Epistemic Critiques of Democracy

With fascism and communism as its most salient opponents, electoral democracy came to enjoy near-universal support in the twentieth century—at least among Western cultural elites. Even as scholars accumulated staggering evidence of the political ignorance of ordinary voters (Campbell et al. Reference Campbell, Converse, Miller and Stokes1960; Zaller Reference Zaller1992), most nonetheless embraced Winston Churchill's view of democracy as the “worst form of government, except for all the others that have been tried.”

Recently, however, some have begun to venture more full-throated criticisms. Jason Brennan argues, for instance, that if people have a right to competent government, we should not leave decisions in the hands of the “ignorant, irrational, misinformed nationalists” of typical electorates (Reference Brennan2016, 23). He then outlines a number of potential alternatives to one-person-one-vote, including restricted suffrage and plural or weighted voting. Given democracy's serious and demonstrable flaws, he argues, we have a responsibility to give such alternatives a try. Daniel A. Bell (Reference Bell2015) raises similar concerns from a rather different perspective, defending a quasi-Confucian political meritocracy modeled on contemporary Singapore and China, which eliminates competitive elections altogether. Such proposals to allocate political power on the basis of knowledge or merit, rather than giving it freely to everyone, are sometimes called epistocracies (Estlund Reference Estlund2008).

Epistemic skepticism of popular rule has an unflattering history, and it is often dismissed as obsolete. Yet electoral democracy faces troubling practical challenges in the twenty-first century (Dresden and Howard Reference Dresden and Howard2016), and serious normative worries have come from diverse intellectual quarters. Libertarians lament widespread ignorance about economic policy (Caplan Reference Caplan2007), for instance, and egalitarians suspect popular commitment to fundamental rights (Dworkin Reference Dworkin1996). Environmentalists bemoan the impact of short-sighted democratic choices on natural systems (Humphrey Reference Humphrey2007), while radicals doubt the ability of electoral democracies to overcome systemic pathologies such as bourgeois ideology, patriarchy, white supremacy, and settler colonialism (Coulthard Reference Coulthard2014; Davis Reference Davis2012; Gramsci Reference Gramsci, Hoare and Smith1971; Malcolm X Reference X1964).

Democrats typically rely on one of two strategies for defending competitive elections with universal suffrage. First, many emphasize their intrinsic value, maintaining that everyone deserves a say over decisions that bind them. Others prefer to stress the instrumental benefits of democratic institutions, observing that modern democratic governments have most reliably promoted peace and prosperity (Sen Reference Sen1999). Both of these common accounts, however, fail to adequately address recent epistemic critiques.

As Brennan and Bell readily admit, for one, highlighting the extent of voter ignorance will rarely sway those who take the intrinsic value of democracy as a foundational normative premise. Yet this premise is neither self-evident nor universally shared.Footnote ¹ If ceding a largely symbolic form of political equality like universal suffrage would really yield dramatic improvements in social, economic, and environmental outcomes, this tradeoff would understandably appeal to many observers around the world. This hardly constitutes a thorough rebuttal of arguments for democracy's intrinsic value, of course, but given that these arguments are both reasonably contestable in theory and widely disputed in practice, democrats would be unwise to ignore the domain of instrumental value altogether.Footnote ² In what follows, therefore, I set aside intrinsic concerns.

Similarly, epistemic critics happily acknowledge electoral democracy's relative instrumental success—especially in the twentieth century (Bell Reference Bell2015, 7; Brennan Reference Brennan2016, 8, 195). Yet it hardly constitutes a guarantee of peace, prosperity, and liberal rights (Levitsky and Way Reference Levitsky and Way2010); much less a thorough “democratization” of social and economic life (Crouch Reference Crouch2004; Tilly Reference Tilly2007). Meanwhile, democracies have also done awful things—especially to noncitizens (Bell Reference Bell2015, 46–47). Isn't it possible, critics ask, that some alternative might perform even better?

Ultimately, I argue, the answer is no: no epistocratic political institutions offer reliable substantive advantages over electoral democracy, all things considered. Especially given pervasive findings of voter ignorance, however, epistemic critics are right to be dissatisfied with the reasoning offered by prevailing accounts. Despite an almost axiomatic faith in democracy, democratic theorists still lack a compelling instrumental explanation of why competitive elections and universal suffrage should not be abandoned in favor of alternative institutions that would maintain attractive features of liberal government while filtering out the ignorance, irrationality, and bigotry of ordinary citizens.Footnote ³ As a result, existing instrumental accounts remain perpetually vulnerable to challenges from novel forms of non-democracy, like those proposed by Brennan and Bell, which are said to transcend the limitations of earlier forms.

Such challenges can be understood as versions of the perennial “benevolent dictator” objection—that is, that if rule by an intelligent and well-intentioned monarch or aristocratic elite could achieve better results than rule by the people, instrumentalists should prefer it to democracy.Footnote ⁴ Many democrats are troubled by this objection, which has often spurred the development of intrinsic accounts of democracy's value (Beitz Reference Beitz1989, 98; Christiano Reference Christiano1996, 16–7, 56; Kolodny Reference Kolodny2014, 202). If instrumentalism yields only a contingent, defeasible commitment to democratic institutions, they conclude, democrats had better rely on the intrinsic value of political equality and self-government. Meanwhile, many instrumentalists embrace the hypothetical prospect of attractive alternatives to democracy. Until recently, however, few have taken this prospect seriously.

We no longer have this luxury. Given the urgent practical and philosophical challenges facing electoral democracy, we cannot be content that it is less awful than the other systems we have tried. We must also explain to challengers like Brennan and Bell—and, more importantly, to reformers and revolutionaries around the world—why we should refrain from trying anything else.

Epistemic Defenses of Democracy

That is why the “epistemic” account of democracy gaining popularity among political philosophers appears so promising (Anderson Reference Anderson2007; Bohman Reference Bohman2006; Estlund Reference Estlund2008; Landemore Reference Landemore2012; Misak Reference Misak2008; Ober Reference Ober2008). Unlike many intrinsic accounts, epistemic views acknowledge that the quality of the outcomes produced by democratic procedures is central to their justification. Unlike their instrumentalist rivals, however, epistemic accounts promise a coherent explanation of why we should expect democratic procedures to produce better outcomes than non-democratic alternatives: only democratic procedures, they claim, can harness the collective wisdom of the people. As a whole, that is, the demos possesses greater wisdom than the wisest philosopher-king or council of experts, and so democracy—a system in which the demos is sovereign—will outperform even the most benevolent of monarchies or aristocracies. Compared to other instrumentalist approaches, then, epistemic accounts stake out a far more robust commitment to democratic procedures. The connection between democracy and superior outcomes is not just a historical accident, but a deep and abiding principle of social organization.

Content neutrality is key here. Most instrumentalist accounts emphasize the connection between democratic procedures and specific substantive outcomes—that is, maintaining liberal rights or avoiding famine and war—which inevitably invites “benevolent dictator” objections. On epistemic accounts, by contrast, democracy is more like the scientific method: we can trust it to reach political truths, in the long run, without knowing those truths in advance (Gaus Reference Gaus2011). As Sean Ingham summarizes, epistemic democrats aim to show that “democratic institutions have a tendency to produce reasonable outcomes [. . .] without presupposing any narrow, controversial view of what the outcomes of democratic procedures should be, much as a good justification of a particular scientific research design does not presuppose the hypothesis that the research aims to test” (Ingham Reference Ingham2013, 136). Rather than establishing their empirical propensity to produce particular results, therefore, epistemic democrats emphasize structural features of democratic procedures; turning to abstract arguments, analogical reasoning, and—most centrally—formal proofs of collective competence (Schwartzberg Reference Schwartzberg2015).

The simplest and most prevalent of these proofs—once called the “jewel in the crown of epistemic democrats” (List and Goodin Reference List and Goodin2001, 283)—is the Condorcet Jury Theorem (CJT). This theorem shows, roughly, that if many people independently make a binary decision that has a correct answer, and each has a greater-than-random chance of being correct, a majority vote is extremely likely to deliver the correct answer. Following its rediscovery in the late twentieth century, early epistemic democrats eagerly applied the theorem to democratic politics (Estlund Reference Estlund1994; Grofman and Feld Reference Grofman and Feld1988).

In the years since this initial enthusiasm, however, the theorem's relevance for democratic theory has repeatedly been challenged (Anderson Reference Anderson2008; Christiano Reference Christiano1996, 33–4; Dietrich Reference Dietrich2008; Gaus Reference Gaus1997), and even many former advocates have now conceded its limited usefulness (Estlund Reference Estlund2008, 223–36). The requirement that individual decisions be independent of one another, for instance, is incompatible with the deliberation and communication that many epistemic democrats understand as crucial to the superior performance of democratic procedures. In assuming a discrete number of choices, moreover, the CJT also ignores problems of agenda control (Fuerstein Reference Fuerstein2008). Perhaps the most devastating objection to the CJT's practical relevance, however, is simply that it proves too much (Ingham Reference Ingham2013). If we suppose the theorem does apply to actual democratic elections, we must implausibly conclude that they almost never deliver the “incorrect” result. Without this supposition, meanwhile, the theorem provides existing institutions with no defense against epistemic critiques.

As a result, contemporary epistemic democrats have typically looked elsewhere for proof of collective competence. Perhaps most prominently, Hélène Landemore has recently championed a theorem known as Diversity Trumps Ability (Landemore Reference Landemore2012; Landemore and Elster Reference Landemore and Elster2012). According to simulations run by Hong and Page (Reference Hong and Page2004), a randomly selected group of problem-solvers will often outperform a group of the best individual problem-solvers, due to the beneficial effects of cognitive diversity. Since this theorem allows for deliberation, unlike the CJT—and furthermore does not assume a discrete number of choices—it is understandably attractive to epistemic democrats looking for a more realistic alternative.

Unfortunately, however, its validity also depends on highly restrictive conditions that cannot be assumed to hold in real political circumstances (Brennan Reference Brennan2016, 180–94; Thompson Reference Thompson2014; Van Hees Reference Van Hees2007). Perhaps most importantly, it assumes cognitive diversity without value diversity, and once the latter is introduced, the theorem fails to show what Landemore claims (Ancell Reference Ancell2017). Moreover, she does not sufficiently address the possibility that epistocratic institutions could easily co-opt this virtue. Indeed, Bell's model for a modernized Confucian political meritocracy foregrounds diversity in the selection criteria for leaders as well as active consultation of diverse bodies of citizens—in both cases building on existing Chinese Communist Party practices (Bell Reference Bell2015, 190).

There are, of course, a range of other analogies and formal results that have been used to demonstrate the epistemic competence of democratic procedures, many of which rely on the value of cognitive, intellectual, or perspectival diversity.Footnote ⁵ We may safely set them aside, however, given the objection just raised. After all, any epistemic advantages thought to be unique to democratic institutions could conceivably be co-opted by a sufficiently well-designed epistocracy. Even if the multitude possesses some special wisdom that no team of experts could match, that is, competitive elections with universal suffrage are not obviously the most efficient way of harnessing it, and could conceivably be improved upon by clever experts.

Extant epistemic approaches, it seems, have failed to deliver on their promise to provide an instrumental defense of electoral democracy that is robust to challenges from epistocracy and other “benevolent dictator” objections. As the longtime epistemic democrat David Estlund admits in his recent work, it is no use denying that “there are subsets of citizens that are wiser than the group as a whole” (Estlund Reference Estlund2008, 40). The problem with epistocracy, rather, lies with deciding which subset to empower. Yet Estlund squanders this insight, in my view, by interpreting the problem in the terms of public justification, arguing that no epistocratic ruling class could be chosen that would be “justifiable in terms acceptable to all qualified points of view” (Estlund Reference Estlund2008, 41). As Landemore (Reference Landemore2012, 52) laments, his account thus resorts once again to intrinsic concerns, relying on epistemic considerations only to show that democracy outperforms random selection of outcomes. Though the formal proofs she favors are unconvincing, therefore, she is right to insist that a robust, fully instrumental defense of democracy is possible.

Given the serious practical dangers of according political power on the basis of epistemic merit, I argue, we can readily meet epistemic critics of democracy on their own instrumentalist turf. To do so, however, we must reject the “epistemic frame” long preferred by democracy's critics, which proposes to compare democracy and epistocracy in terms of the wisdom of those supposed to have ultimate sovereignty—that is, the demos and the knowers, respectively. Given that all regimes are in reality composed of complex, pluralistic power structures, the notion of “ultimate” sovereignty only obfuscates (see Pettit Reference Pettit2013, 12–5, 220–9). As I elaborate below, democracy and epistocracy are not diametrically opposed: in fact, successful democracies rely on certain epistocratic institutions, while the proposals of Brennan and Bell avowedly include democratic elements. Instead of evaluating the wisdom of the “sovereign” in each regime, therefore, I urge sustained attention to the incentives and opportunities facing those whose power is, in relative terms, most concentrated.

Political Meritocracy and the Necessity of Competition for Power

We may begin with Bell's (Reference Bell2015) arguments for political meritocracy. Despite what many assume from his book's title (i.e., The China Model), Bell does not claim that the contemporary Chinese government fulfills the meritocratic ideal, nor even that it outperforms existing electoral democracies (19). Instead, the version of political meritocracy he proposes as a “model” will have eliminated the political repression and corruption characterizing the current regime by implementing independent checks on centralized power (116, 124, 150), a freer press (134, 174), intraparty competition (138), more democratic participation in local affairs (169, 189–91), improved Confucian moral education (124), greater economic equality (132), and diversified methods of meritocratic selection (130, 135, 193–4), among other reforms. Top-level leaders in Bell's imagined political meritocracy, in other words, would be prevented from abusing their power by a number of norms and institutions that many will recognize as liberal. Citing precedents from East Asian history and philosophy, however, Bell rightly observes that many of these devices are not exclusively Western or liberal in origin. Notably, this includes meritocracy itself, which has origins in both Western and Confucian traditions as a mechanism for constraining the power of the nobility and other privileged elites (65–6).

Bell's emphasis on independent checks on centralized power is not surprising. The evils of unconstrained authoritarian rule—or, more classically, tyranny—have long been obvious to all serious observers. As the power of modern centralized states has continued to expand, moreover, effective constraints have only become more important (El Amine Reference El Amine2016). Even if unconstrained rulers are initially benevolent, the political institutions they create can be used by less benevolent successors (or usurpers) in tremendously harmful ways. Meanwhile, opposition leaders have no feasible non-violent path to power and are thus more likely to engage in violent resistance; inducing coercive repression and/or civil war (Przeworski Reference Przeworski, Shapiro and Hacker-Cordon1999). Finally, the absence of independent constraints is associated with “extractive” institutions and reduced long-run economic growth (Acemoglu and Robinson Reference Acemoglu and Robinson2012; Fukuyama Reference Fukuyama2014; North, Wallis, and Weingast Reference North, Wallis and Weingast2009).

Like most historical advocates of rule by the few, therefore, contemporary epistocrats happily concede that rule by unconstrained, extractive elites is worse than rule by the many. They insist, however, that non-democracies are not necessarily tyrannical in this way. As Bell points out, for instance, the leaders of contemporary China and Singapore are already constrained by a range of norms and institutions, and his reforms would constrain them further. Moreover, effective constraints predated competitive elections in most early liberal states, while implementing elections without supporting reforms in certain contemporary non-democracies could be counterproductive. As a result, I readily concede that elections need not always be our first practical priority.

What I maintain, however, is that no large-scale, long-term political vision is plausible without open political competition. Within democracies, suspending elections may be necessary when the electoral framework itself faces an existential threat, but this measure must always be temporary (Kirshner Reference Kirshner2014). Meanwhile, non-democracies should generally strive to achieve stable electoral institutions as quickly as realistically possible. This is not because elections are intrinsically valuable, nor because they always yield superior outcomes. Instead, the primary reason political meritocracy does not present an attractive alternative to electoral democracy is that it is far more likely to lead to unconstrained authoritarianism or tyranny. Though constraints on centralized power can also be eroded in a system with electoral competition, the magnitude of this risk is much greater under political meritocracy—even given highly advantageous conditions—than it is under any ordinary consolidated democracy. The key difference is that meritocratic leaders have much more effective tools for entrenching their power.

In any political system, many things influence whether incumbents and their factional allies retain power. In both meritocracies and democracies, for instance, the approval of a broad class of economic and cultural elites is an important contributing factor. In both systems, similarly, massive popular discontent bodes poorly for incumbents. In any political meritocracy, however, the incumbents themselves—as those atop the centralized meritocratic hierarchy—have far greater capacity to affect the outcome. Given the significant discretion afforded by the unavoidably ambiguous standard of political “merit,” incumbents can gradually shift the balance of power in their favor, appointing cadres at every level who will support their factional interests, and replacing those who will not.Footnote ⁶ Compared with elected democratic politicians, therefore, they have far more latitude in shaping the structural conditions supporting their own political power, and (consequently) much greater capacity to entrench their rule.

Of course, democratic incumbents also use their control over the state apparatus to increase their chances of retaining power—reliably employing techniques such as gerrymandering, for instance, where available. They may also attempt to capture independent agencies by appointing factional allies or undermine the economic and organizational foundations of opposition parties (see Levitsky and Ziblatt Reference Levitsky and Ziblatt2018, 72–96). Yet there are at least three crucial factors restraining such efforts at entrenchment: the uncertainty of electoral outcomes, the organizational strength of the opposition, and independent commitment to the institutional framework of liberal democracy.

First, incumbents know that despite their efforts to stack the deck, their opponents may yet come to power and retaliate. Given sufficient uncertainty about the results of subsequent elections, an equilibrium in which all incumbents respect certain limits to such self-serving strategies is beneficial for all parties (Przeworski Reference Przeworski1991, Reference Przeworski2005; Shapiro Reference Shapiro2016, 49–50).Footnote ⁷ Attempts to exceed these limits may then be foiled by independent power centers within the state—perhaps through conflict between executive and legislative branches, as Madison famously envisioned, or perhaps through other mechanisms like judicial and bureaucratic independence, federalism, and institutionalized protections for opposition parties. As Levinson and Pildes (Reference Levinson and Pildes2006) insist, the key principle is “separation of parties”—the distribution of state power among different factions—rather than the separation of powers as such (which is perfectly compatible with unified partisan control). Second and more generally, then, rival parties will be able to coordinate opposition to incumbent entrenchment—at the limit including armed rebellion. Indeed, the organizational strength of opposition parties is one of the key conditions for achieving and consolidating democracy (Levitsky and Way Reference Levitsky and Way2010, 68–70).

Finally, these incentives will be supplemented by some degree of genuinely principled commitment to liberal democracy and the rule of law—especially among those with strong professional socialization such as lawyers and journalists. This commitment will bolster the existing motivation of opposition officials to frustrate incumbent entrenchment, for instance, as well as any existing preferences for stable equilibrium among the incumbents’ allies. Indeed, mechanisms grounded solely in political incentives would undoubtedly be less reliable in the absence of genuine principled commitments, and their significance should not be discounted. At the same time, polarization reliably weakens their motivational force relative to immediate substantive concerns (Svolik, n.d.). Perhaps even more troublingly, it can be difficult to disentangle principled from expedient motives in the first place (Bagg Reference Bagg2018a, Reference Bagg2018b; Lodge and Taber Reference Lodge and Charles2013). Given these vulnerabilities, it would be unwise to rely entirely on the production of principled commitments through education and socialization to maintain the integrity of notoriously fragile constraints on centralized power.

That, however, is precisely what Bell's political meritocracy would do. Even if the independent power centers he proposes were initially effective, their continued independence would not be supported by electoral uncertainty and organized opposition, and could be relatively easily co-opted by top-level leaders disposed to do so. As Bell himself concedes (Reference Bell2015, 123–4), the stability of constraints on power in a meritocracy ultimately relies on the forbearance of those at the top of the centralized hierarchy, which he proposes to secure through improved Confucian moral education.

As I have emphasized, I do not deny the efficacy of such principled commitments. Yet, as Madison famously insisted, “enlightened statesmen” should not be relied upon to sustain crucial constraints on power in perpetuity.Footnote ⁸ Without reinforcement from strong independent incentives—such as those provided by institutionalized uncertainty and the separation of parties—principled commitment to maintaining independent power centers will eventually succumb to the siren song of political expediency. Institutions that foster legitimate opposition and open competition for power thus render the capture of the state by tyrannical, extractive elites less likely than institutions that effectively allow incumbents to select their own replacements.Footnote ⁹

In my view, the severity of this danger is more than enough reason to abandon the idea of political meritocracy. It is also worth noting, however, that the increased risk of tyrannical rule is not offset by great prospective rewards. Even when incumbents do not actively consolidate power, they will still be subject to more mundane forms of capture and corruption, undermining meritocracy's claim to possess significant advantages over democracy.

Consider the strict conditions enabling the progress of natural science—perhaps the most obviously successful meritocracy in existence. Especially within Kuhnian “normal science,” standards of merit are relatively uncontroversial. Where disagreement exists, it is usually resolved methodically within structures of open competition, which are themselves supported by a decentralized disciplinary structure. Finally, few scientific questions have significant consequences for the distribution of social power, meaning that scientific institutions are rarely targeted for capture by outside interests.

In politics, by contrast, the distributive stakes are far higher. Political meritocracies are thus, predictably, the target of intense competition for capture among factions seeking to advance their interests. Given the inevitable limits to open competition under a centralized hierarchy, these factions are forced to displace the weight of their political disagreements onto the definition and application of meritocratic standards. Yet these standards are far more ambiguous than those of science—thanks partly to wider disagreement on the ultimate ends to be served (Jenco Reference Jenco2016)—and consequently less capable of withstanding such strain. Though scientific meritocracies have enabled remarkable progress through the methodical resolution of disagreements, therefore, this is only in virtue of several features that cannot be replicated by a centralized political meritocracy.Footnote ¹⁰

Despite these vastly more favorable conditions, of course, even science is hardly a perfect meritocracy. From its origins to the present, systematic biases and exclusions have profoundly affected its character (Longino Reference Longino2013; Marks Reference Marks2009). In other realms where it is often invoked, meanwhile, the ideal of meritocracy is even further from realization, and is often simply used to legitimize unjust inequalities (McNamee and Miller Reference McNamee and Miller2013). These challenges do not render the concept of meritocracy incoherent or inherently reactionary—on the contrary, meritocracy is a necessary feature of any modern political system, including electoral democracy. Yet no meritocracy is immune to bias, capture, or corruption (Fricker Reference Fricker2009; Medina Reference Medina2012), and given the intense strains they must bear, such defects will be especially significant in political meritocracies.

Acknowledging the extensive corruption of China's existing political structure, Bell often draws on the success of political meritocracy in Singapore and Hong Kong to demonstrate that these flaws might be remedied (Reference Bell2015, 31–3, 113, 117). According to many other scholars of the region, however, the opposite trend is more likely. Political meritocracy in Singapore and Hong Kong is already decaying, they observe—and will continue to do so, absent democratization—while China's is unlikely to become less corrupt (Hui Reference Hui2015, Reference Hui2016; Tan Reference Tan2008). Ironically, they conclude, meritocracies function best under democratic political conditions.

That said, we need not deny that political meritocracies can govern well—perhaps even outperforming electoral democracies in certain cases. Rather, my claim here is that political meritocracies are unlikely to enjoy either systematic or substantial advantages over comparably situated electoral democracies in the long run—especially those that successfully integrate healthy meritocratic institutions at other levels. Given this, the increased risk of authoritarian consolidation easily counts as a decisive argument against political meritocracy.

Franchise Qualifications and the Necessity of Universal Suffrage

We turn next to the franchise qualifications proposed by Brennan (Reference Brennan2016, 204–30; see also Caplan Reference Caplan2007, 197–8). He acknowledges the injustice of historical restrictions grounded on morally irrelevant factors such as race, sex, or possession of property. Given the epistemic flaws of democracy, however, he asks: why not conditionalize suffrage upon morally relevant epistemic qualifications? Prospective drivers must pass a driving test, and prospective doctors must attend medical school. Why not issue voting licenses in a similar way, through competence testing or educational requirements?

As above, my response to these suggestions is thoroughly pragmatic. While conceding the conceptual possibility of unobjectionable epistemic qualifications, I highlight the practical risks of giving political leaders the power to design them. Though not quite as dangerous as the ability to appoint successors directly, the ability to implement epistemic qualifications would also enable incumbents to entrench their rule—and the hypothesized benefits here are even less likely to materialize. Thus, Brennan's epistocracy ought to remain off the table as well.

Of course, democracies already enable incumbents to manipulate the pool of eligible voters in various ways. Some restrictions, like those concerning age and citizenship, are likely inevitable. Others, however—including voter identification requirements, felon disenfranchisement, and the gerrymandering of single-member districts—are routinely used to entrench the power of incumbents by tilting the electoral odds in their favor (Levitsky and Ziblatt Reference Levitsky and Ziblatt2018, 183–6, 208–11). Such policies are typically justified on other grounds, and some may be valuable on balance, yet in each case, this feature gives us reason to be suspicious (Issacharoff and Pildes Reference Issacharoff and Pildes1998; Shapiro Reference Shapiro2016, 86–8). By comparison, then, Brennan's proposals would grant politicians far greater leeway in choosing their own voters, and should be regarded as far more dangerous.

Consider the possibility of conditionalizing suffrage on educational attainment. Brennan himself speculates (Reference Brennan2016, 223) that in the US, Democrats would favor very low or very high qualifications, while Republicans would prefer a mid-range qualification, such as a high school diploma. Yet he fails to recognize what a dramatic effect this could have: for either party, implementing a properly targeted policy would instantly translate a small temporary majority into a massive and durable electoral advantage. Where traditional gerrymandering faces natural limits, moreover—imposed by the need to distribute a fixed voting population among geographically contiguous districts—“epistemic” gerrymandering could continuously compound incumbent advantages through ever-more-targeted tests, recursively altering the voting population itself. Even if their immediate effects were not so extreme, finally, such policies would still give incumbents strong incentives to manipulate educational attainment patterns for partisan ends.

Implementing a competence test presents no solution, as incumbents would have similarly strong incentives (and even greater leeway) to manipulate its content. Once established, moreover, parties would likely scramble to prepare their supporters, much as they already conduct voter registration drives and get-out-the-vote campaigns in places without automatic registration and mandatory voting.Footnote ¹¹ The resulting arms race to “teach to the test” could hardly be expected to yield substantial benefits in vote quality—indeed, given the massive expenditures required, the most significant consequence would be to intensify parties’ dependence upon concentrated wealth.

Brennan concedes that his proposals could have drawbacks, including the potential for manipulation (Reference Brennan2016, 222–30). However, he asserts that they also promise vastly superior outcomes. Caricaturing any unwillingness to experiment with epistocracy as a species of Burkean conservatism, he insists that we cannot know in advance whether such tradeoffs will be worthwhile. I disagree. As I have already argued, we know enough about politicians to predict that some would use these tools to entrench their power. Perhaps more importantly, we also know enough about voters to predict that disenfranchising those with low political knowledge would not substantially improve the epistemic quality of electoral results. Thus, even if incumbents (or independent bodies) imposed perfectly benign epistemic qualifications, the benefits Brennan projects would be very unlikely to materialize.

Consider that voters in modern democracies do not choose directly between different policies. Instead, they choose between parties or candidates. There is thus substantial distance between a voter's overall level of political knowledge and the quality of her vote choice. Yet Brennan makes very little effort to demonstrate that higher political knowledge scores would be associated with greater likelihood of voting for the “correct” party or candidate.Footnote ¹² Instead, he relies on an intuitive sense that political knowledge must raise vote quality.Footnote ¹³

Perhaps, on the margin, it does. As Brennan himself insists throughout his book, however, our intuitive understanding of voting behavior cannot be trusted. In particular, it seems, rational assessment of the arguments for competing policies—which political knowledge tests purport to measure—plays a small role (at best) in vote choice. As extensive research on “motivated reasoning” has demonstrated, better information rarely changes our mind, it simply gives us better tools for defending our intuitions (Lodge and Taber Reference Lodge and Charles2013). Meanwhile, education primarily alters our policy preferences indirectly, by changing our social context and identity. Where they exist, then, correlations between political knowledge and vote choice are largely explained by unrelated, non-rational factors—centrally including the set of identity groups to which we belong (Achen and Bartels Reference Achen and Bartels2016). By far the most significant consequence of disenfranchising those individuals with low epistemic qualifications, therefore, would be to disenfranchise the (already disempowered) groups to which they disproportionately belong.

Brennan glibly dismisses this “demographic objection.” He announces repeatedly, for instance, that by his “objective,” “morally relevant” standards, black women are the least “competent” voters in the US, and that his proposals would therefore disenfranchise them in large numbers (Reference Brennan2016, 33, 132–3, 148, 227–8). Rather than taking this opportunity to reexamine his epistemological premises, however, he assures readers of his excellent implicit bias scores, and asserts that disadvantaged citizens cannot vote in ways that promote their interests “unless they have tremendous social scientific knowledge” (227).Footnote ¹⁴ Remedying their disadvantages thus requires disenfranchising them—and thereby entrusting their interests to better-informed voters.Footnote ¹⁵

Such claims are, quite frankly, preposterous, and one may be tempted simply to ignore them. In my view, however, it is worth showing why they are not supported by the empirical evidence Brennan cites, since this research does challenge many common assumptions about democracy, and it should not be dismissed along with Brennan's conclusions. It is well established, for instance, that most voters know too little about policy to make meaningful choices between parties on that basis—meaning that there is no “bedrock” of public opinion to which representatives might faithfully respond (Disch Reference Disch2011). According to Achen and Bartels (Reference Achen and Bartels2016), moreover, the minimalistic model favored by many political scientists—retrospective accountability—is also typically oversold (146–76). At best, it seems, voters hold representatives accountable for a fairly narrow set of outcomes (Healy and Malhotra Reference Healy and Malhotra2013); at worst, for nothing at all—or for outcomes representatives cannot control. This research thus warrants serious attention, if not a thorough rethinking of democratic theory (e.g., Green Reference Green2009). Yet even the most resolutely skeptical accounts do not remotely imply that low-knowledge voters would be better off disenfranchised.

For one, democratic realists like Achen and Bartels readily admit some correspondence between public opinion and policy outcomes. So long as they must fear electoral defeat, incumbents will adhere to “vague limits of permissiveness” set by public opinion, and though electoral punishment of specific misdeeds is rare, it does occur in extreme cases (Achen and Bartels Reference Achen and Bartels2016, 318–9). Uncertain about what will tip the balance from victory into defeat, politicians carefully manage their reputations, aiming to satisfy a range of interest groups without upsetting others too profoundly (Oppenheimer and Edwards Reference Oppenheimer and Edwards2012, 206–18). Even if the franchise does not meaningfully empower individuals, therefore, it does give non-negligible leverage to groups.

That said, we may grant that enfranchisement often brings surprisingly few benefits. For instance, nineteenth-century socialists and reactionaries both assumed working class suffrage would yield far more extensive redistribution than it did. We may also grant that most voters are sociotropic, meaning that their votes are directed at a common good, rather than their own self-interest. Yet even Brennan acknowledges that there is clearly some connection between enfranchisement and group interests, observing for instance that “if we deprive all black people of the right to vote [. . .] this will help facilitate people of other races in exploiting, dominating, and oppressing blacks” (Reference Brennan2016, 97–8). Nothing could be more obvious, of course, but given Brennan's assumptions, it cries out for explanation: Why might this be? Alas, it seems, not everyone votes sociotropically—and even when we do, our views of the “common good” are inevitably biased (Bagg Reference Bagg2018b).

In sum, empirical research in public opinion and political psychology does not dispute that groups—even those with below-average political knowledge—can use the electoral system to advance their interests. What the research does throw into question is the rational grounding such processes are often supposed to have. If “most citizens support a party not because they have carefully calculated that its policy positions are closest to their own, but rather because ‘their kind’ of person belongs to that party”—as Achen and Bartels (Reference Achen and Bartels2016, 307) summarize—members of disadvantaged groups may not always vote in ways that promote their interests. This is concerning, if unsurprising. Yet it hardly implies that their voting behavior is entirely insensitive to their interests—much less that their interests will be more reliably protected by others. Voters driven by identity rather than policy will effectively promote their interests as long as the identities they find salient lead them to support parties and candidates whose policies promote their interests.

To be sure, this is not always the case. But neither is it particularly implausible—voters do not choose their identities at random—and it seems especially likely when groups suffer collective disadvantages that render that identity affiliation salient. In such cases, ordinary group members will often follow the recommendations of respected leaders, many of whom would pass any epistemic qualification Brennan could propose. Because this compliance is largely driven by identity rather than informed and independent consideration of the issues, of course, leaders have latitude to extract rents and engage in various deceptions. Voter ignorance does have real costs. Barring exceptional circumstances, however—such as a sinister conspiracy among group leaders—it need not render their votes counterproductive, as Brennan suggests. Perhaps working-class voters could have done better than the social democratic parties of the early twentieth century—just as contemporary African Americans could do much better than the Democrats—but it is idle to pretend that such allegiances have no basis whatsoever in genuine protection of group interests.

We may admit, then, that members of disadvantaged groups often vote in suboptimal ways—just like everyone else. As we have seen, however, it is highly unlikely that instituting epistemic conditions on the franchise would reliably and significantly improve average vote quality. It is far more likely that they would be used to entrench the power of incumbents—potentially so effectively as to endanger the competitiveness of elections. Restricting the political rights of low-knowledge disadvantaged groups, therefore, would bring risks for those groups (and for the polity as a whole) that are far out of proportion to any hypothesized benefits. Insofar as Brennan's epistocratic proposals are intended for the real world, we must resoundingly reject them.

Limited Government and the Necessity of Robust Discretionary Power

We turn, finally, to the idea of limited government—that is, constraining and decentralizing political power through mechanisms like federalism, local governance, and judicial review, while encouraging greater reliance on competitive markets and interjurisdictional foot-voting. As Ilya Somin argues, this could “reduce the problem of public ignorance by reducing the number of issues to be decided by government to a level that voters would find more manageable” (Reference Somin2013, 141; see also Caplan Reference Caplan2007, 192–7). Indeed, something like this solution may appear as the logical extension of my argument: if epistocracy is dangerous because it enables state actors to accumulate excessive centralized power, perhaps the solution is to place tighter constraints on those actors.

As I demonstrate, however, the problems of entrenchment and capture are not so easily resolved. While certain limits on discretionary state power are advisable, others would actually facilitate the entrenchment of various elites, and ultimately the capture of the state itself. Thus, I argue that resisting state capture requires a delicate balance between “defensive” strategies that limit the state's discretionary power, and “offensive” strategies that use this discretionary power to target the potential perpetrators of capture—that is, the possessors of concentrated private power.

Consider first that even a “minimal” state—one constitutionally limited, perhaps, to defining and enforcing property rights—would still play a crucial role in the distribution of power, and would hardly be neutral between differently situated actors. Such a state would enable privileged elites to perpetuate their social and economic advantages quite effectively, while coercively preventing any redistributive efforts—legal or extra-legal—undertaken by coalitions of the disempowered. More generally, state protection of private property is fundamental to the social order in nearly all modern societies and, within this context, rules or structures prohibiting centralized redistribution between groups will predictably facilitate the entrenchment of elite power. Accordingly, such limits can be considered mechanisms of state capture.

Advocates of limited government may object to this description. Even if elites benefit disproportionately from state protection of property, they do not have arbitrary control over state action. Given a situation of highly asymmetrical social power, indeed, it might seem that increasing the state's centralized power would only give elites more powerful tools with which to dominate others. We should not brush this concern aside, given that discretionary state power has often been used in precisely this way (e.g., Somin Reference Somin2015). Nevertheless, historical experience also provides evidence of opposing tendencies.

In the infamous Lochner era between 1897 and 1937, for instance, the US Supreme Court repeatedly overturned redistributive legislation favored by wide majorities. Interestingly, some justices saw themselves as defending the “general” interest—namely, economic growth—against labor's attempts to capture the state in service of its “partial” interests (Rahman Reference Rahman2016, 64). In hindsight, however, it is far more plausible to understand these conservative justices as agents of state capture, defending the partial interests of industrial capitalists at the expense of nearly everyone else. For one, growth inevitably creates winners and losers, and cannot be understood as a “neutral” policy that is always in the “general” interest. Meanwhile, labor's eventual victories did not demonstrably slow growth, and in fact seemed to facilitate widely shared prosperity. Of course, “state capture” is no more self-interpreting than any other normatively laden concept, and as I have acknowledged, there will always be disagreement about how to apply it in practice. Nevertheless, it is broadly accepted that Lochner-era jurisprudence protected the interests of a small economic elite at the expense of a vast majority (Sunstein Reference Sunstein1987).

Nor is it the only example of US elites employing techniques of “limited government”—that is, constraining and decentralizing political power—to entrench their dominance. Southerners have long appealed to federalism in defending the institutions of white supremacy, for instance, while metropolitan elites routinely use the principle of local governance to thwart the efforts of disadvantaged groups to claim a greater share of regional resources (Anderson Reference Anderson2010; Hayward Reference Hayward2013). Here too, of course, opponents of redistribution by centralized federal or municipal authorities may claim that it represents a pernicious form of state capture—and here too, their claims are thoroughly implausible. Though the federal government coercively imposed the Civil Rights Act upon states, for instance, few today would dispute that it was the state-level Jim Crow laws overturned by the Act which more clearly embodied the use of coercive centralized power by a privileged class to perpetuate its (perceived) interests at the expense of others.

The point of these examples is not to show that limits on discretionary power are always mechanisms of state capture. As Somin and others correctly observe, constraining and decentralizing political power can often make capture more difficult or less consequential. Meanwhile, there is much to be said for democratic empowerment through “exit” rather than “voice” (Warren Reference Warren2011). As these examples show, however, limiting state power can also serve to enable dangerous forms of elite entrenchment. In particular, when popular efforts to counteract social and economic inequalities are routinely foiled by privileged classes using tools like judicial review, federalism, and local governance, their ability to prevent centralized redistribution certainly looks suspiciously like arbitrary control of the state.

Regardless of whether we extend the concept of “state capture” to this kind of obstruction, moreover, the increasing concentration of private power it enables will often be used to perpetrate capture in even more straightforward ways. As the social and economic power of privileged classes, interest groups, corporations, and individuals continues to grow, their political clout will inevitably expand as well. Once they become powerful enough, no well-intentioned constraints will effectively prevent them from using the state to implement what Lindsey and Teles (Reference Lindsey and Teles2017) call “upward redistribution.” If a state's commitment to strict limits on its discretionary power prevents it from resisting the rise of such actors, therefore, this commitment will ultimately prove self-undermining.Footnote ¹⁶ To stipulate otherwise is to engage in the sort of wishful thinking characteristic of central planners who assume incorruptible state agents, or meritocrats who rely on moral education to sustain independent power centers.

As a result, a state seeking to resist capture must not only “play defense” by means of self-imposed limits on their discretionary power, which can enable rampant concentration of social and economic power, in turn creating private actors and privileged classes with dangerous capacities to perpetrate capture.Footnote ¹⁷ In addition, it must go on the offensive, undermining these threats before they become impossible to resist. A state that merely refrains from perpetuating oligarchy or racism—without proactively combating these systemic forces—will become their instrument eventually.

What might constitute an appropriate balance between offensive and defensive strategies? Drawing on Progressive-era thinkers like Louis Brandeis, Sabeel Rahman (Reference Rahman2016) outlines a promising approach that avoids the pitfalls of both “laissez-faire” and “managerialist” solutions, applying it to the important question of financial regulation. Rather than letting markets run riot or engaging in capture-ripe fine-tuning of their daily operations, he argues, “structuralist” regulation would seek to eliminate the shared interests enabling financial giants to collude and capture regulators in the first place—perhaps by separating commercial and investment banking, for instance, or breaking up structurally important firms. In combination with non-electoral modes of democratic oversight, Rahman claims, this will allow us to minimize capture within the regulatory state while retaining crucial capacities for prophylactic action against concentrated private power.

My aim in this section, of course, has not been to defend any particular configuration of “offensive” and “defensive” strategies for resisting state capture—such as that proposed by Rahman—but simply to show that some such balancing act is necessary. Discretionary state power may be dangerous, but so are many of the limitations often proposed as solutions. As a result, some gambles will inevitably be lost. Some expansions of state prerogative intended to curb concentrated economic power, for example, may ultimately strengthen the oligarchs’ position. But surrendering the ability to take such prophylactic action is both unnecessary and ultimately counterproductive. Though some constraint and decentralization is advisable, it cannot represent a general solution to entrenchment and capture.