The virtues of truth: On democracy’s epistemic value

Introduction

Miranda Fricker’s Epistemic Injustice: Power and the Ethics of Knowing (2007) has now been widely regarded as a significant and original contribution to contemporary political philosophy. Besides Fricker’s (2013) own discussion of the relationship between epistemic justice and political freedom, Jose Medina (2013) has elaborated on Fricker’s analysis and developed a social connection model of shared responsibility in epistemic resistance, whereas Elizabeth Anderson (2012) argues that ‘the virtue of epistemic justice for institutions is otherwise known as epistemic democracy: universal participation on terms of equality of all inquirers’ (2012, 174). Generally speaking, existing literature has treated Fricker’s idea of epistemic injustice as a moral wrong that ought to be corrected. To the extent that this argument provides a justification for a particular set of social and political arrangements, it is an internal one addressed primarily to those who are already committed to the egalitarian ideal of free and equal epistemic participation. Insufficient attention has then been paid to the fact that Fricker’s genealogical narrative, presented most noticeably in chapter 5 of that book, is a pragmatic one built directly upon the previous effort by Bernard Williams in Truth and Truthfulness: An Essay in Genealogy (2002). ¹ Its purpose is to vindicate Testimonial Justice, understood as a third virtue of truth in addition to Williams’s Accuracy and Sincerity, more in functionalist and less moralized terms, not merely as a virtue-based remedy for epistemic injustice, but also as a fundamental epistemic virtue arising out of the basic epistemic needs that are present in any human society, whether egalitarian or not (Fricker 2007, 108).

In this article, I want to push the functionalist explanation offered by Fricker and Williams a bit further, using them to develop an external justification of democracy that is not dependent on a prior commitment to political liberty and equality. This is not to say that one should not appeal to the intrinsic value of political liberty and equality or their expressive properties in her normative justification of democracy. Instead, I only want to point out that there exists a justificatory division of labour targeting different audiences. Specifically, my goal here is to demonstrate to those who are otherwise sceptical of the intrinsic and expressive values of democracy that the regime also has epistemic value, especially for a modern pluralistic society. This is because the principles of political liberty and political equality, when properly institutionalized, help the government to take fuller account of diverse forms of knowledge possessed and acquired by its citizens. Drawing on Williams’s Truth and Truthfulness and Fricker’s Epistemic Injustice, I argue that the virtues of truth required for the social production and aggregation of knowledge are more likely to be found in a democracy than in an autocracy or an epistocracy.

The organization of the article is as follows. In the first section, I reconstruct and elaborate on the genealogical narratives of Williams and Fricker, explaining what the virtues of truth are, why they are essential to the social production and aggregation of knowledge and how they are positively correlated with a political regime’s epistemic capabilities. I then compare democracy, respectively, with autocracy and epistocracy in the next two sections, claiming that it is likely to be, within the context of a modern pluralistic society, an epistemically superior regime in the sense that it creates more favourable conditions for people to display those virtues required for the pooling of epistemic resources. Finally, I conclude by considering how my argument helps to bridge the gap between epistemic and non-epistemic approaches in democratic theory and establish a multi-dimensional framework of democratic legitimacy. Within this framework, democracy’s epistemic value is directly tied with both the safeguard against elite domination and the development of citizens’ ethical and intellectual capabilities. In this regard, the wisdom of the multitude goes together with the power and virtue of the multitude.

Before proceeding to the main argument, three things need to be clarified. First, without providing a full explanation regarding what should be included in diverse forms of knowledge, I want to note at least three types of knowledge that are relevant for political decision-making: specialized knowledge, local knowledge and perspectival knowledge. Specialized knowledge covers knowledge of various sciences, disciplines and practices, including those of economics, epidemiology, education and public administration that are essential to good governance. This is also the type of knowledge that epistemic critics of electoral democracy argue ordinary voters often fall short of, but which, as I will demonstrate in the third section, could be even more difficult to collect in an epistocracy. Local knowledge refers to information about particular circumstances of time and place, social and economic conditions required for the formulation of policy proposals. As pointed out by Friedrich Hayek (1945), this type of knowledge does not exist in any coherent whole but is widely dispersed across the society. ² Perspectival knowledge, which could also be understood as knowledge about beliefs or ‘soft knowledge’, describes what the wider population think of certain propositions and how they might react to them (Rehfeld 2005, 26). It is needed by almost all governments because the effectiveness of a given policy (e.g. raising taxes) is dependent on both the proposal itself (e.g. setting out new tax rates) and the behavioural responses of the citizenry (e.g. the extent of tax evasion under the new proposal). While the significance of perspectival knowledge does not mean that public officials should follow and respond to their citizens’ preferences, it does require them to have an accurate grasp of most people’s perspectives when evaluating the practicability of a given proposal.

Second, I need to clarify what I mean by democracy, as well as the two non-democratic alternatives that I choose to compare it with, autocracy and epistocracy. ³ As I understand it, democracy is a regime that is committed to both people’s political liberty and political equality. Now this understanding is not necessarily a liberal conception of democracy. After all, as Josiah Ober (2017) has pointed out, both critics and defenders of democracy as it was practiced in classical Athens and other Greek city-states, that is, before liberalism, regarded ‘a conjoined commitment to both liberty and equality as preeminently characteristic of democratic states’ (2017, 110). Moreover, depending on social, historical and cultural circumstances, those two principles could take various forms. For example, in the classical Athenian democracy, political liberty was reflected in the notion of isēgoria, or the freedom to speak out in decision-making bodies, as well as a constitutional framework in which the popular courts, assembly and council check one another’s power. Political equality, on the other hand, were further instantiated by the notions of isonomia and isokratia that establish equal access to law, public offices and other political institutions. In a modern representative democracy, however, political liberty usually takes the form of individual rights that guarantee freedom of speech, press and association, whereas universal suffrage, in the form of ‘one person, one vote’, can be viewed as a public, though very imperfect, acknowledgement of citizens’ status as political equals. ⁴

Still, the difference between democracy and the other two non-democratic alternatives should be in principle clear. Specifically, I define autocracy as a regime violating both political liberty and political equality, where people enjoy neither the freedom from abuses of political power nor equal political standing. By contrast, I grant that epistocracy respects political liberty and has mechanisms in place to check abuses of political power. Yet, it does not treat people as political equals, but categorizes them and distributes political power according to an epistemic labelling of political competence. Like democracy, autocracy and epistocracy can also take various forms. As a result, despite my use of some empirical examples, I will analyse those three regimes mostly as ideal types in this article, via their relationship with political liberty and political equality. I thereby acknowledge the limits of my argument that additional work is needed to determine whether it would apply to all versions of democracy, autocracy and epistocracy, as well as to what extent actual regimes match my ideal types.

Finally, because I seek to provide an epistemic argument for democracy on the basis of its ability to incentivize more people to display the virtues of truth, I will focus mainly on the instrumental value of those virtues and the role they play in the social production and aggregation of knowledge. This is not to say that the cultivation of such virtues should be based only on an instrumental understanding of them. In fact, doing so would be self-defeating, given that their instrumental value lies largely in the one they have for the society as a whole and is thereby highly vulnerable to the free-rider problem (Queloz 2021, 167). In other words, although the virtues of truth would significantly improve a community’s ability to pool its epistemic resources, their utilities to each of its individual members may be outweighed by other considerations. The fear of missing out certain truths, for example, is often not enough to convince a person to overcome his intellectual laziness, dishonesty or prejudices and to behave in an epistemically virtuous manner. Partly because both Williams and Fricker recognize the motivational gap of pure instrumentalism, their respective works have also considered various reasons why people should value Accuracy, Sincerity and Testimonial Justice for their own sake, as opposed to merely for their epistemic benefits. ⁵ However, due to the limited scope of this article, I will not examine those intrinsic reasons here. This does not mean that I regard them as less important than instrumental ones related to epistemic efficiency.

Accuracy, sincerity and testimonial justice

In Truth and Truthfulness, Williams argues that our dispositions of truthfulness emerge first from the basic epistemic needs that are present in any human society. To illustrate the point, he constructs a fictional State of Nature, a minimal human society where a group of individuals who share a common language have to cooperate with each other so as to facilitate their own survival. It will then be profoundly advantageous for each of them to have at her disposal not only the information that she herself is in the best position to acquire, but also what other individuals in different positions could have acquired. Here Williams (2002) introduces the idea of purely positional advantage that ‘a speaker can tell someone else about a situation because he is or was in it, while his hearer is not or was not’ (2002, 42). For example, I can see and tell you that a predator is coming because I am up on a tree whereas you cannot because you are lying on the ground. Consequently, one’s individualistic need for true information gradually develops into a community’s collective need to participate in a division of epistemic labour, where each member contributes what she knows to a shared pool of information to which everyone else also has access.

However, such a division of epistemic labour is sustainable only under certain conditions. Specifically, it requires those who provide the information to display two kinds of dispositions concerning truthfulness, which Williams called ‘Accuracy’ and ‘Sincerity’: ⁶

One kind of disposition applies to their acquiring a correct belief in the first place, and their transporting that belief in a reliable form to the pool. The other desirable dispositions – desirable, that is to say, from the social point of view of those using the pooled information – are necessary because reflective creatures will have the opportunity within this structure for deceit and concealment; they will also have the motives for them, as when a hunter has found a prey which he would rather keep for himself and his immediate family. (This is the force of Voltaire’s famous remark to the effect that men have language in order to conceal their thoughts.) This second group of dispositions centrally contains the motivation, if one is purporting to tell someone something and the circumstances are right, to say what one actually believes. (Williams 2002, 44)

Williams’s account of Accuracy and Sincerity has also demonstrated two important points of social epistemology. First, the vast majority of our knowledge is second-hand knowledge learnt not by ourselves, but from others’ testimonies, including those of our parents, friends, teachers and colleagues who have displayed the virtues of truth. Second, most of our communicative acts can be regarded as some kind of epistemic activity that is minimally defined by a concern with fidelity to how things actually are, rather than what we wish them to be or how we take them to be. In this regard, neither the Platonic dichotomy of doxa versus episteme nor the Rawlsian public reason’s claim to epistemic abstinence can be fully replicated in a proper functioning human society, since part of the reason why people expect their conveyed opinions to be taken seriously by others is that they constitute sincere assertions aiming at the truth. ⁷

Considering that any political community’s ability to perform its basic functions (e.g. to identify internal and external threats that endanger civil peace and political stability) requires the practice of pooling epistemic resources, the virtues of truth become practically even more necessary if we move from Williams’s fictional State of Nature to an actual political community. This is especially the case when we take two other factors into account. First, no political community is an island entirely unto itself. Instead, they exist in a competitive ecology of states, where each often needs to actively deploy the epistemic resources of its citizenry to make ‘good enough’ decisions and hold its place (Waltz 1979). Second, as I have pointed out in the introduction, the types of knowledge relevant for political decision-making include not only specialized knowledge which can, at least theoretically, be commanded by a smaller group of experts (though in practice less so, especially in a modern pluralistic society where most decisions require a very broad range of specialized knowledge) but also local and perspectival knowledge that are widely dispersed across the society. Indeed, to the extent that a political community could survive and flourish in a competitive international environment through its use of those different forms of knowledge possessed and acquired by its citizens and that the virtues of truth allow its government to have fuller access to such knowledge, we could claim that the regime which encourages the display of those virtues is likely to outperform the one that does not.

But to have fuller access to those different forms of knowledge, there may have to be another virtue besides Accuracy and Sincerity. As Fricker (2007) has pointed out in Epistemic Injustice, one problem with Williams’s State of Nature story is that it is told entirely from the speaker’s perspective and does not include any material needed for ‘gathering truths form others’ (2007, 114). If there would be no inaccuracy and insincerity left once some people began to show Accuracy and Sincerity, then those who received the information might not need to do anything at all and could simply take others’ testimonies to be accurate and sincere. Yet this would certainly be an overly idealistic view of any human society and would thus deprive Williams’s State of Nature story of its explanatory power. In reality, the pooling of epistemic resources does demand something more than Accuracy and Sincerity: not only should those who provide the information acquire a correct belief and tell others what they actually believe, but those who receive the information must also be able to properly discriminate between different informants so that only correct information is being transmitted from one person to another. The latter practice, which is about accepting truths and rejecting falsehoods in a reliable manner, should thus be regarded as the third virtue that is essential to the social production and aggregation of knowledge.

According to Fricker (2007), one problem which this third virtue has to address is the counter-rational influence exerted by epistemically flawed social categorizations (2007, 116). To be clear, most of human interactions require people to engage in social categorization, which, if done properly, allows the transfer of tacit knowledge through the use of ‘trust in testimony’ (Benson 2019, 77). For example, when a doctor recommends a treatment to a patient, the patient, who lacks relevant medical knowledge, cannot directly judge the reliability of the treatment itself. Instead, she has to rely upon her view of the doctor (e.g. whether he fits into the category of a ‘dependable doctor’) to decide whether she should follow his recommendation (Benson 2019, 84). When being based on an empirically reliable generalization (e.g. the doctor is a dependable doctor because his treatments have already cured many other patients), social categorization constitutes an epistemically productive exercise, which enables the hearer to judge the Accuracy and Sincerity of the speaker and to transmit or act upon his knowledge even when the knowledge itself has not been explicitly expressed (e.g. the patient followed the doctor’s treatment and found herself cured despite not having been told how and why the treatment would work).

However, social categorization will incur epistemic costs if what it embodies is not an empirically reliable generalization, but rather an unreliable generalization – of which one prototypical example would be an identity-based partiality or prejudice. For instance, we can imagine that persons living in Williams’s State of Nature may out of pure contingencies be divided into several subgroups and thereby come to have in their mind concepts such as ‘trustworthy insiders’ and ‘ignorant outsiders’. A may then believe that B is correct whereas C is wrong not because B is correct and C is wrong, but because he happens to share the same group identity with B rather than with C. In other words, people’s judgements of others’ Accuracy and Sincerity can be easily affected by certain basic, either positive or negative, identity-based stereotypes, which, if they remain unchecked, will undermine their ability to properly discriminate between different informants and eventually contaminate the society’s public pool with misguided information. Consequently, the third virtue of truth has to be a corrective virtue, which helps us to match the level of credibility we attribute to our interlocutors to the evidence that their testimonies are sincere assertions aiming at the truth, rather than our positive or negative perceptions of their social identities. To harmonize with Williams’s account with capitals, Fricker (2007) has named this reflexive awareness in the evaluation of others’ testimonies as ‘Testimonial Justice’ (2007, 116). Practiced by the hearer rather than the speaker, it aims to neutralize the counter-rational influence which identity-based stereotypes tend to have on people’s credibility judgements (Fricker 2007, 116). As the essential counterpart to Accuracy and Sincerity, Testimonial Justice emerges also from the State of Nature as a fundamental virtue that helps to sustain the pooling of epistemic resources.

Now I want to move again from the fictional State of Nature to an actual political community, so as to allow us to also appreciate Testimonial Justice as an important remedial virtue that enables a political community to expand and better exploit its epistemic resources. First, considering the vast array of identity-based stereotypes that have been operative in almost all political communities, the existence or lack of Testimonial Justice could well determine the size of epistemic resources which the government would have at its disposal. For example, for long periods of human history, being a rich male landlord constituted probably the positive indicator of epistemic authority. By contrast, women and men of other occupations were generally not treated as persons who could acquire or possess epistemic resources simply because they were women and men of other occupations. In other words, they were not granted with Testimonial Justice by those in power, who also, under most circumstances, turned out to be rich male landlords. The epistemic resources which the government could then deploy were comparatively limited because the vast majority of its population had been preemptively excluded from making any epistemic contribution and were not even given the opportunity to display Accuracy and Sincerity. This is a problem pointed out, for example, by John Stuart Mill (1989) in The Subjection of Women, where he claims that ‘the loss to the world, by refusing to make use of one-half of the whole quantity of talent it possesses, is extremely serious’ (1989, 199). A similar argument has been made by Ober in his description of classical Athens as an epistemic democracy. While demonstrating the city’s economic and military superiority over most of its less democratic or non-democratic rivals, Ober (2008) also contended that the exclusion of women, slaves and long-term residents from political participation had prevented Athens from having full access to vital epistemic resources that might have enabled it to perform even better within a highly competitive international environment (2008, 258–63).

Furthermore, it is through the practice of Testimonial Justice that we establish the kind of discursive environment which can incentivize more people to display Accuracy and Sincerity. After all, one problem brought about by either positive or negative identity-based stereotypes is that the credibility which people receive will not be proportional to their epistemic contributions. Instead, some receive more than they should have – a credibility excess – whereas others get less than they should have – a credibility deficit (Fricker 2007, 17). Both of those scenarios could significantly undermine Accuracy and Sincerity and thus reduce a political regime’s epistemic capabilities. On the one hand, the person who is endowed with credibility excesses is likely to develop some kind of epistemic arrogance that leaves him less concerned about Accuracy and Sincerity. He does not need to care much about acquiring a correct belief or telling others what he actually believes because many people will simply assume what he says to be accurate and sincere. In some extreme cases, one who has long been overly esteemed in his capacity as a knower may eventually become a cognitively disabled individual, a ‘Mr. know it all’ who could not even recognize, let alone admit, any of his own mistakes and limitations (Fricker 2007, 20). On the other hand, the person who suffers credibility deficits has to overcome additional barriers because of others’ prejudices regarding her social identity. To make the situation worse, she may well be tempted to give up on intellectual inquiry (Accuracy) or choose to not share what she knows with others (Sincerity) if it becomes clear that whatever she conveys is unlikely to be taken seriously by the society at large. In this regard, prejudices operating against socially disadvantaged groups do often have a self-fulfilling power, so that those who have been denied Testimonial Justice eventually end up further confirming the stereotypes which others have about them (Fricker 2007, 55). As a result, the entire political community remains trapped within a vicious cycle, where incentives for intellectual laziness and dishonesty continue to multiply.

However, such a situation will begin to change if enough people can overcome the counter-rational influence exerted by identity-based stereotypes and show Testimonial Justice in their communicative acts. Along with the decline of credibility excesses and credibility deficits will emerge the kind of discursive environment that encourages the display of Accuracy and Sincerity. Not only will the disadvantaged be more likely to have the epistemic authority that they deserve to have from their epistemic contributions, but the advantaged will also have to work harder if they want to keep their positions that were previously not fully earned. Mill (1989), for example, has envisioned precisely such a scenario when arguing for the emancipation of women that the overcoming of gender stereotypes would not only give freedom to ‘one-half of the whole sum of human intellect’ but also bring ‘the benefit of the stimulus that would be given to the intellect of men by the competition, or (to use a more true expression) by the necessity that would be imposed on them of deserving precedency before they could expect to obtain it’ (1989, 99).

We have thus seen, from a reconstruction of Williams and Fricker’s genealogical narratives, what the virtues of truth are, why they are essential to the social production and aggregation of knowledge and how they are positively correlated with a political regime’s epistemic capabilities. The entire account is also a functionalist one demonstrating that those virtues are practically, rather than morally or metaphysically, necessary. In the following two sections, I seek to offer an epistemic argument for democracy on the basis of its ability to incentivize more people to display such virtues. By doing so, I also hope to investigate the kind of political structure that creates more favourable conditions for individual epistemic virtues to work (Anderson 2012, 168). My ultimate goal, however, is to develop an external vindication of democracy, derived from the more basic epistemic needs that are present in almost all political communities, whether one identifies political liberty and political equality as intrinsic goods or not.

Autocracy and the problem of untruthfulness

In chapter 9 of Truth and Truthfulness, Williams (2002) examines the general relations between truthfulness and politics. He focuses specifically on what he called the anti-tyranny argument for truthfulness in political life that it is in citizens’ interests to have true information, so that the governments cannot conceal their illegitimate actions as they wish (2002, 207). Moreover, to the extent that democracy in its modern constitutional forms is valued in the name of liberty and that the falsification or suppression of information impedes the exercise of liberty, there seems to be a particular connection between a modern democratic regime and ‘a demand for governmental truthfulness’ (Williams 2002, 210–11). Indeed, one of the conditions for democratic legitimacy is that ‘the government should not (regularly, except for special cause, etc.) deceive ordinary citizens’ (Williams 2002, 211).

However, building on the argument presented in the last section, I want to suggest that Williams’s own discussion of the issue has probably been too one-sided. Insofar as he is concerned about the value of truthfulness in politics, it is because the governments might hide true information from the citizens so as to conceal their own illegitimate actions. But his genealogical accounts of Accuracy and Sincerity have also shown that the basic epistemic needs of a political community also require its government to have true information from its citizens. Consequently, we may switch the perspective and see the societal function of political liberty not as demanding governmental truthfulness, but as encouraging more people to be truthful with themselves and each other. To see this point, let us consider how the lack of political liberty discourages the display of Accuracy and Sincerity and impedes a society’s ability to pool epistemic resources or what I would call the problem of untruthfulness faced by autocracy.

First, to the extent that the governors, as pointed out by David Hume, have nothing on their side but opinion (for force is always on the side of the governed), it is in the interests of almost all governments, whether autocratic, epistocratic or democratic, to pool perspectival knowledge, so as to accurately grasp the degree of support which they and their policies have among the general public. Yet this is more difficult to accomplish in an autocracy, where the lack of political liberty leaves ordinary citizens to be fearful of their government and ultimately be insincere with it, that is, not revealing their true level of support for the regime and its policies. As a result, governmental officials find it easier to indulge in wishful thinking, which eventually leads them to become inaccurate with themselves. The sudden collapse of the former Warsaw Pact regimes, for example, caught most people, including many leaders of those regimes, by surprise partly because the majority of the population within those countries had long been untruthful with their governments and engaged in preference falsification, which in turn allowed those in governments to give in to self-deception and ignore the fact that their regimes were substantially more vulnerable and less popular than they seemed (Kuran 1997, 266). Indeed, one of the classical arguments against autocracy revolves precisely around this information problem faced by the autocrat that it may be costly for him to trust even those closest to him because few persons would be sincere with him. As Xenophon put in the short dialogue between Hiero and Simonides:

To the private man it is immediately a sign that the beloved grants favors from love when he renders some service, because the private man knows his beloved serves under no compulsion. But it is never possible for the tyrant to trust that he is loved. For we know as a matter of course that those who serve through fear try by every means in their power to make themselves appear to be like friends by the services of friends. And what is more, plots against tyrants spring from none more than from those who pretend to love them most. (Xenophon, reprinted in Strauss 1991, 7–8)

Second, insofar as the lack of political liberty insulates the ruling elites in a cocoon where less true information gets through to them and fewer persons dare to be sincere in challenging their inaccurate beliefs, it will also be more difficult for an autocratic government to acquire specialized and local knowledge. In this regard, the epistemic case against autocracy is not that it cannot establish institutions such as deliberative forums consisting of cognitively diverse individuals or persons possessing relevant expertise (for it theoretically can and sometimes will), but that people are less likely to display Accuracy and Sincerity required for the pooling of epistemic resources within those institutions when there is no effective mechanism to check abuses of political power.

Calvert Jones’s (2019) study of the role played by expert advisors in the Arab Gulf monarchies serves as a good example that helps to illustrate the point. During the past few years, the Gulf monarchies, including Saudi Arabia, Kuwait, Bahrain, Qatar, the United Arab Emirates and Oman, increasingly choose to consult a wide range of local and foreign experts in their governance, hoping to use their epistemic resources to build a knowledge-based economy in the preparation for a post-petroleum era (Jones 2019, 12). Yet such a strategy has largely failed to produce the desired outcomes. Far from bringing relevant knowledge, data and experiences to the government, the increasing use of experts actually has actually made political decision-making more irrational, by encouraging an unrealistic and inaccurate belief among authoritarian ruling elites, that they can accomplish almost anything, no matter how rapid or unreasonable. The actual cause of this epistemic dysfunction can again be traced back to the lack of political liberty and the disincentives it creates for Accuracy and Sincerity. On the one hand, those expert advisors, after learning that they inhabit in an authoritarian political environment with a high potential for arbitrary dismissal and a complete lack of checks and balances, soon become insincere in the advice they offer and choose to ‘self-censor, exaggerate successes, and downplay their misgivings in response to the incentive structures they face’ (Jones 2019, 3). On the other, the authoritarian ruling elites, confronted with few challenges to their preconceived beliefs, find it more difficult to resist the temptation of believing the agreeable and eventually end up being further trapped into wishful thinking, with ‘excessive levels of optimism about the choices before them and the possibilities for change’ (Jones 2019, 19).

The problem of untruthfulness faced by autocracy in turn illustrates the instrumental value of political liberty in helping to pool epistemic resources. We can also see that this value may become more significant in a modern pluralistic society, where both domestic governance and international competition require the government to have fuller access to diverse forms of knowledge. Still, I need to distinguish my argument from Mill’s liberal belief in a ‘marketplace of ideas’, which Williams (2002) has regarded as being too optimistic (2002, 212). ⁸ What I am advocating here is a more limited and modest case for political, rather than individual, liberty. Instead of claiming, as Mill does, that a social system permitting the maximum degree of individual freedom is most conductive to the emergence of truth, I am only suggesting that the freedom from abuses of political power helps to correct the distorting effects that such abuses usually have on the social production and aggregation of knowledge. I thereby do not take a position on whether liberal rights more broadly understood, including the right to make disorderly, offensive and prejudicial speech, are a net positive for discovering, sharing and transmitting the truth. ⁹ Indeed, precisely because my conception of democracy is not necessarily a liberal one, I have no theoretical objection against a regime that regulates or puts limits on those rights, especially if they do, given particular social, historical and cultural circumstances (e.g. the proliferation of echo chambers in the age of internet), come into conflict with the more basic epistemic needs of the political community. ¹⁰

Nevertheless, my vindication of political liberty still does not constitute an epistemic argument in favour of democracy, since I have not yet considered epistocracy, the regime that respects political liberty, but not political equality. To this question I turn in the next section of the article, where I will bring Fricker’s Testimonial Justice, as well as the incentives it creates for Williams’s Accuracy and Sincerity, into the analysis. Specifically, I am going to show that political equality, understood as the presumption of equal political competence, can be defended on prudential grounds once we have clarified the epistemic difficulties involved in the very act of categorizing political competence in a modern pluralistic society, as well as how such a categorization will undermine the practice of Testimonial Justice and hence ultimately Accuracy and Sincerity. Moreover, I will suggest that democracy is likely to be epistemically superior to epistocracy over the long term, since the presumption of equal political competence helps it to avoid the epistemic risks posed by epistocratic categorization.

Epistocracy and its epistemic risks

It is important to first point out two major problems with existing epistemic critiques of democracy. One is that most of them are targeted at purely electoral forms of democracy, whose democratic credentials are problematic even under ideal circumstances (Landemore 2020). The social scientific literature on the problem of voter ignorance, for example, is based on the theoretical insight that it is rationally irrational: given the minuscule impact that any single vote has on the eventual policy outcome, it is instrumentally rational for a voter to be epistemically irrational, and thereby not make reasonable efforts to avoid fallacious reasoning, or keep an open mind for new evidence (Caplan 2007). Yet, by its own logic, this claim would be inapplicable in a randomly selected, periodically renewed citizen assembly empowered to make policies, where each member will have a much stronger incentive to ‘invest in acquiring the relevant knowledge and competence’, precisely because her views will matter a great deal (Abizadeh 2020, 11). In addition, this assembly as a whole would be cognitively more diverse, which enables it to deliver additional epistemic benefits through improvements also in collective intelligence (Landemore 2012). But perhaps most importantly, such a use of sortition, which Helene Landemore (2020) has described as ‘lottocratic representation’, constitutes a fuller realization of political equality than purely electoral representation via ‘one person, one vote’, since all citizens would have an equal prospect of being selected to office (Manin 1997, 56–57). ¹¹ How and to what extent a modern state should transit towards a more democratic form of government incorporating lottocratic representation are separate and largely empirical questions that I cannot address in this article. Rather, my point is that even if we grant existing epistemic critiques of electoral democracy, they may still not be valid with respect to alternative, and possibly more egalitarian, forms of democracy.

Second, epistemic critiques of democracy often paint an extreme caricature of democracy, as a regime that is inherently hostile to relevant expertise. A more realistic view, however, reveals that democracy could incorporate, in both its modern electoral and ancient non-electoral forms, certain epistocratic mechanisms that are designed to elicit competent government, yet compatible with political equality. With respect to the former, Dimitri Landa and Ryan Pevnick (2020) have shown that at the heart of modern electoral democracy lie the treatment and selection mechanisms that when properly institutionalized, it can generate epistemic benefits in a way that adheres to the egalitarian commitment of democratic citizenship. As for the latter, Ober (2013) has demonstrated how ‘relevant expertise aggregation’, where options presented by domain experts are subjected to judgement by ordinary citizens, has allowed the classical Athenian democracy to effectively respond to external threats such as the Persian invasion. Moreover, even contemporary epistemic democrats, who have been very optimistic about the collective wisdom possessed by average citizens, are not against the idea of deference and delegation to particular groups of experts on particular issues. As Landemore (2012) has pointed out, the relevant comparison for an epistemic theory of democracy is ‘not between democracy and that technocratic branch of the government but between democracy and oligarchy when both are equipped with a competent technocracy of that kind’ (2012, 204).

Consequently, when comparing democracy with epistocracy, one thing that needs to be clarified is what political equality actually means in epistemic terms. As I understand it, no democrat would reject the significance of relevant expertise and presume all citizens to be equally competent on all matters. Instead, what democrats are committed to is something much more specific: all citizens are presumed to be equally competent on political matters that deal with how they shall live well together. A good example of this view is found in Plato’s Protagoras, where the celebrated sophist Protagoras, whom some view as the first democratic theorist in the history of philosophy, argues, against Socrates, that citizens who are unequal in other aspects deserve an equal right of speech when it comes to discussing about what is best for their political community. By contrast, Plato himself, widely regarded as the prototypal theorist of epistocracy, not only treats people as fundamentally unequal with respect to their political competence, but also, as shown in the Statesman, takes political competence to be a kind of ‘master knowledge’ that is capable of instructing and coordinating all other ‘subordinate knowledge’, including rhetoric, military and juridical expertise (Moore 2017, 18). It is this latter view that I believe will discourage the display of Testimonial Justice and hence ultimately Accuracy and Sincerity. To see the point, let us consider the epistemic risks involved in the epistemic labelling of political competence.

Because it serves as the ‘master knowledge’, political competence, at least within the context of a modern pluralistic society, has to actually cover a broad range of often unrelated competencies, which include epistemic claims ranging from economics and natural science to government operations and foreign affairs. Hence, contrary to what supporters of epistocracy usually assume, the epistemic labelling of political competence is not like medical licensing that evaluates only a person’s level of relevant knowledge within a particular domain. Instead, it is something that requires the constant tracking and weighting of different kinds of knowledge across multiple domains (Bhatia 2020, 724). For example, suppose there is a person who knows a lot about macroeconomics but fails to accurately name most governmental agencies, it will be epistemically difficult to determine her level of political competence. On the one hand, she may be ignorant in the sense that she lacks certain knowledge regarding public administration. On the other hand, she also seems to possess the specialized knowledge required for formulating sound fiscal policies. To make the situation more complicated, the types of knowledge relevant for political decision-making, as I have already pointed out, include not only specialized knowledge which can, at least theoretically, be identified through a general knowledge survey (though in practice less so, considering that any such survey could ask only a limited number of questions and underscore specialized knowledge of certain disciplines), but also local and perspectival knowledge that are widely dispersed across the society and much more difficult to be identified beforehand. Due to those epistemic difficulties, the categorization of political competence constitutes an epistemically daunting task in a modern pluralistic society, one which tends to be inadequately executed in practice and is hence especially vulnerable to the cognitive shortcuts provided by existing identity-based stereotypes. In other words, even if we leave aside its significance for the distribution of political power and the strong incentives socially advantaged groups will have in manipulating its criteria, epistocratic categorization is already where people are least likely to exhibit reflexive awareness and display the virtue of Testimonial Justice.

Moreover, precisely because political competence is regarded as the ‘master knowledge’, the very act of categorizing persons as politically competent or incompetent could have a spillover effect, which influences how their epistemic abilities would be perceived in general. In this regard, credibility judgements about political competence, which may already be affected by existing identity-based stereotypes, could well end up further entrenching those stereotypes. As a result, people who have not been granted with Testimonial Justice in terms of their political competence are also less likely to be granted with Testimonial Justice in other walks of life. Epistocratic categorization could then incur epistemic costs on at least two fronts: not only will it wrongly categorize members of socially disadvantaged groups as disproportionally incompetent on public affairs, while unjustifiably amplifying socially advantaged groups’ epistemic credentials, but it will also further increase the credibility excesses and deficits to which people are already subject in their private lives. This mutually reinforcing relationship between credibility judgements about political competence and the counter-rational influence exerted by identity-based stereotypes allows us to defend political equality, understood as the presumption of equal political competence, on prudential grounds: it serves as an equalization of an otherwise unequal, and often unjustified, access to the markers of credibility.

As it does in the case of political liberty, this instrumental value of political equality is likely to become more significant in a modern pluralistic society, where both the government needs to collect more diverse forms of knowledge, and the epistemic labelling of political competence turns out to be epistemically more difficult. We may then identify three stages of what might be called the emergence of the epistemic value of political equality. First comes the Platonic ‘philosopher king’ stage, in which the ‘wise’ few are regarded as the only ones who possess the types of knowledge relevant for political decision-making. Then arises the second ‘mixed government’ stage, where individual members of different social groups are unequally represented in political institutions because (1) both domestic governance and international competition require the government to have fuller access to forms of knowledge that are not possessed by the ‘wise’ few and (2) individual members of some groups are still viewed as always being much more ‘knowledgeable’ than those of others. Finally, we can see a democratic stage, when members of a modern pluralistic society, realizing their propensities to misjudge one another’s degree of political competence due to the epistemic difficulties involved and the counter-rational influence exerted by identity-based stereotypes, as well as the long-term, accumulative epistemic costs of such misjudgements, take epistocratic categorization out of their own hands and presume everyone to be equally competent on political matters.

After we reach this third stage, the problem with epistocratic categorization also becomes clear: it undermines the practice of Testimonial Justice and hence ultimately Accuracy and Sincerity that facilitate the pooling of epistemic resources. In other words, one does not have to appeal to any intrinsic harm of political inequality for a person’s moral worth to see that the presumption of equal political competence in democracy avoids the epistemic risks posed by epistocracy. Neither is this epistemic argument in favour of democracy directly associated, as the one based on the Condorcet Jury Theorem is, with particular democratic procedures that are adopted under specific social, historical and cultural circumstances. Instead, what it emphasizes is a more general account of the symbolic value of political equality: it reduces the epistemic arrogance of socially advantaged groups and makes them to engage with disadvantaged groups on a more equal footing, thereby helping to create the kind of discursive environment in which more people can expect to be granted with Testimonial Justice and thus be incentivized to display Accuracy and Sincerity.

Once enough people have displayed the virtues of truth, epistemic resources distributed across the society can be pooled in a modern state with the help of democratic representation. Here I would endorse Michael Fuerstein’s (2020) account of democratic representatives as epistemic intermediaries, who occupy a natural mediating position between experts and non-experts and can thereby facilitate the integration of the specialized knowledge commanded by the former and the local and perspectival knowledge held by the latter. In this regard, the epistemic value of concrete democratic practices, including in the case of electoral representation: universal suffrage and in the case of lottocratic representation: equal eligibility, lies not primarily in the role they play at the moment of authorizing representatives but in the discursive forms of interaction that they help to activate between members of different social groups, representatives and represented, in the public sphere as well as formal state institutions, with the aim of integrating widely and asymmetrically distributed knowledge about political matters.

By contrast, let us consider how the epistemic risks carried by epistocratic categorization could impede a society’s ability to pool various forms of knowledge. With respect to local and perspectival knowledge, the risk is that an epistocratic government would be less attentive to them and thereby lose accurate grasp of what is happening on the ground and how effective its policies have turned out to be. This is because less educated citizens who lack specialized knowledge are less likely to be granted with Testimonial Justice by epistocratic categorization. As a result, they could be disincentivized to display Accuracy and Sincerity in acquiring and sharing the local and perspectival knowledge that are usually within their reach. For example, manufacturing workers who lack economic knowledge and support protectionist policies could well have accurate perceptions regarding the effects of plant closures on their local communities. Yet, such information about what has gone wrong with the current economic arrangement is more likely to be received and taken seriously by a democratic government than by an epistocratic one, since the former’s public officials have stronger incentives, both epistemic (because the presumption of equal political competence does not grant anyone the epistemic superiority in assuming that certain segments of the population are politically incompetent and hence unworthy of epistemic respect) and strategic (because theoretically all people have an equal chance in affecting political outcomes), to engage with those supposedly ‘uneducated’ citizens, as well as showing adequate epistemic respect toward their concerns and complaints.

When it comes to specialized knowledge, we have to bear in mind that politics constitutes a realm of uncertainty where we cannot know, ex ante, whether the same group of experts who succeeded in addressing the problems posed by the world of yesterday will have similar success in addressing the problems posed by the world of tomorrow. According to Landemore (2014), ‘there is no epistemic reason to give anyone’s voice or vote more weight in the deliberative and aggregative phase of the decision-making process’ (though decision-making can be delegated to relevant experts when the nature of the task at hand has been identified) because we are just unable to predict what kind of problem our political community is going to face in the future, regardless of how much knowledge of the past we may have acquired (2014b, 174–75). What I want to point out, however, is that such an uncertainty would further increase the epistemic risks carried by epistocratic categorization. Once an epistocracy has labelled particular experts as politically competent and distributed political power to them accordingly, it is likely that they will develop a group consciousness and come to identify specialized knowledge solely as what they as a group have known most in common, even when it turns out to be less effective in resolving the new wave of problems under changed circumstances. In this regard, epistocratic categorization could generate new identity-based stereotypes in favour of the current knowledge elites, but against potential intellectual innovators who may possess specialized knowledge that are different from those whom it now regards as the ‘knowledgeable’, leaving the latter more susceptible to be denied Testimonial Justice and thus less inclined to display Accuracy and Sincerity. ¹² To make the situation worse, such a society may eventually reach a stage of epistemic stagnation, when the size of its knowledge stock could no longer keep growing because a decreasingly number of its members would still be able to exhibit the virtues of truth.

In his comparison of democracy and oligarchy, Daron Acemoglu (2008) argues that democracy is better-off for long-run economic performance. According to him, oligarchic societies, which provide a higher degree of protection to current holders of productive assets by placing additional political power in their hands, also enable those powerful economic elites to create more entry barriers against new groups of entrepreneurs, so that they could maintain a monopoly position for themselves (2008, 3). By contrast, although democratic societies may initially discourage entrepreneur investment through higher redistributive taxes, they allow for a relatively level playing field because of their more equal distribution of political power (Acemoglu 2008, 3). The economic advantage of democracy thus reveals itself when the comparative advantage in entrepreneurship begins to shift away from the incumbents, so that the entry barriers erected in oligarchy become increasingly more costly than the disincentive effects of redistributive taxation in democracy (Acemoglu 2008, 3).

Based on my argument above, I want to suggest that an epistemic analogue of Acemoglu’s thesis could be applicable when evaluating the overall epistemic performance of democracy against that of epistocracy. Epistocracy may at first be able to outperform democracy, since its system of epistocratic categorization has placed additional emphasis on acquiring certain types of knowledge and distributed political power accordingly to people who possess such types of knowledge. However, precisely because of the unequal distribution of political power in their favour, those whom are regarded as the ‘knowledgeable’ are likely to develop some kind of epistemic arrogance, which leaves them less inclined to grant Testimonial Justice to others, and others less inclined to show Accuracy and Sincerity in return. As a result, the epistemic advantage of epistcoracy gradually disappears and may eventually turn out to be a disadvantage, when the types of knowledge prioritized under a system of epistocratic categorization become ‘entry barriers’ to the social aggregation and production of other, new, perhaps more useful and relevant, types of knowledge. In this regard, democracy is more likely to be epistemically superior to epistocracy over the long term.

Conclusion: The wisdom, power and virtue of the multitude

In concluding this article, I want to move beyond Williams and Fricker and consider two broader implications which my epistemic account of democracy based on their virtues of truth has for democratic theory. The first is that it helps to unite different instrumentalist approaches to democracy. The epistemic approach has never been the only or even the main approach that seeks to defend democracy on instrumental grounds. The other, perhaps more influential, strand of instrumental justification, advocated mostly by neo-republicans and neo-Schumpeterians, claims that democracy should be conceived primarily as a political arrangement that helps to resist various forms of elite domination (Green 2009; McCormick 2011; Pettit 2013; Shapiro 2016). ¹³ On the surface, it seems that those two approaches could not be more different from each other. While one focuses on how to produce superior substantive outcomes, the other deals with the need to manage power relations through political institutions. Indeed, so far as most epistemic democrats choose to base their proofs of collective wisdom on some mathematical theorems and formal results, they have been rightly criticized for not only ignoring the complex dynamics of political power, but also giving additional credit to the ‘epistemic frame’ long preferred by supporters of epistocracy (Bagg 2018). However, I do not think my epistemic argument for democracy could be subjected to such criticisms. Instead, it breaks the false dichotomy between the politics of epistemic reliability and politics against domination by placing the social production and aggregation of knowledge firmly within the context of political power. Democracy has epistemic value for a political community precisely because its commitment to political liberty and political equality help to reduce the epistemic costs associated with the problem of untruthfulness in autocracy and the epistemic risks in epistocracy. In other words, the power of the multitude serves, at least in my account, as an essential precondition for the wisdom of the multitude.

Second, by exploring the virtues of truth required for the social production and aggregation of knowledge, my argument helps to bridge the epistemic dimension of democracy with its ethical dimension. Because of their excessive focus on collective wisdom, epistemic democrats have also been criticized for obscuring a more promising Aristotelian approach to democracy that democratic authority rests not on the pooling of diverse knowledge, but on the aggregation and amplification of moral and intellectual qualities (Balot 2016; Cammack 2013; Frank 2005; Lane 2014). As Daniela Cammack (2013) has pointed out, ‘the great interest of Aristotle’s account is that he does not base political authority on the contingent possession of potentially useful information, but rather on a range of ethical and intellectual capacities that, given the right training and external conditions, can be widely shared’ (2013, 194). Yet, the epistemic account of democracy which I have offered in this article actually shows that the virtue of the multitude tends to go together with the wisdom of the multitude. Epistemic decision-making has never merely been about just the pooling of diverse knowledge. Instead, it requires people to first display Accuracy, Sincerity and Testimonial Justice, all of which might be regarded as morally and intellectually commendable qualities in themselves. To the extent that democracy could have full access to all potentially useful information and thereby outperform other non-democratic alternatives, it is because it has successfully created the social and political conditions under which most of its citizens can share those morally and intellectually commendable qualities. Moreover, my epistemic case against epistocracy could as well be interpreted as an ethical one that it is an epistemically flawed regime because it leads to the amplification of intellectual vices such as epistemic arrogance. In this regard, an epistemic democracy also turns out to be a democracy of virtue.