Much discussion of democracy in the United States, popular as well as scholarly, employs simple, descriptive models of that democracy. The most commonly encountered of these is what I call the "vote-centered" model of democracy. Under this vote-centered model the public policy outcomes produced by legislatures are traceable to equally weighted voter inputs. Another model that makes frequent appearance in the literature about democracy is an "interest group" model, under which democratic outcomes are depicted as "equilibrium" states in struggles among competing powerful, organized groups. For a variety of reasons, I do not think that either of these models(1) does a very satisfactory job of integrating the phenomena of American democracy. I will have something to say later about the interest group model, but the deficiencies in the dominant vote-centered model are particularly glaring and, because of its dominance, particularly important. The vote-centered model will be my principal foil in this Comment, as I advance another possibility, what I call "democracy as meaningful conversation," under which the citizenry is engaged by ongoing public conversation about public policy, and it is this engagement that is the stabilizing force in the system.
The existing descriptive models are not typically referred to as "models." They are not in any sense formal models. Indeed, they are usually implicit in discussions of democracy, rather than explicit. Even if implicit, however, they are models in the sense that they encapsulate American democracy by reference to certain central features. And they are simple models by virtue of the fact that the features they employ are few in number. As these simple models are made explicit, certain difficulties in the modeling process come into focus that are probably best highlighted from the outset.
Models can be descriptive or normative, or even both at once. While the line between the two is in principle tolerably clear, it is also difficult to heed. Descriptive models describe what is, frequently ascribing causal connections among parts of what is modeled, and even predicting results to be expected if changes are made.(2) In contrast, normative models provide an ideal to be strived for, or perhaps only dreamt of, but that need not now exist, or even be attainable. Still, models advanced as description are often likely at least to insinuate normative judgments. For all descriptive models are selective. They choose some features of the system they purport to model to the exclusion of others. The simpler the model, the more selective it will be. And if what a model identifies as encapsulating the modeled system is seen as a desirable feature rather than an undesirable or a normatively neutral one, then the model perforce has a normative twist.
Holding the line between descriptive and normative models is additionally complicated by the fact that models consciously designed as normative are seldom greatly divorced from the reality they seek to instruct. If the distance is too great, the task of bridging it will likely seem too substantial to justify the bother. For this reason there will typically be a high degree of correspondence between normative models and the portion of the real world in view, so that even normative models may easily be mistaken for description--by the consumers of commentary based on models, but also on occasion by the commentators.
The difficulty of holding the line between description and prescription is especially acute when modeling democracy,(3) probably because the appeal of democracy in the modern day is at once so great and so badly in need of explanation. Whatever the reasons, the descriptive and the normative are thoroughly intermixed in existing uses of the vote-centered model. My criticisms will largely be on a descriptive plane, but it may occasionally be that the criticisms unjustifiably mistake for description what was intended as prescription. And while the inspiration for the conversational model is description, the prescriptive pull has proved irresistible for me as well. What democracy as meaningful conversation purports above all to describe and explain is stability in the system, which it traces to widespread conversational involvement of the citizenry. Stability is usually viewed as a desirable feature of political systems, especially when, as with the conversational model, it is not the product of coercion. For that reason I will often ascribe virtue to conversational phenomena. But I want to be clear that any normativity in the conversational model is decidedly limited. Stability is surely not the only, or even necessarily the highest, virtue in systems of government, nor do I claim that the conversational model identifies the only source of stability in democracies.
Even on a descriptive level, simple models of complex things can easily mislead, for they will inevitably fail to account for a good deal of the complexity. For large scale social phenomena like democracy in the United States, controlled experimentation is unavailable, so that ascribing causal connections must always be done cautiously and tentatively. If two models describe and explain different things, moreover, there will be no common metric by which to judge one more "accurate" than the other, and also no objectively verifiable basis for saying that the one rather than the other has described the "right" thing. Despite these difficulties--or perhaps because of them--the more complex a phenomenon of interest, the more inevitable and useful simple models will be. They break the complexity down into a more useable form, one that can facilitate understanding and discussion. In such a setting the appeal of a model will necessarily rest not on some systematic basis for choice, but on informed intuition about whether the model identifies important things and provides a measure of coherence in its account of diverse phenomena. It is on this basis that I believe that democracy as meaningful conversation succeeds. The description it provides is by no means comprehensive, but the model does point the way to connections among important features of American democracy that the vote-centered model, its most prominent competitor, misses or muddles.
I begin in Section II with some sounds of silence that got me thinking about deficiencies in the vote-centered model and about alternatives to it. I begin with the almost unquestioning acceptance of the apportionment of the United States Senate.
The absence of controversy over the apportionment of the Senate is striking. In the Great Compromise of 1787 that opened the way for agreement on the Constitution, the House of Representatives was apportioned among the states by population (with the significant qualification that each state is entitled to at least one representative), while the Senate, was apportioned by states, with each entitled to two Senators. This equal Senate representation of the states is declared by the Constitution to be unalterable without a state's consent.(4) The obvious consequence is that populous states have less representation per capita in the Senate than thinly populated ones (while citizens in territories and the District of Columbia have none at all). In 1787 the disparity in population between the more and less populous states was already significant(5)--that is what necessitated the compromise--and it has grown over the years. Today the ten most populous states have more than fifty percent of the nation's population. California alone has more than sixty-five times the population of Wyoming, while each has the constitutionally prescribed two Senators.(6) Suggestions are occasionally advanced with apparent seriousness that California be split into two states, but to the best of my knowledge the focus of concern is exclusively intrastate, asserted antagonisms or incompatibilities of north and south, and not at all dissatisfaction with the state's apportionment in the United States Senate.(7)
This seems all the more remarkable given the reapportionment decisions of the United States Supreme Court, especially the 1964 decision in Reynolds v. Sims.(8) In Reynolds the Court decided that both houses of bicameral state legislatures were required to be apportioned by population. In doing so the Court was naturally called upon to explain how it was that the national legislature--most importantly the Senate--is not only allowed but required to deviate from that pattern. There is, of course, specific constitutional language governing the apportionment of the Senate, while the Constitution is basically silent on the structure of state government, save that the states are guaranteed a "Republican Form of Government"(9)--the framers' term for what we would be more likely to call "representative democracy."(10) The Court based its decision not on that "guarantee," which had long been held non-justiciable,(11) but on the Equal Protection Clause of the Fourteenth Amendment. And the Court in Reynolds provided an explanation for the difference in constitutional treatment of the Congress and state legislatures that on its own terms seems persuasive enough:
The system of representation in the two Houses of the Federal
Congress is one ingrained in our Constitution . . . conceived
out of compromise and concession indispensable to the
establishment of our federal republic.
Political subdivisions of States--counties, cities [on which
legislative districts were often based], or whatever--never were and
never have been considered as sovereign entities.... [T]hese
governmental units are "created as convenient agencies for
exercising such of the governmental powers of the State as may be
entrusted to them," and the "number, nature and duration of the
powers conferred upon [them] ... and the territory over which they
shall be exercised rests in the absolute discretion of the State." The
relationship of the States to the Federal Government could hardly
be less analogous.(12)
Persuasive as this distinction might be as a matter of law, it does not explain the absence of controversy surrounding the Senate. Indeed the Court's rationale for the basic holding of Reynolds compounds the puzzle. According to the Court, population-based apportionment of State legislatures--the Court's famous shorthand formula was "one man one vote"--is necessary to give each citizen an "equally effective voice in the election of members of his state legislature," as part of an "inalienable right to full and effective participation in the political processes of his State's legislative bodies."(13) If "full and effective participation" requires population-based apportionment in the state context, however, no reason appears from what the Court had to say in Reynolds why the Senate apportionment would not be defective as well, in which case one would expect to see agitation about its apportionment, or at least some substantial sign of dissatisfaction, even supposing that the obstacles to constitutional change are insuperable.(14) Also striking is that the Court's requirement of population-based apportionment for state legislatures has itself been relatively uncontroversial, even while coexisting in apparent ease with a Senate apportionment it seems--with its talk of "inalienable right"--to brand as in principle foreign to the central tenets of American democracy.
The more I puzzled about popular acceptance of a malapportioned Senate, the more it became clear to me that the notion the Court articulated in Reynolds of what is central to democracy in the United States is pretty far off base. Many difficulties with what the Court had to say have been noted over the years. Thus, as public choice theorists repeatedly tell us, in itself the vote is a decidedly ineffective means by which an individual citizen participates in politics. Demographic and other groupings of voters may tend to vote alike, and as groups they may prove decisive in candidate elections. The Court's emphasis in Reynolds, however, was on the individual's right, and in any reasonably populous district no individual voter has any significant chance at all of having his vote determine the outcome of an election.(15) In that sense, the vote of each individual was already (more or less) equally (in)effective in virtually all state electoral districts, essentially regardless of malapportionment.
If the Court was really striving to equalize the degrees of (in)effectiveness of the individual's vote, moreover, it was on a fool's errand. In the tallying of votes in each district each person's vote was already given the same weight. Each "person" (that is, "voter") already had "one vote." Beyond that, "effectiveness" of the vote as an instrument of electoral decision in the district or state (putting aside for the moment matters other than the vote that might make an individual citizen's voice "effective," or "participation" in legislative processes "full and effective") would be dependent on the distribution of politically salient sentiment. Holding size of the electorate constant, a vote in a district that is politically homogenous stands a much smaller chance of affecting the outcome in the district election than does a vote in a district where contending forces are about evenly balanced." And insofar as "effectiveness" in the state legislature is associated with representation there of a voter's favored political party, nothing that happens in a district election matters one whit unless the state legislature comes to be close to equipoise in party representation. These matters the Court did not purport to address, and if it had attempted to do so, perhaps by requiring that each district reflect the political complexion of the state as a whole, it would have risked sinking the federal courts (and the country) in a quagmire of apportionment litigation. Being essentially powerless to produce either effectiveness or equal effectiveness of the individual's vote in candidate elections, what the Court did was impose a formal equality of individual voting power in the state as a whole on top of the formal equal distribution of voting power in a district election that already existed.(17)
As I got more deeply into these matters, increasingly it seemed that the conceptual problems with the Court's rationale in Reynolds run deep. The decision appears to be rooted in a traditional model of representative democracy in the United States that emphasizes the vote for representatives as its defining and dominating characteristic, as the essence of self-government in our "republican" form of government--a model I call "vote-centered."
In this vote-centered model, there are two discrete stages by which important choices of public policy are made, first election of representatives, and then public policy decisions by those representatives in the form of legislation. The electorate has its decisive say through the vote in the first stage. This vote in candidate elections is advanced as the mechanism within the model of popular sovereignty and self-governance, and it is associated with political equality through equal weighting of votes, and with majoritarianism, often taken to be a corollary of political equality.(18) Separately and together these elements are often advanced in a normative spirit. Popular sovereignty, self-governance, political equality, and majoritarianism are taken to be elements of what a political system should be. At the same time, however, they are taken to constitute a basically accurate description of contemporary American democracy.
Under this model, the electorate's active role is largely completed in the first stage. Despite this, popular sovereignty, self-governance, political equality, and majoritarianism are assumed to carry through descriptively to the second stage as well, to characterize the entire system and the decisions it reaches. If the people as a whole are "sovereign," and that sovereignty is distributed equally, then the legislative decisions are theirs too, and theirs equally, which means that each voter exercises an equal portion of influence in the eventual decisions of democratic government, those taken by the representative assembly.
The mechanism by which this is accomplished under this vote-centered model is that the interests of the electorate are "represented" in the second stage by the representatives voted into office in the first. In this view representative democracy is adopted as an admittedly inferior substitute for bringing the entire people together to make the decisions of government, as the only efficient way in a large scale democracy to provide sovereign decisions by the sovereign people. The accuracy of the representation of the interests of the electorate by representatives is aided by open competition among candidates in the candidate elections, so that the most faithful representers are chosen. Free speech and free press are important then, and also between elections, to shore up the information base of the representers about the interests of the electorate. Still, the process between elections is in the nature of things imperfect, because most people remain silent then. It is the vote where virtually all adult citizens have the opportunity to speak, and it is the vote through which the sovereign people rule. Under the vote-centered model, the process is taken systematically to turn citizen preferences into "majoritarian" legislation. Votes are the crucial inputs in the vote-centered model, but much of the appeal of the model is in its depiction of public policy outputs as referable rather directly to equally weighted input of votes. Under the model citizen respect--and the stability it engenders--is garnered because outputs are systematically related to inputs in this way.(19)
I do not claim that any single theorist of American democracy has explicitly embraced as descriptively accurate all the elements of such a vote-centered model.(20) To the contrary, the problems with it are many and to a great extent both apparent and noted.(21) The vote-centered model is rather something that I have fleshed out from what others say. But something like such a model in both a normative and a descriptive mode seems to have a substantial hold on those who theorize about, and those who just seriously discuss, contemporary American democracy.(22) It is through use of such a vote-centered model, with its projection of the election stage ideology onto the system as a whole, that I am able to make some sense of the Court's quest for "full and effective participation" in "political processes of ... [a] State's legislative bodies," under a banner of "one person one vote" in candidate elections. John Rawls also seems to be working basically within a vote-centered model when he describes a "principle of (equal) participation" requiring that "all citizens are to have an equal right to take part in, and to determine the outcome of, the constitutional process that establishes the laws with which they are to comply."(23) Rawls' theory, like the Court's, has a normative thrust, but each must also think that its normative vision is within shooting distance of the way in which representative democracy actually operates. The spell of the vote-centered model as description is probably most apparent in the repeated characterization of the American system of government and the results it reaches as "majoritarian."(24) Legislative bodies generally take action by majority vote, of course, but that cannot be what is meant when the American system is characterized as "majoritarian," for so do other bodies like the United States Supreme Court that are said b X some of the same commentators to be "countermajoritarian."(25) If the system is said to be majoritarian that must mean that it produces some (more or less) systematic translation of the interests or sentiments of societal majorities into legislation in the image of the vote-centered model.(26)
The problems with the description of the American system of government as "majoritarian" are legion. In the national government of the United States--unlike a number of individual states(27)--not a single decision is entrusted to a majority vote of the electorate.(28) Nor is there any basis for finding an effective majoritarianism in the process.(29) Even apart from the apportionment of the Senate(30) (and also putting aside the role of constitutional rights as "trumps" over legislative decisions, and of the Supreme Court in elaborating and enforcing those rights), any translation of majority sentiments of the electorate at the candidate election stage into majority sentiments of the electorate at the stage of legislating is frustrated in a large variety of ways.
To begin with, the vote for a candidate is a terribly blunt instrument for expressing sentiments on particular issues of governance. It is cast in the secrecy of a polling booth,(31) so that those elected do not know much for certain about the identity of the large majority of their supporters or about the motivations behind their votes. In virtually all elections for the national legislature in the United States, a candidate represents a bundle of positions, and even if he only campaigns about one or a few he will be called upon to vote on many. The vote for a candidate is thus opaque with regard to most of the actual decisions that legislatures will make.(32) Indeed in most elections the vote is probably cast as much with an eye on the past and the present as on the future, when there are still decisions to be made. To be sure, there is communication between voters and their representatives between elections, and the contemplation of a future is always providing incentives for the here and now. But it is entirely unrealistic to think that the members of the legislative body collect the sentiments of majorities in their districts as the basis for decision on many, or perhaps even any, of the votes they will cast. They do not have the ability to do so, but they also do not have the motivation.
The principle reason for the absence of motivation is associated with what is called "the interest group theory of legislation."(33) The basic insight of interest group theorists is that relatively small numbers of voters with large individual stakes in some issue or set of issues will have the motivation to wade in politically on those issues, while voters with smaller individual stakes will not, …