In his Economic Theory of Democracy, Anthony Downs presents three quite different, but related, theoretical claims

STREET-LEVEL EPISTEMOLOGY AND DEMOCRATIC PARTICIPATION*

Russell Hardin, Stanford and New York Universities

In post-war public choice theory there have been at least four devastating theoretical claims against the coherence of democratic theory. The first of these was Kenneth Arrow’s ([1951] 1963) impossibility theorem, which concludes that there is no general, acceptable way to aggregate from individual level to collective level preferences. One might have supposed that democracy does not require this kind of aggregation anyway, because it requires only majority rule, not collective rule. But an implication of Arrow’s theorem is that with simple majority rule we may have cycles and no genuine majority. Suppose we face three independent policy issues. One majority can carry allocation A across these issues over allocation B; another majority can carry B over C, and yet a third can carry C over A. Hence, at least in principle, majoritarianism cannot solve our problem.

In his economic theory of democracy, Anthony Downs (1957) presented three additional theoretical claims, which are quite different from each other but related. First, he gave us a model of how candidates must locate themselves in order to maximize their chances of being elected. This is the median voter model, which says that a candidate must take a position at the median of a normal distribution of voters. A candidate who does not do this can be outflanked by another candidate who takes a position between the first candidate and the median voter. This model assumes away Arrow’s problem by supposing that all policy issues aggregately reduce to a single left-right dimension.

Second, Downs supposed that voters actually have little incentive to vote, because they cannot expect to have any impact on the outcome of any given election. Indeed, they have so little impact that any costs of voting, such as suffering through long queues or foul weather, trump any direct benefit from voting. This claim is a specific instance of the logic of collective action, as generalized later by Mancur Olson (1965).

The third major theoretical claim of Downs is that individual citizens have no incentive even to learn enough to be able to vote their interests intelligently. This immediately follows from the second claim if we suppose that gaining relevant knowledge entails some costs. Oddly, however, this claim seems to run against the first of Downs’s theses, which seems to entail that candidates should attempt to influence voters’ knowledge.

I wish to push these four claims by relating them and, in particular, by subjecting them to an economic theory of knowledge. Knowledge is prima facie central to the Arrow problem, the median voter model, and the issue of voters’ incentive to learn enough to vote intelligently in their own interests if they do vote. One can also argue that Downs’s second thesis, that the individual has no incentive to vote, itself poses a problem of knowledge of a somewhat different kind. The general argument of the logic of collective action and Downs’s narrower version of that argument are both relatively recent discoveries that are not well understood by many people. In general, the individual-level knowledge demands of these issues are no different in kind from the individual-level knowledge demands for ordinary pragmatic choice in daily life and in planning future actions, when the costs of information and the costs of mastering relevant quasi theoretical understandings can be higher than the seeming payoff that will come from them.

I will generally focus on issues of voters’ interests and will not attempt to take on normative concerns, such as moral commitments to foreign aid, opposing abortion, punishing victimless crimes, or retributive views on punishment. These other issues often override concern with interests for particular citizens. A full account of democratic theory would have to deal with them as well as with interests. Unfortunately, knowledge issues around mere interests already call such theory into question.

After first briefly outlining an economic theory of knowledge, or a street-level epistemology, I will discuss knowledge issues in political participation in the following order. First, I will briefly discuss the widely held view that voters do not know enough to vote intelligently. Second, I will discuss candidates’ efforts to place themselves at the median voter’s position. Third, I will discuss the problem of knowledge of the logic of collective action, or at least of its specific application to voting. Fourth, I will discuss the newly multi-dimensional issues of contemporary politics as themselves in part a problem of knowledge that can confound voters’ choices and as a re-invocation of Arrow’s problem in a somewhat new guise. Finally, I will conclude with some remarks on the implications of these arguments for liberties, which are a peculiar class of collective goods, and on the relation of liberties to democracy.

Street-Level Epistemology

The theory of knowledge that I propose to apply to these issues is a street-level epistemology, a theory of the knowledge of the ordinary person. It differs substantially from standard philosophical epistemology. The latter is a theory of how to justify truth claims. It has been developed especially in the context of attempting to understand physics and other sciences. The tenets of such an epistemology are about what criteria make a claim true. The focus is on the matter that is supposedly known, not on the knower. For example, it is about whether Einstein’s theory of relativity is true. We could think of it as a theory of the knowledge of a super-knower or of the distributed knowledge of a society.

An economic theory of knowledge is a theory of why the typical individual or even a particular individual comes to know various things. In an economic theory, it makes sense to say that you know one thing and I know a contrary thing in some context. I might eventually come to realize that my knowledge is mistaken and therefore correct it, especially after hearing your defense of your contrary knowledge. But there is no role for a super-knower who can judge the truth of our positions. We are our own judges. If we wish to seek better knowledge, it is we who must decide from what agency or source to seek it. Street-level epistemology is not about what counts as knowledge in, say, physics, but rather with your knowledge, my knowledge, the ordinary person’s knowledge.

Most of the knowledge of an ordinary person has a very messy structure and cannot meet standard epistemological criteria for its justification. With characteristic force, David Hume spells out our problem: “Our thought is fluctuating, uncertain, fleeting, successive, and compounded; and were we to remove these circumstances, we absolutely annihilate its essence, and it would, in such a case, be an abuse of terms to apply to it the name of thought or reason” (Hume �). Rather than a standard philosophical epistemology, we therefore need a street-level epistemology to make sense of the morass of ordinary knowledge. Street-level epistemology is a subjective account of knowledge, not a public account. I wish here not to elaborate this view but to apply it to the problems of representative democracy. I will briefly lay out the central implications of a street-level epistemology and then bring it to bear on democratic citizenship.

Much of the work on voting behavior and the apparent ignorance of many voters treats the issues as problems of psychological foibles in decision making (see, e. g., Popkin [1991] 1994, esp. chap. 4). Many — although not all — of these foibles can easily be seen as essentially economic constraints on learning how to judge complex issues, but I will generally not discuss the psychological approach to these problems here. Much work also gives a fairly straightforward account of the problem of the status-based economics of knowledge. For example, Robert Dahl (1961: 1) notes that “knowledge, wealth, social position, access to officials, and other resources are unequally distributed” in American politics. I will also not take up this issue but will generally discuss only the general problem of coming to know relevant things for political participation. Bringing unequal positions into such an analysis would be valuable for a complete account of actual participation. Such a move could be fully consistent with street-level epistemology.

In standard philosophical epistemology, it would commonly be incoherent to speak of my mistaken knowledge. Knowledge is, in some epistemologies, “justified true belief.” If I am mistaken in my belief, then I most likely lack justification for the belief. Hence, it is not knowledge. And in any case, the category of justified true belief is a category of somehow public knowledge, not personal knowledge. In a street-level epistemology, there may be no ground for claiming in general that my knowledge is philosophically justified in any such sense. There is commonly only a story to be told of how I have come to have my belief. There is therefore little or no point in distinguishing between belief and knowledge, and I will not do so. Typically, at the street level, the term “belief” is commonly used when the substance of the knowledge is a particular kind, such as religious knowledge. There is often no other systematic difference in degree of confidence in knowing those things that are labeled as knowledge and those that are labeled as belief. Indeed, people with strong religious convictions would commonly claim to know the truth of the things they believe religiously far more confidently than the truth of many simple, objective things they might also claim to know. It is true that we sometimes use the term belief to allow for doubt, as when we say, “I believe that’s the way it happened, but I might be wrong.” But this hedge applies to virtually all our knowledge.

Standard philosophical epistemology is concerned with justification, that is, justification of any claim that some piece of putative knowledge is actually true. Street-level epistemology is economic; it is not generally about justification but about usefulness. It follows John Dewey’s (1948: 163) “pragmatic rule”: In order to discover the meaning of an idea, ask for its consequences. In essence, a street-level epistemology applies this to the idea of knowledge, with consequences broadly defined to include the full costs and benefits of coming to know and using knowledge. Note that the pragmatic or street-level epistemology is essentially an economic theory. But it is not the economic theory that presumes full knowledge, as in rational expectations theory or much of game theory. And it is not merely about the costs of information, as in some economic accounts.

Again, standard philosophical epistemology focuses on the matter of belief, for example, on the orbit of a planet. It is about truth and the justification of truth claims. An economic theory of knowledge focuses on the individual believer or knower, on the costs and benefits of coming to know, which, of course, vary from person to person and context to context. Perhaps the chief way in which standard epistemologies do not fit much of our ordinary knowledge is that the bulk of our knowledge — perhaps virtually all of it — depends on others in various ways. We take most knowledge on authority from others who presumably are in a position to know it. Indeed, we take it from others who themselves take it from others who themselves take it from others and so forth all the way down. There are finally no or at best vague and weak foundations for most of an individual’s knowledge.

Trudy Govier (1997: 51-76) argues that our knowledge therefore depends on trust. It might be better to say it depends on the trustworthiness of our authoritative sources, although even this is too much. Very little of our knowledge seems likely to depend on anything vaguely like an ordinary trust relationship. I personally know none of the authoritative sources for much of what I would think is my knowledge in many areas. It is not so much that I take that knowledge on trust as that I have little choice but to take it. If I do not take it, I will be virtually catatonic. I am quite confident that much of what I think I know is false, but still I rely on what I know to get me through life because I have to.

Hence, the knowledge that you or I have is from a vast social system, not from anything we actually checked out. Much of it can only be generated by a social system. We depend on knowledge by authority because it is efficient and because, without division of labor in generating our knowledge, we would have no time for putting much of it to use. Since what we mainly want is to use it, we settle for taking it on authority rather than seeking to justify it. We have to rely on others or massively restrict our lives. As Ludwig Wittgenstein (1969: §170) says, “My life consists in my being content to accept many things.” Henry Sidgwick similarly noted that to live at all is prior to living well and if we are to live at all we must accept many things that don’t have reason as their source.

Knowledge How To Vote

The central epistemological concern in representative democracy is what the typical citizen knows about the actions of public officials. If, in general, we make the effort to know something in large part because it serves our interest to know it, then we cannot expect people to know very much about what their representatives do. In the argument of the economic theory of democracy, a citizen typically does not have very much interest in voting. One vote has a miniscule chance of making a difference, so miniscule that, even when it is multiplied by the value of making a difference and getting one’s preferred candidate or policy, the expected value of the vote is miniscule (see further below, “Understanding Whether To Vote”). Hence, if there is any real cost involved in casting a vote, that cost swamps the expected benefit to the voter of voting. Hence, by the pragmatic rule, there is little point in knowing enough actually to vote well.

The conclusion that we have no incentive to learn enough to vote well was part of Downs’s argument, and it had been even more central to the earlier argument of Joseph Schumpeter. As Schumpeter wrote, implicitly invoking his own pragmatic rule, “without the initiative that comes from immediate responsibility, ignorance will persist in the face of masses of information however complete and correct” ([1942] 1950: 262). I may have reason to acquire knowledge because it gives me pleasure, but not because it will be useful in my causing good public effects through my role as citizen.

Most of the subsequent research and debate on voting have focused primarily on the incentive to vote rather than the incentive to know enough to vote intelligently.[1] The latter is at least logically derivative from the former, because it is the lack of incentive to vote that makes knowledge how to vote well virtually useless, so that mastering that knowledge violates the pragmatic rule. Just because my vote has miniscule causal effect on democratically determined outcomes, there is no compelling reason for me to determine how to vote by assessing the causal effect of my vote on such outcomes. Or, to put this the other way around, the fact that I would benefit from policy X does not give me reason or incentive to know about or to understand the implications of policy X unless, by the pragmatic rule, I can somehow affect whether policy X is to be adopted.

If the citizen has no interest in voting, then the citizen has no interest in making the effort to learn enough to vote well. Something that is not worth doing at all is surely not worth doing well. If the problem of knowing enough to judge government officials is already hard, the lack of incentive to correct that problem is devastating. Indeed, the costs of knowing enough about government to be able to vote intelligently in one’s own interest surely swamp the modest costs, for most people in the United States, of actually casting a vote, at least on commonplace issues of public policy outside moments of great crisis. The economic theory of knowledge or street-level epistemology therefore weighs against knowing enough to vote well because the incentives heavily cut against investing in the relevant knowledge. The typical voter will not be able to put the relevant knowledge to beneficial use.

In what follows, I will simply take for granted that typical citizens do not master the facts they would need to know if they were to vote their interests intelligently. There is extensive evidence on this claim, although there is, of course, also great difference of opinion on its significance for electoral choices. For example, Popkin canvasses problems of voter ignorance in American presidential elections but he then refers to “low-information rationality,” which is rationality despite abysmal factual ignorance (Popkin [1991] 1994: 78). He also argues for a Gresham’s law of political information: bad facts drive out good facts. This law is that “a small amount of personal information [on a candidate] can dominate a large amount of historical information about a past record.” The personal information might be some minor thing that comes up during a campaign. The trouble with the large amount of historical information that is, at least in principle, available is that voters do not typically know much of it because it would be silly for them to invest the time needed to learn such information.

As evidence of how little voters even seek better information before voting, consider the difficulty candidates have in getting their message across to voters. Richard Fenno (1978) elegantly displays the burden that candidates for the US House of Representatives face in finding people to talk to.

Even professional political scientists, who have a strong interest in knowing more about politics than their mere interest in the outcomes of elections would suggest, find it hard to keep up with much of what happens. Weekly tallies of votes in the US House of Representatives and Senate, for example, are reported in some newspapers, but with such brevity that their meaning is often opaque to anyone who has not followed the relevant issues very closely, more closely than many of those newspapers do.

Results of referendum votes on even relatively simple issues suggest astonishing misunderstanding by voters (see further, Hardin forthcoming). California voters displayed cavalier irresponsibility in a recent referendum on a so-called three-strikes sentencing law that mandates harsh minimum prison terms for repeat offenders (Estrich 1998). In an early case to which the new law was applied, a one-slice pizza thief was sentenced to a term of 25 years to life, with no possibility of parole before serving at least twenty years, for his “felony petty theft” (New York Times, 5 March 1995). And they have apparently displayed complete misunderstanding of a referendum to undo a mistaken law on open primaries. That law, passed in ignorance of its consequences by the state legislature, would disallow California representation at the national party nominating conventions in the year 2000. After both the legislature and the electorate failed on this issue, administrative devices were proposed to enable the state to distinguish Democratic and Republican voters in primary elections (New York Times, 15 November 1998). In this failure, democracy was a charade and, when it failed from ignorance, we let a knowledgeable bureaucratic agency act against the democratic result.

Median Knowledge

The argument of the median voter model of elections is demoralizing. It implies that a mere census of voters and their positions would define the median voter and therefore the outcome of any election. Then what is the point of the elections? Ostensibly, each candidate uses the electoral process to influence voters by convincing them that their real positions center on the positions of that candidate. To put this in a positive light, we could say that election campaigns are about giving voters the knowledge they need to vote intelligently in their interest. To put it in a negative light, we could say they are about deceiving voters into thinking their interests are other than they are or that a candidate’s position is other than it is. Or perhaps there is a middle way and campaigns are about giving voters any insight at all.

Suppose voters are generally quite ignorant of the nature of issues of objective importance to them and of the stances of candidates on those issues. In such an ignorant population, the median of the distribution of voters is not well defined. Moreover, it can be volatile. Because voters are ignorant, their positions may be relatively unstable and subject to sudden change in response to new information (Popkin [1991] 1994: chap. 4). This means that a candidate’s effort to inform voters can be risky. If voters were well informed, however, campaigns would have little effect unless candidates could generate new issues. The very fact of campaigns suggests the general ignorance of voters. The quality of campaigns, perhaps especially in the United States, suggests that candidates believe that voters are abysmally ignorant.

In principle, if a candidate could do a good enough job of convincing voters where they — the voters — stand, that candidate could even overcome the Arrow problem by violating its condition of universal domain. That condition stipulates that all conceivable preference orderings might actually be held in the population. An effective politician might shift enough people into some particular preference ordering as to make that ordering the majority preference. One might suppose, however, that this would typically be a false achievement, because that candidate’s success would be de facto a matter of deceiving voters about the voters’ own interests.

Understanding Whether To Vote

When I have taught the logic of collective action, it has often taken a lot of persuasive effort to get the argument across at the general level to many of the students. Even after I have seemingly managed to do that, however, some students have immediately argued against it in particular applications, such as voluntary payment of union dues in order to gain the collective benefits of union protection of workers. This is, in a sense, a surprisingly different result from the intellectual history of understanding the logic. Typically, the logic has been understood in a particular context, such as John Stuart Mill’s ([1848] 1965: 958; Hardin 1988: 92-94) discussion of the need for legal enforcement of a shorter work day in order to overcome the inherent incentive of individuals to freeride on a voluntary agreement to work less by getting paid extra for overtime.

Consider the logic in the context of voting in elections in substantial polities. The odds against a voter’s ever making a difference are overwhelming. There was a tie vote in a local election in New Jersey in 1994 — this otherwise trivial vote became national news because it was the exceedingly rare case of a tie in which one more vote could have made a difference. There have also been votes that were de facto ties in larger elections in which the counting error is too great to know who really has won in a very close count. In the New Hampshire election for the United States Senate in 1974, Louis Wyman and John Durkin were virtually tied at about 111,000 votes each, with various state agencies giving a slight edge to the Republican Wyman and others to the Democrat Durkin during the contest over the vote. Eventually, the US Senate declared the election undecidable and declined to seat either candidate. The vote was then retaken in a special election (Durkin won by a substantial margin). This odd election shows that merely for practical reasons of the impossibility of counting votes accurately, one more vote is unlikely to make a difference in an election even in as small a polity — less than a quarter of a million voters — as New Hampshire, one of the smallest states in the United States. The individual voter essentially does not count.

This relatively common sense claim may actually mislead us on just how little a single vote matters. The very best chance of my vote making a difference would come if, de facto, all other voters one by one tossed a fair coin and voted for A when heads turned up and for B when tails turned up, while I voted definitively for A. With a hundred million voters, my vote would have so little chance of breaking a tie even in this extraordinarily evenly matched election that, even if I valued the election of A at $1000, my vote would be worth only eight cents (Hardin 1982: 60).

Despite such numbers and such supposed logic, many voters claim forcefully, and seem to believe, that it is in their interest to vote. Moreover, they demonstrate their commitment to this view by actually voting. The incentive problem is conspicuously overcome for about half the voters in US presidential elections and for even more of the voters in most other contemporary democracies. One might give an account from a theory of knowledge of why people believe their votes matter. If such an account is successful, it then overcomes Downs’s logic, and we might understand why there are substantially higher voting turnouts than that logic, if only it were well understood, would allow.

In typical economic choices, it seems to make sense to suppose that people understand their own interests reasonably well. In collective action contexts, however, this may not make sense. A standard quibble with rational choice theories is that they require individuals to do what best serves their objective interests. If individuals act otherwise, then the theory is thought to fail. Unfortunately, this move short circuits the mental task of weighing one’s objective interests in acting one way rather than another. If I conclude that my interest is other than what some objective analysis by a critic says it is, it does not follow either that I am right or that, if I am wrong, I act irrationally in acting according to what I believe to be my interest. A fully adequate account of my rationality must account first for my beliefs and then for my action from these beliefs.

We generally have no difficulty with this two-stage requirement on judging someone’s rationality in certain contexts, such as medical decisions. For example, it would be odd to say that George Washington acted irrationally in allowing the best medical people available to him to bleed him, possibly killing him or hastening his death, during his final, perhaps minor illness. He did the best one could do within reason, which is to say within the economic or rational constraints on what he could come to know about medical care. When he died, his doctor may have concluded that he had not bled the great man aggressively enough and might therefore have determined to bleed people who suffered from, say, a cold and fever, more aggressively lest he lose them too. Given his sadly mistaken view of the efficacy of bleeding, that might even have been the right response.

Our task here is to assess the rationality of individuals’ beliefs about the rationality of acting for collective benefit in the face of the logic of collective action. Many of those who vote do so for moral reasons of their duties or the fairness of their doing their part. But many seem genuinely to think it their own interest to vote. They invoke a rational choice version of the generalization principle in ethics (Singer 1961; Hardin 1988: 65-68). That principle is a response to the query: What if everyone did that? For example, what if everyone took a short cut through a lovely lawn or garden as I wish to do in this moment because I am lazy or in a hurry? Well, if everyone did that, there might be an ugly path through the splendor of the lawn or garden. But one could often answer the query by noting that everyone does not, and evidently will not, do that.

Many voters seem to believe in a pragmatic (non moral) version of the generalization argument. They feel responsible if, after they fail to vote, their party loses. And if their party loses after they do vote, they console themselves with the realization that at least they tried. If they had merely a moral commitment to voting, they should feel guilty for not voting independently of whether their party wins or loses. To feel regret because one’s party loses makes no sense unless one supposes one might actually have made a difference.

I have tried to explain the logic of collective action as it is played out in voting to many people on many occasions, including many politically active and sophisticated people. I was not trying to dissuade them from voting but only to defend a casual comment on how voting is motivated, if at all, by moral commitments. Or I was trying to give an analysis of why costs of voting dissuade many people. For example, it is much harder to vote in New York City today than it was to vote in Chicago or Philadelphia when I lived in those cities. And the turnouts in New York are much lower than in those cities. I have attempted to explain that this fact of differential costs, and not some cultural claim about diverse New York, made sense of the variance in behavior. Very sophisticated people have simply rejected the entire argument. And they have commonly asserted a pragmatic generalization argument in favor of voting as in their and everyone else’s interest. Utterly against the self-evident facts of low turnouts, they often even held that everyone understands their principle and must reject my argument.

For some years, I thought that people would eventually come to understand the nature of this problem. I no longer believe that. I believe it is easier to understand the logic of collective action and to apply it to real problems of choice than it is to understand, say, the theory of relativity, quantum mechanics, or the workings of DNA. But for many people these are all equally incomprehensible — which is to say, not at all comprehensible. A major reason for their failure of understanding is that people typically suffer no grievous consequences from not understanding these things. As a result of their lack of understanding, some people may put a bit more effort into voting than they might under difficult circumstances, such as vile weather during a New Hampshire primary in February. But that is not a great loss. It is, however, a striking fact that, although they seem not to understand their actual incentives in voting, people nevertheless do respond to those incentives to some extent, as the comparative behaviors of voters in New York City and Chicago shows.

It would be a greater benefit to many people to understand the tax laws better than to understand the logic of collective action better. And they do not even master the tax laws very well. If many of my academic colleagues are indicative, they typically pay far more than they should under the law (and I probably do so as well). Then why should they be bothered with the logic of collective action? It is a professional hobby of rational choice theorists to understand it and maybe that fact makes it easier for them to understand it. Or maybe rational choice theorists are self-selected from people who find it relatively easy to follow such logic and economic reasoning more generally. They are then motivated by the interest they have in getting arguments right in contexts in which getting them wrong leads to public embarrassment. They are not motivated very much by the actual usefulness of putting the logic to work in their lives — although a few have become adamantly determined to live by its conclusions and not to vote.

Suppose we conclude that it is plausibly rational for a person not to master the argument of the logic of collective action and that they therefore vote despite lack of objective interest in doing so. Then why do they seem to follow the logic in not investing in the knowledge they would need to vote intelligently? This is arguably the central theory-of-knowledge question in political participation. As a first answer to it, we can note that any kind of sustained action might fail because it gives too little positive feedback to fool people into believing their own action is benefiting them. But a relatively simple, single-shot action with a more or less immediate outcome might make understanding the interests at stake in the action much more difficult. The difference between voting and knowing enough to vote one’s interests is partly a difference between a single-shot action and a sustained pattern of actions. The sustained pattern of investments in knowing about political candidates may never be related to any sense that it made a difference, whether mistaken or correct.

Another difference between the two investments — in voting and in knowing enough to vote intelligently — is that there is a substantial public discussion of the first but very little of the second. People come to have an active belief about the value of voting, but they may have none of any consequence about the value of knowing enough to vote intelligently. It is easy enough to understand that people learn wrong theories and then even apply them, especially when the consequences of their application are not substantially painful. It is not merely that people have failed to grasp the logic of collective action but that they have been actively proselytized for a contrary belief. Rather than investigate that logic for themselves, they take the view of the authoritative proselytizers.

To change the focus of these observations, we can address the question from the benefits side and from the costs side. There are no significant objective benefits either of voting or of mastering the knowledge to vote intelligently in one’s interest. That there are no benefits of mastering the knowledge merely for the purpose of choosing how to vote follows, of course, from the claim that there are no benefits from voting. For many people there seem to be, however, perceived benefits from voting — they evidently stop working through the issue before reaching the Downsian conclusion. This view may be a mistake, but it is not irrational, anymore than it is irrational of me to get the arithmetic wrong in balancing my bank statement. Moreover, that conclusion was apparently not formally reached by anyone until very recently.

If we are being proselytized to believe in the benefits of voting, then the costs of coming to perceive the logic against voting have been raised, perhaps very high for many people. It would be much harder to proselytize people for the view that massive study of current politics and politicians is in their interest, because the real costs of such study would be substantial and readily felt. In conclusion, we can say that there are real differences in both the benefit and the cost functions of voting and of learning enough to vote intelligently. The differences in both of these suggest more reason to expect people to vote than to expect them to be well informed enough to vote intelligently in their interest.

Finally, we may note that people often do understand the logic of collective action, perhaps especially in relatively local contexts in which the cooperative contributions would be costly and readily perceived. The slogan, “Let George do it,” is grounded in a recognition of the logic. It may have taken a very substantial level of proselytizing to get people not to see that logic in the case of voting, although the proselytizers may themselves have been led to believe in their own preachment.

Among those who have done best at understanding the logic of collective action have been union leaders and members. This makes sense from the fact that theirs has been a problem of collective action from the beginning. They have long been forced to recognize that voluntarism does not work very well and that coercive laws are necessary to induce contributions of dues and efforts. It was, as noted earlier, in the context of collective action for workers that Mill recognized the logic. Samuel Gompers, an early leader of the American union movement, asserted the logic clearly in 1905 (Gompers 1905). He had learned the logic through extensive, no doubt painful experience at a level most of us will never even vaguely match. Perhaps the first I ever heard of the logic of collective action was from a union neighbor before I was a teenager. I asked him how he had voted on election day. He replied that he had not bothered to vote because his vote would not have mattered and it would have been a lot of trouble because he would have had to knock off work early or miss his pre-dinner beer on the patio. He took a dime from his pocket and said, “My vote is not worth this much.” Yet he understood the election very well in the sense that he knew how he should have voted if he had voted. He predicted, rightly, that Dwight David Eisenhower would defeat Adlai Stevenson with ease because, he supposed, people liked Ike and were not smart enough to see that Stevenson would serve the interests of most of them better.

When the costs of voluntaristic collective action are substantial and clear and, perhaps especially, when the significance of joint action is driven home repeatedly in some context, such as union or neighborhood organizing, people can grasp the logic of collective action clearly. Generalizing it, however, is quite another matter. Only with Olson’s generalization in 1965 did the logic become generally clear to large numbers of social scientists with a strong professional interest in understanding it. The best and the brightest of them had typically got it wrong before. Downs himself famously slipped on his own argument when he supposed that the final reason people do vote must be to prevent the collapse of democracy. This bad argument — arguably some orders of magnitude worse than the argument that it is in my interest to vote in order to affect the outcome of the present election — is commonly and perversely cited by critics as a refutation of Downs’s central arguments.

In sum, many people do vote despite the absence of a personal benefit from doing so. Hence, the Downsian instance of the logic of collective action is de facto resolved for a substantial fraction of the electorate. Therefore, the problem of investing in enough knowledge to vote intelligently may well be the more fundamentally serious issue in democratic theory, as Schumpeter seemed to think. To assess whether it is, we would need even more extensive studies of the fit between votes and objective interests than we have had so far.

Multi-Dimensional Issues

Arrow’s theorem differs fundamentally from Downs’s in its assumptions. In Arrow’s preference orderings, there is no reason to suppose that there is a simple left-right dimension along which individuals can order their rankings of states of affairs. There could be as many dimensions as there are issues that might be relevant to defining states of affairs. In some respects, Arrow’s vision of the nature of the issues over which a collective is choosing is a much better fit to European and North American national politics today than is Downs’s. The left-right model of Downs’s normal distribution arguably fit the politics of the 1950s better than it fits today. I therefore wish to discuss the median voter model in the context of an electorate facing issues that are not simply arrayed along a single dimension. Because of this characteristic of contemporary issues, there may be much severer knowledge demands for a voter today than there were in the earlier era, as I will argue below.

In a sense, the era in which the Downs model was a plausible approximation was a golden age of simplicity, an age that may have lasted on and off for a couple of centuries in the Anglo-Saxon world, and for a century or more in other industrial states that became democratic later. But that era has now passed, perhaps only temporarily but perhaps permanently. Its passing may have significant consequences for the form that politics will take in the near future, because simple left-right parties can no longer represent the key issues at stake.

One reason for contemporary multidimensionality is that there are many non economic issues. I wish to continue to focus, however, on issues that primarily concern interests rather than those that are matters of moral commitment that is not based on interests and that might even run against interests. One might suppose that these issues must therefore really just be matters that can be arrayed on a left-right dimension, just as general economic policy once was thought to fit such a model. But where knowledge and understanding of the impact of various policies — for example, on environmental issues, military expenditures, basic research, health care, and on through the list of most major economic policies of contemporary industrial states — is at issue, we cannot simply put all these issues into an additive function that goes from, say, more to less government expenditure. I want more spent on the environment, you want less on the environment and more on health care, another wants to cut both these if necessary to spend more on defense or education. If these were all marketable goods to be consumed at the individual level, then, subject to our resource constraints, I could buy what I want and you could buy what you want. But they are not, and we typically decide collectively on the levels of provision of these things, and we all get roughly the same levels.

If we could believe that there is something like a basic welfare or utility function that we all share, as George Stigler and Gary Becker (�) have argued, we might be able to array these issues, at least in principle. But even then, the ordinary citizen cannot perform such a trick and there will be remarkably strong disagreement on where to put our public money. Hence, although all these issues might be clearly seen as merely about interests, they define different dimensions because individuals have such different evaluations of them. For private consumption goods in modern economics, this would merely mean that we would have reason to exchange with each other to improve our welfare. I prefer your bit of something to my bit of something else and you have the opposite preferences, so we trade. This is the great simplifying move of modern market economics: We can each trade off bits of some things for bits of others, all to increase our own welfares.

This simplifying move fails for many public allocations. For public provisions of essentially collective goods we have to agree on single allocations of various things for everyone at once, and we cannot then trade with each other to come closer to our own preferred outcomes. We all get the same outcome. If we all started with the same endowments of private consumption goods, we could improve our lots by trading. That is not the position we are in for collective goods once these are provided. I cannot trade part of my share of environmental protection for part of your share of health care coverage. If we start with the same endowments of collective goods, we keep the same endowments. As a result, we each must view the collective allocations as occurring on de facto different dimensions, because I do not aggregate them in the same way you do. The aggregate of our collective allocations might seem splendid to you and miserable to me in comparison to plausible alternatives at the same total public cost.

Even though they commonly cannot affect their own welfare through their private actions without interaction effects from the actions of others, people surely have more control in general over their own welfare through the private than through the political sphere. They therefore have much greater incentive to understand their private concerns than to understand public concerns. This is an analog of the common claim in utilitarianism that we should typically focus our beneficent efforts on those close to us, because we can be surer of acting effectively for the good this way than if we try to act for the general welfare.[2] Ideally, we could overcome this problem of unequal incentives to a large extent if we could reduce collective goods to individual goods. This is not possible for collective bads, such as pollution, which are not deliberately produced but are, at best, external effects of desirable activities. And the reduction could not be efficient for many goods that are collectively provided by governments and other agencies.

It may also generally be harder to understand public concerns for such things as environmental protection than it was to understand at least the supposed implications of policies to enhance economic activity. For the short term of the present generation, workers might generally think it their interest to increase wages even at cost to investment, while owners and the relatively wealthy might think it their interest to increase investment even at cost to wages. But neither workers nor owners might have a clear sense of whether general environmental protections are in their interest, although they might readily conclude that protections that specifically burden their industry are not in their interest. Even workers exposed to carcinogens, such as benzene and the gases involved in the production of polyvinylchloride film (plastic wrap for food and other things), might think it their interest to keep their jobs with such exposure rather than to escape the exposure by enclosing various processes and, coincidentally, eliminating workers.

One might read recent elections in the United States, the United Kingdom, and Germany as responses to the fading of the traditional economic divide between left and right parties, between concern for wages and concern for profits, as leaders of both left and right parties begin to suppose that relatively free market devices are best for running contemporary economies. The focus in recent years has been on other issues, including the single issue focus of many groups on such issues as abortion and gun control in the United States, and the environment in Germany. Indeed, in Germany there is a specifically Green party, because environmental issues are not the natural concern of either the conservative Christian Democrats or the Socialists. These other issues are generally not conspicuously tied or related to the traditional divide over wages and profits. Hence, neither traditional party is able to capture them for its agenda to help trump the opposing major party. Traditional parties and voters alike now face a world in which multiple issues de facto represent multiple dimensions (see further, Hardin forthcoming).

The main left-right economic issue defined the two main parties in the United States and in many other democratic nations for much of this century. Other issues were many and varied, but they commonly did not dominate the central economic issue.

It is the fact that various contemporary issues, which themselves seem like matters of mere interest, are collectively determined or provided that makes them each an independent dimension for collective choice. Hence, so long as these diverse issues are politically important, traditional left-right party alignments may not fit actual political issues very well. Therefore, the more of these issues we can get off the collective agenda, the better for making collective choice coherent. With many of these issues on the collective agenda, we may increasingly witness the break-up of any major party focus and, de facto, sharpen the relevance of the problem of Arrow’s multiplicity of quasi-orthogonal dimensions of major issues. The response of candidates to such a change cannot be that of Downsian candidates. Instead it is likely to be to become unspecific and bland, without major policy positions other than those that can gain relatively general support. But of course, policies that gain general support cannot differentiate candidates.

Concluding Remarks on Liberty

We can amplify some of the discussion above through consideration of a particular set of goods that are collectively secured through government: political and personal liberties. The liberties I have are almost entirely in the context of my private life, not in the context of my public or political life. I may have the political liberty to vote and even to run for office. But most citizens do not typically have the liberty actually to make any difference to their own welfare through politics. I may indeed come to have substantially enhanced welfare from my political activities and I may even vote with the majority to achieve benefits for us. But if my vote is worthless, then the liberty to cast it is of little value either. Having the liberty to cast it is roughly as valuable as having the liberty to cast a vote on whether the sun will rise tomorrow.

In general, the ratio of liberty to restraint is greatest for those in a frontier context in which they need not be bound by any cooperative or coordinative arrangements. But that is not a very desirable state of affairs because it is likely to be impoverished, and liberty with poverty does not enable one to do much or to prosper well. To be much better off one must submit to substantial restraints of social order for mutual benefit. The anarchist’s liberty comes at a dreadfully high price. This is an instance of Brian Barry’s (1980) argument that, in democratic politics, it is better to be lucky than powerful. It is better to just happen to be with the majority than to have a little bit of influence over public policy. One does not then cause one’s preferred policies to be adopted, but one benefits from their adoption. Most of us who have liberty are merely lucky to have it; we did not bring it upon ourselves.

The liberty we get from democratic politics is, usually, the liberty we get from constitutional government and its protections. When democracy fails in the sense of producing an anti-constitutional regime, as it did in Germany in the 1930s, liberty may fail with it. An anti-constitutional regime can be democratic in the strongest sense of the term in that it can be wildly popular, as such regimes have been in many fascist and other nations historically. But it would be odd to suppose that the popular election of an anti-constitutional regime is itself a failure of liberty rather than of democracy.

The fact that we can readily recognize the infraction of a personal liberty gives reason to defend liberties through individual actions — that is, of necessity, through courts rather than through majoritarian devices. It is not merely that we need to fear a tyranny of the majority but also that we should doubt the capacity of a majority to act on its own behalf in general defense of liberties. The generalized logic of liberties might seldom impress itself upon citizens. The specific violation of my liberty, however, will immediately impress itself on me. Hence, although we have a genuinely collective interest in various liberties, we should want to have them enforced through individual initiative.

In this respect, liberties can be handled to some extent the way we might want other collective allocations to be handled: individually. That is to say that, ideally, we could reduce collectively provided goods to individually provided goods. This move is not possible for collective bads and it might be inefficient for many collective goods. As long as there is great demand for such goods and for the regulation of such bads, therefore, we can expect politics to be multidimensional and we can expect voters to be rationally ignorant of their own interests and of candidates’ positions on all the dimensions they face. Finally, we can also expect that they can no longer simply rely on traditional left-right parties to represent their interests on all these dimensions.

Problem of voter ignorance is compounded by the nature of our agency relationship to government. Our agents are our rulers. Przeworski et al. (Democracy Accountability, and Representation, editors’ overview), say that there are two big conclusions from their contributions First, citizens do not know enough to instruct or judge governments; and second, the institutional structure of government is fundamentally important and is not well matched by the electoral system. Our crucial problem therefore is how to design government in a way that its parts control each other and make the whole act in the best interest of the public.

References

Arrow, Kenneth J. [1951] 1963. Social Choice and Individual Values. New Haven, CT: Yale University Press, 2nd edition.

Barry, Brian. 1980. “Is It Better To Be Powerful or Lucky?” Political Studies (June and September 1980) 28: 183-194, 338-352.

Dahl, Robert A. 1961. Who Governs? New Haven, CT: Yale University Press.

Dewey, John. 1948. Reconstruction in Philosophy. Boston: Beacon Press.

Downs, Anthony. 1957. An Economic Theory of Democracy. New York: Harper.

Estrich, Susan. 1998. Getting Away with Murder: How Politics Is Destroying the Criminal Justice System. Cambridge, MA: Harvard University Press.

Fenno, Richard F. Jr. 1978. Home Style: House Members in Their Districts. Boston: Little, Brown.

Gompers, Samuel. 1905. “Discussion at Rochester, N. Y., on the Open Shop — ‘The Union Shop Is Right’ — It Naturally Follows Organization.” American Federationist 12 (April, no. 4): 221-223.

Govier, Trudy. 1997. Social Trust and Human Communities. Montreal: McGill-Queens University Press.

Hardin, Russell. 1982. Collective Action. Baltimore: Johns Hopkins University Press for Resources for the Future.

–––––. 1988. Morality within the Limits of Reason. Chicago: University of Chicago Press.

–––––. Forthcoming. “The Public Trust.” In Susan J. Pharr and Robert D. Putnam, eds., What's Troubling the Trilateral Democracies. Princeton, NJ: Princeton University Press.

Hume, David. �

Mill, John Stuart. 1965 [1848]. Principles of Political Economy, ed. John M. Robson. Toronto: University of Toronto Press.

Olson, Mancur, Jr. 1965. The Logic of Collective Action. Cambridge, MA: Harvard University Press.

Popkin, Samuel L. [1991] 1994. The Reasoning Voter: Communication and Persuasion in Presidential Campaigns. Chicago: University of Chicago Press, 2nd edition.

Schumpeter, Joseph A. [1942] 1950. Capitalism, Socialism and Democracy. New York: Harper, 3rd edition,

Singer, Marcus George. 1961. Generalization in Ethics. New York: Knopf.

Stigler, George, and Gary Becker. �. “De Gustibus non Disputandum.” �

Wittgenstein, Ludwig. 1969. On Certainty, ed. G. E. M. Anscombe and G. H. von Wright. Oxford: Basil Blackwell.