two players prefer different rules despite the fact that both prefer either rule to no rule. Because of the disagreement over which rule to adopt, however, there may be delay and dispute between the participants, each of whom will seek to maximize the distributional advantage promised by a choice among alternative rules.

The differential advantage placed on differing rules by different persons should not be overemphasized. To the extent that rules are long standing and that persons anticipate that they may occupy different positions in sequential plays of the game, the players may tend to reach agreement on the rule to be adopted much more quickly than the simple analysis implies. In our example, if the players predict that each will sometimes approach intersections from the right and sometimes from the left, the interaction can be modeled as in Matrix 1.2 rather than in Matrix 1.4.4

[4. ]For a general discussion of the principle that agreement on which rule to adopt is less difficult to achieve than agreement on strictly defined distributional allocations, see James M. Buchanan and Gordon Tullock, The Calculus of Consent (Ann Arbor: University of Michigan Press, 1962).