¹ Some political egalitarians would not limit the view's scope in this way, but it simplifies matters to consider this narrower view. If it is not correct, then the broader version couldn't be either.

² I describe an epistemic approach to democratic legitimacy called epistemic proceduralism in my essay “Beyond Fairness and Deliberation: The Epistemic Dimension of Democratic Authority,” in Deliberative Democracy: Essays in Reason and Politics, ed. Bohman, James and Rehg, William (Cambridge: MIT Press, 1997).Google Scholar I explain the pertinent idea of an independent standard there, and at greater length in “Making Truth Safe for Democracy,” in The Idea of Democracy, ed. Copp, David, Hampton, Jean, and Roemer, John (Cambridge: Cambridge University Press, 1993), 71–100.Google Scholar

³ My criticism of fair proceduralism in “Beyond Fairness and Deliberation” leaves the question of political egalitarianism open. Here I take it up directly.

⁴ On the distinction between justice and legitimacy in a liberal theory, see my essay “The Survival of Egalitarian Justice in John Rawls's Political Liberalism,” Journal of Political Philosophy 4, no. 1 (1996): 68–78.Google Scholar

⁵ Robert Dahl endorses political egalitarianism in Democracy and Its Critics (New Haven: Yale University Press, 1989), 109, 114–15.Google ScholarCohen, Joshua, in “Deliberation and Democratic Legitimacy,”Google Scholar reprinted in Bohman, and Rehg, , eds., Deliberative DemocracyGoogle Scholar, sketches an “ideal deliberative procedure” which “is meant to provide a model for institutions to mirror” (73). One feature is that “the participants are substantively equal in that the existing distribution of power and resources does not shape their chances to contribute to deliberation …” (74). Sunstein, Cass, in “Political Equality and Unintended Consequences,” Columbia Law Review 94 (1994): 1394CrossRefGoogle Scholar, says: “Disparities in wealth ought not lead to disparities in power over government.” Thomas Christiano writes: “Justice requires that individuals have political equality, that is, equal resources to influence decisions regarding the collective properties of society” (Christiano, , The Rule of the Many [Boulder, CO: Westview Press, 1999], 87Google Scholar). Harry Brighouse advocates “equal availability of political influence,” which “requires the insulation of the political process from [income and wealth] inequalities” (Brighouse, , “Egalitarianism and Equal Availability of Political Influence,” Journal of Political Philosophy 4, no. 2 [1996]: 120CrossRefGoogle Scholar). Jack Knight and James Johnson (“What Sort of Political Equality Does Deliberative Democracy Require?” in Bohman, and Rehg, , eds., Deliberative DemocracyGoogle Scholar) advocate “equal opportunity of access to political influence” (280), including “equality in the resources that any participant be allowed to employ in the deliberative process” (293). In addition, Rawls and Dworkin may be committed to versions of it; see notes 10 and 41, respectively. In criticizing political egalitarianism I do not expect to have fully refuted any of these authors, whose views differ in interesting ways from each other and from the simplified version of political egalitarianism that I discuss.

⁶ For quotations of several epistemic arguments used to disenfranchise the propertyless, blacks, and women, see Ortiz, Daniel, “The Democratic Paradox of Campaign Finance Reform,” Stanford Law Review 50 (02 1998): 906–9.CrossRefGoogle Scholar

⁷ This label is meant only to name a view of political equality. I leave aside the question of the relation between this view and a more general libertarianism.

⁸ The difference principle states that inequalities can be justified if they benefit even the least well-off. See Rawls, John, A Theory of Justice (Cambridge, MA: Harvard University Press, 1971), 302Google Scholar; and Rawls, , Political Liberalism (New York: Columbia University Press, 1993), 56.Google Scholar

⁹ I will assume that where there is formal equality there is also full compliance.

¹⁰ Rawls stops short of insisting on perfect equality, but mentions approximately equal value of political liberties (Political Liberalism, 358Google Scholar). His name for this is “fair,” not “equal,” value, and this may reflect the view that equality is not the point. The challenge, then, as I see it, is to find a salient standard between merely formal equality and strictly equal substantive influence. It is not clear whether Rawls would include the fair value requirement of the first principle among the requirements of legitimacy, which is generally a lower standard than justice in Rawls's theory. (I discuss this in “The Survival of Egalitarian Justice in John Rawls's Political Liberalism.”)

¹¹ I am not making any assumptions about the particular nature of the discussion, or its civility, etc. Some of the literature on deliberative democracy revolves around implausible standards of politeness and reciprocity that I do not want to commit to here. Moreover, the term “discussion” is not meant to exclude nondiscursive contributions to public debate, such as protests or political art. There are important nondiscursive components of any discussion, including public political discussion.

¹² See, e.g., Ackerman, Bruce, “Crediting the Voters,” The American Prospect 13 (Spring 1993): 71–80.Google Scholar I discuss Ackerman's view in more detail in Section V.

¹³ Rawls, , A Theory of Justice, 78.Google Scholar

¹⁴ See Brighouse, Harry, “Political Equality in Justice as Fairness,” Philosophical Studies 86 (1997): 166–67.CrossRefGoogle Scholar

¹⁵ By Pareto-superiority (political or not), I have in mind improvement in all positions in the distribution rather than effects on actual individuals. If two people switched places, so that one was worse off and the other better off, this would not yet count as a change at all from the perspective I want to consider. One distribution is Pareto-superior to the other if their graph lines flowing from lowest holdings to highest never cross, though they may touch (except in the case of strong Pareto-superiority).

¹⁶ One well-known competitive good structure is the “zero-sum game.” But that is a special case. Competitive goods need not have a “zero” or fixed sum.

¹⁷ I address this simplification below, in a subsection titled “Input as money.”

¹⁸ See the discussion below in Section V.

¹⁹ Of course, it would be unfair to formally give the two sides different time limits. But suppose both were allowed sixty minutes, but one side lacked leisure or strength enough to go beyond fifty minutes.

²⁰ For some purposes, it is useful to distinguish between this general instrumental value, and a specifically epistemic route to it, e.g., via reasoning, knowledge, understanding, etc. This distinction is not important here.

²¹ Rawls, , Political Liberalism, 137.Google Scholar

²² See my “Making Truth Safe.” As William Galston and Steven Wall pointed out to me, citizens who reject the public conception of justice that I hold proper voting to address are very unlikely to promote it through their votes (whatever the value of their political speech), and this could be agreed upon by all reasonable citizens. This is because reasonable citizens all accept the conception. Thus, there is some uncomfortable pressure on my theory to disenfranchise those who reject the conception, or to weight their votes more lightly for legitimate epistemic reasons. There are several lines of reply worth considering, but here I will mention only that if, as I believe, the public conception of justice will consist in rather abstract principles that do not have straightforward practical consequences without interpretation, there may be no method that is beyond reasonable objection for identifying those who reject the conception.

²³ There are complications. Certainly some bureaucratic or representative positions, e.g., entail greater power. The rule against invidious comparisons does not preclude hiring or electing the best-qualified applicants for such positions. But this would not be basic inequality so long as the position and its power and the hiring criteria are authorized, at least indirectly, by a legitimate democratic process that relies on no invidious comparisons.

²⁴ Rawls holds that the fair value of the political liberties “is essential in order to establish just legislation” (Political Liberalism, 330).Google Scholar

²⁵ Brighouse, , “Egalitarianism,” 123Google Scholar; Christiano, Thomas, “The Significance of Public Deliberation,”Google Scholar in Deliberative Democracy, ed. Bohman, and Rehg, , 256.Google Scholar

²⁶ Christiano, , “The Significance of Public Deliberation,” 341.Google Scholar

²⁷ Estlund, , “Beyond Fairness and Deliberation.”Google Scholar

²⁸ We might distinguish between fairness to participants and fairness to candidates or potential beneficiaries. I am discussing fairness to participants.

²⁹ Rawls, , A Theory of JusticeGoogle Scholar; Scanlon, Thomas, “Contractualism and Utilitarianism,” in Utilitarianism and Beyond, ed. Sen, Amartya and Williams, Bernard (Cambridge: Cambridge University Press, 1982)Google Scholar; Beitz, Charles, Political Equality (Princeton: Princeton University Press, 1989), esp. ch. 5.Google Scholar

³⁰ Verba, Sidney, Schlozman, Kay Lehman, and Brady, Henry E., Voice and Equality (Cambridge, MA: Harvard University Press, 1995).Google Scholar

³¹ Christiano (in “The Significance of Public Deliberation”) explores the epistemic basis of political egalitarianism in addition to fair proceduralism. “[E] quality in the process of democratic discussion … improves the quality of the outcomes of democratic decision making” (256). Since Christiano advocates equal access rather than equal actual input (253), it is not entirely clear how his approach would have epistemic advantages.

³² By a more equal distribution, I shall mean as measured by the so-called Gini coefficient (a measure of inequality sometimes used by economists). There are various alternative measures, some more appropriate than others for different purposes, but the Gini measure is simple and has no significant disadvantages that I know of for our purposes. The choice between various measures will not matter for my main points, and readers can safely proceed with only an intuitive idea of greater or lesser distributed equality. See Temkin, Larry S., Inequality (Oxford: Oxford University Press, 1993)Google Scholar, ch. 5, for a critical discussion of several alternative measures of inequality; see esp. p. 129f. for a discussion of the Gini coefficient.

³³ See the discussion below in the subsection entitled “Is more better?”

³⁴ Increased quantity at a constant level of Gini inequality (see note 32 above) may seem to guarantee political Pareto-improvement, but it does not. Here is one category of counterexample: Consider a distribution, call it Distribution 1, among ten people. Suppose the bottom person has 0 units of input and the top person has 1,000, and each of the middle eight people has 50. Gini inequality rounds to .58, and the total is 1,400. Now the level of inequality can be maintained even through a Pareto loss, as follows: reduce the middle eight to 48 each (the Pareto loss), but raise the bottom person to 40.8 and the top person to 1,200. This is Distribution 2. This also rounds to Gini = .58 (and the remaining difference can be completely expunged by precisely tinkering with the numbers, but I am keeping it simple here). But the total has gone up to 1,424.8. Distribution 2 has a Gini-constant increased quantity, but is not Pareto-superior to Distribution 1.

³⁵ One very simple way to model the decreasing epistemic value of input is to suppose that, so long as the level of inequality is kept constant, a given extra unit of input produces an epistemic increase that is some constant fraction of the gap between the ex ante epistemic value and infallibility. So, for example, each extra unit of input (suppose this is some amount of money spent) might get you 10 percent of the remaining way toward infallibility. The next unit gets you 10 percent of the remaining way, which is less progress than the first unit. Any value for the marginal unit (e.g., $1, $1,000), and any constant setting of this fraction of the gap (1 percent, 20 percent), will allow increasing quantities to converge on infallibility. I will proceed on the assumption that the marginal epistemic value of input at a given level of inequality has this structure. This says nothing about how fast an extra unit of input increases epistemic value, since this can be set very low or very high consistent with my assumption.

Decreasing Marginal Epistemic Value of Input: Assuming inequality stays the same, then for any quantity, and any epistemic value, there is some constant fraction F, such that a unit of extra input moves the epistemic value of the scheme forward by removing that fraction F of the gap to infallibility.

This assumption is one way of representing the intuitive thought that quantity improves epistemic value, other things being equal. I offer no argument for placing no limit on this improvement short of infallibility, though this seems the simpler position in the absence of any reason to believe there are more severe limits. A weaker claim, placing such limits, would probably be sufficient for practical purposes, but it is difficult to know which weaker claim. With this in place, we are entitled to our assumption that quantity compensates inequality.

One odd feature of this model is that it is oblivious to the ex ante level of input, but notices only the ex ante epistemic value. In one scenario, then, the total input might be $1 million with an epistemic value of .5. Another scenario might find total input at $100 and an epistemic value of .5. If $1,000 moves a distribution 10 percent of the way toward infallibility, it would do this in both cases. This might seem to ignore the apparently greater epistemic value of each unit of input in the latter scenario, which has the same epistemic value as the former but with less input. I leave these complexities aside, but perhaps a more refined model should take account of them.

³⁶ See note 35.

³⁷ If the left-out group is very small, we could increase inequality only a little, increase quantity a lot, and still fail to share downward. Is it plausible to say this does more harm than good? What if only one person out of a million is left out of the increase? Here we should allow that the quantity probably outweighs the inequality. We can avoid this problem if we limit our purview to changes that affect groups that make up a substantial fraction of the whole. Then leaving one group out of an inequality-increasing input increase will raise inequality significantly so long as the quantity is increased significantly. Let us also assume that we mean no more here by social groups than groupings according to amount of political input. So the membership of, for example, the lowest group can change. If two sets of people were to end up with completely exchanged levels of input (set A now has as much input as set B had, and vice versa), the distribution of input would not have changed for present purposes.

³⁸ Dworkin, Ronald, “The Curse of American Politics,” New York Review of Books, 10 17, 1996, 21.Google Scholar

³⁹ See, for example, Ansolabehere, Stephen and Iyengar, Shanto, Going Negative (New York: Free Press, 1997)Google Scholar: “As we have shown in several chapters of this book, television actually fosters the democratic ideals of an informed and reasoning electorate” (145). Their worries about political advertising lie elsewhere.

⁴⁰ Ansolabehere and Iyengar also argue (in Going Negative) that negative campaigning drives voters away from the polls. This might itself have epistemic disvalue to be weighed against the value of the information provided. But this is not a point specially about high levels of campaign spending, unless for some reason high levels of spending increase negativity. And even then, in order to suppose that limits could be imposed without epistemic loss, it would need to be shown that the bad epistemic effects of the marginal speech outweigh the good.

⁴¹ Dworkin is not explicit about how much equality of influence should be sought through campaign finance regulations or other means. But he does say: “Each citizen must have a fair and reasonably equal opportunity … to command attention for his own views” (Dworkin, , “The Curse of American Politics,” 23).Google Scholar

⁴² I say “roughly” because he concentrates on primary goods, which are not limited to what is usually understood as economic matters.

⁴³ The similarities between these two incentive arguments for inequality are sufficient, however, to force us to answer an important objection advanced by G. A. Cohen. Cohen argues that it is not clear that a proper citizen in a fully just society could have the motives that the incentive argument assumes. He wonders what would justify a citizen's holding out for more pay than others when he or she is capable of doing the work without it. Cohen has developed this criticism in a series of essays, including “Incentives, Inequality, and Community,” in The Tanner Lectures on Human Values, vol. 13, ed. Peterson, G. (Salt Lake City: University of Utah Press, 1992), 263–329Google Scholar; “The Pareto Argument for Inequality,” in Contemporary Political and Social Philosophy, ed. Paul, Ellen Frankel, Miller, Fred D. Jr., and Paul, Jeffrey (Cambridge: Cambridge University Press, 1995), 160–85Google Scholar; and “Where the Action Is: On the Site of Distributive Justice,” Philosophy and Public Affairs 26, no. 1 (Winter 1997): 3–30.Google Scholar I have defended Rawlsian inequality against Cohen's arguments in “Liberalism, Equality, and Fraternity in Cohen's Critique of Rawls,” Journal of Political Philosophy 6, no. 1 (03 1998): 99–112Google Scholar, and those arguments apply fairly directly here as well.

⁴⁴ For example, suppose Ackerman's Patriot plan (discussed below, in Section V) is already in effect at the maximum equal level.

⁴⁵ The name “Progressive Vouchers” might connote three relevant things: that the voucher plan promotes quantity of input in a politically Pareto-superior way; that it promotes quality of decisions by independent standards of, e.g., justice; and that it involves progressive rates for marginal vouchers.

⁴⁶ The Progressive Voucher plan suggests that it is possible to epistemically improve upon an equal distribution of actual input. However, political egalitarianism, the egalitarian alternative we are considering, advocates an equal distribution of availability of, or opportunity for, input, not an equal distribution of actual input. Even if Progressive Vouchers can epistemically beat E-max, which involves equal input, can it beat the pattern of actual input that would emerge under equal availability of input?

The epistemic value of democracy under political egalitarianism depends upon the level and distribution of actual input, even though what it seeks to equalize is availability of input. Since Progressive Vouchers can beat the epistemic value of E-max, the highest level of equal actual input, then it would suffice to show that E-max epistemically beats actual patterns of input under political egalitarianism. This is easily shown, and derives from the fact that a scheme of equally available input has no tendency to allow a greater quantity of input than would be present under E-max. In that case, since the distribution of this lower or equal quantity will be a less equal distribution than E-max, which is exactly equal, then by the principle of Epistemic Value of Equality, the epistemic value under political egalitarianism will be less than that under E-max. Therefore, if, as I claim, a Progressive Voucher program can epistemically beat E-max, it can also beat poiitical egalitarianism.

⁴⁷ This grants something congenial to opponents of my thesis. If more bought vouchers, it would simply mean that my conclusion would follow more easily.

⁴⁸ As Ackerman points out, vouchers can be spent with no opportunity cost, unlike ordinary contributions of money. This fact should vastly increase participation. Still, full spending of the vouchers is an unrealistic assumption made to simplify the model.

⁴⁹ This $5,128,560 figure is arrived at by adding to the original $1 million (from E-max) the proceeds from the sale of vouchers. Two thousand voters pay at each of the following five levels: $50 for one voucher, $137.50 ($50 plus $87.50) for two vouchers, $290.63 for three vouchers, $558.60 for four vouchers, and $1,027.55 for five vouchers. Rounding to the nearest dollar, this yields $4,128,560, which, added to the $1 million from E-max, gives us a total of $5,128,560.

⁵⁰ Only about 8 percent of eligible voters contribute any money to political campaigns. See Warren E. Miller, and the National Election Studies, American National Election Studies Cumulative Data File, 2952–1992 (computer file), 6th release (Ann Arbor, MI: University of Michigan, Center for Political Studies [producer], 1994; Ann Arbor, MI: Inter-University Consortium for Political and Social Research [distributor], 1991).

⁵¹ It might seem useful to know the Gini index for any such set of data. But no single Gini level can be assumed to be epistemically too much or too little. It follows from the Compensation of Quantity for Inequality assumption that higher levels of Gini inequality are epistemically acceptable at sufficiently higher quantities.

⁵² Ackerman, , “Crediting the Voters.”Google Scholar

⁵³ Ibid., 72.

⁵⁴ Note the political problem of the wealthier being required by law to transfer resources to the less wealthy in order to subsidize political activity that may well be antithetical to their interests. But this may be just, and thus so I will not harp on the fact that my Progressive Voucher plan would avoid this political problem.

⁵⁵ Buckley v. Valeo, 424 U.S. 1, 48–49 (1976).Google Scholar

⁵⁶ Mill, John Stuart, Representative Government (New York: E. P. Dutton and Co., 1950), ch. 8.Google Scholar

⁵⁷ Political egalitarianism apparently cannot account for this kind of inequality.

⁵⁸ For a more difficult case, suppose it could be established beyond reasonable dispute that the quality of political decisions would be enhanced by selecting a small number of voters from the millions who register, and depriving all others of voting privileges. I do not think this has much plausibility, but suppose it were true. Giving only a few adult citizens the right to vote must be reckoned a version of formal political inequality. Now if the voting citizens were selected according to whether they were college-educated, or Christians, or pet-owners, and if this were done on the hypothesis that members of these groups were likely to make better decisions, or more fully deserved the power to vote than others, then the scheme would be utterly illegitimate. But if instead they were chosen randomly, and this still were reliably shown to have advantages for the quality of political decisions, it is no longer clear that the formal political inequality is objectionable. In truth, such a scheme would probably do more epistemic harm than good; but if it had overriding epistemic advantages, implementing it would not express disrespect for anyone.