What’s wrong with an epistocratic council?

The epistocratic challenge

Many epistemic democrats are uneasy with the possible implications of their epistemic starting point (see, in particular, Estlund, 2008; Martí, 2006). They seem afraid that their reasoning, unless qualified properly and topped with procedural arguments, might lead to the justification of epistocratic institutions, that is, institutions in which those who are more knowledgeable, or wiser, have more formal political power than the rest of the population. Arguably, it is already the case in our democracies that some people enjoy more political power than others. People having a good knowledge of political institutions are more likely to access positions of power than others. And people with a certain expertise or communicative skills generally enjoy more political influence than others. Yet despite these political inequalities – which are not necessarily undesirable (Dworkin, 2000: ch. 4) – democratic regimes are based on a fundamental, formal political equality (one person, one vote) in the selection and retrospective judgement of political representatives. In an epistocracy, this would be the first salient difference: the selection and/or assessment of the rulers, if there is one, would not be grounded in formal political equality.

Two different aspects might differentiate an epistocracy from a democracy in this respect:

The second aspect introduces a distinction between democratic accountability and other forms of accountability, such as accountability to a subset of adult citizens or to an unelected (or non-representative) body. Although epistocratic institutions can include some form of accountability, this will typically not be accountability to the whole population. An epistocratic body could, for example, be accountable to other unelected bodies. The border cases are unelected bodies accountable to elected ones. Some may consider them democratic because, indirectly, they derive legitimacy from an authorized body. Others will consider them as epistocratic. According to this wider definition of epistocracy, institutions such as central banks or the European Commission can be regarded as epistocratic institutions embedded in (more or less) democratic arrangements. In a pure epistocratic regime, though, none of the decision-making bodies (if there is more than one) would face inclusive and egalitarian accountability. Otherwise, they would lose their independence from public opinion. Yet this independence is justified by their higher wisdom or better knowledge and necessary to insulate this wisdom from public pressure.

In addition to these undemocratic aspects, an epistocracy would differ from other non-democratic regimes by some mechanism selecting people who distinguish themselves from the mass by their wisdom or expertise to rule, or at least to enjoy an important degree of political power. A particularly interesting example is Brennan’s (2016: 215–220) idea of an ‘epistocratic council’. Members of this council would be selected on a meritocratic basis, passing a competency exam. And in order to minimize the risks of a biased test, all citizens could have a say and an equal vote in the determination of the required competencies.

What makes this idea particularly challenging is its relative political plausibility compared with other epistocratic arrangements disenfranchising a set of the population or overtly rejecting political equality as in plural or weighted voting. With an epistocratic council, there would be some degree of popular participation and political equality in the determination of the required competencies. There would also be formal equality in the access to political power. Differences in skills would of course translate into unequal real opportunities, but this is also the case with elections. Hence, the main difference with electoral representation is that there would be no institutionalized democratic accountability: during their mandate, epistocrats would have more or less free rein – at least vis-à-vis public opinion. For the rest, electoral representation has also partly been conceived as a way to select the wisest (Manin, 1997). In this respect, another difference with an epistocratic system is that the latter may be considered as more likely to effectively bring the wisest to power. Although people can make wrong choices in their selection of the criteria of wisdom or expertise, they do not run the risk, as is the case with elections, to be fooled by a candidate’s irrelevant personal traits. As Brennan argues, it is much easier to identify the criteria of expertise than the best experts (Brennan, 2016: 224–226). And this is precisely why the selection of the latter should, according to him, be made through a competency exam. Furthermore, this option has the merit of reducing the influence of money on the selection of rulers. Although a competency exam is likely to favour the more educated, and hence (under existing conditions) the higher social classes, it will not be necessary to raise money to make a chance to be selected.

For all these reasons, an epistocratic council might appear more acceptable than other epistocratic alternatives. This is why I will take it here as the most serious competitor to standard democratic institutions. Before comparing these alternatives, however, we need to make clear what we expect from our political institutions. We need standards to assess and compare rival options.

What makes the epistemic value of a political decision?

What we expect from our political institutions is of course a controversial issue. Economic prosperity, civil peace, social harmony, or social justice are possible candidates. Therefore, several theorists somehow hide their opinion behind the unspecified notion of ‘right decisions’. This can be justified by the willingness to start from uncontroversial premises (Estlund, 2008: 169). Admittedly, it would be problematic to assess democracy’s epistemic potential drawing on a controversial theory of justice, democracy being supposed to arbitrate our disagreements about what a just society is (Waldron, 1999). However, an unspecified criterion of right decisions may not bring us very far in the discussion and can also generate ambiguity. Experts can make better decisions in light of some criteria and be outperformed by the many in light of others.

More specification could therefore be helpful. Yet we do not need to agree on a very substantive standard of epistemic quality (a theory of justice or an index of prosperity, for example) to assess the epistemic potential of an institution or regime. No set of institutions or regime can guarantee specific decisions anyway. In Rawlsian terms, pure procedural justice is not possible (Rawls, 1971: 83–90). Thus, all we can try to assess is whether an institution or regime (would) foster(s) general qualities such as rationality, impartiality, or civil peace, which are less controversial candidates for a standard of good decisions.

Here, I will focus on rationality and impartiality. Whatever the ultimate goal that we would like politics to pursue, I think we can all agree without much argument that these two qualities matter. Rationality, understood in an instrumental sense, refers to the capacity of a decision-making system to select the appropriate means for the ends pursued (see also Przeworski, 2018: 80). Whether we want peace, prosperity, or justice, we will all agree that it matters to be able to identify the best ways to pursue our aims.

Impartiality, understood in a moral and substantive sense, as a property of public policies and of a political order, can be simply defined as not favouring some groups or individuals over others for morally arbitrary reasons (Barry, 1995: 7–8; Elster, 2012: 150). The opposite of an impartial policy is a discriminatory or biased one.

Admittedly, most policies have a different impact on different social groups. But this is not what matters here. The idea is that the fundamental interests of no one should receive higher importance than those of others – that everyone should be entitled to the same rights and opportunities, for example. This understanding of impartiality does not forbid group-targeted policies but any form of unfair discrimination or favouritism. Rothstein and Teorell (2008: 170) illustrate this distinction with the example of policies supporting poor families with children: ‘The enactment of such policies would not break the principle of impartiality, while denying such allowances for families from a certain ethnic group or parents with a certain sexual orientation when implementing the policy would’.

Although the means to pursue such ideal of impartiality are extremely controversial, this fundamental moral requirement lies at the core of most modern ethical theories (Dworkin, 2000: 1–2; Rawls, 1971: 5; Sen, 1995: 3–4, 16–19), be they egalitarian, utilitarian, or libertarian. When libertarians argue for property rights, for example, they believe that all citizens are equally entitled to these rights and that it would be unjust for some to have a property right and others not. This is why I will assume here the validity of this standard rather than argue for it.

Nonetheless, two further clarifications are needed. Substantive impartiality thus defined is distinct from procedural impartiality, that is, not favouring any person or voice in the decision-making procedure. Democratic arrangements are procedurally impartial but may produce less substantively impartial decisions than non-democratic alternatives. It also differs from impartiality vis-à-vis people’s preferences, which would also beg the question in favour of democracy. I will assume here that what is just and substantively impartial is potentially distinct from what most people actually prefer. As highlighted by Martí (2006), a standard of good decisions must be at least partly independent from the procedure in order to help assessing the latter’s results.

Existing responses to the epistocratic challenge

In the existing literature on the epistemic value of democracy, the moral dimension of a good political decision – its normative validity – usually receives less attention than instrumental rationality (Vandamme, 2016). The focus is often laid on the cognitive intelligence of ordinary citizens, on their ability to solve problems or make predictions, for example (Landemore, 2013), or on their understanding of economics (Brennan, 2016; Caplan, 2011). Although the importance of morally just decisions is often stressed (Brennan, 2016; Estlund, 2008: 17–18; Landemore, 2013: 11; Martí, 2006: 43–44), the significance of this moral dimension in our assessment of democratic and epistocratic institutions is somewhat neglected, with important implications: it might explain the frequent difficulty to rebut the epistocratic challenge, as I will try to show.

José-Luis Martí, (2006, 2013), for example, argues that deliberative democracy is epistemically superior to alternative forms of democracy, but could be outperformed by epistocratic institutions. Hence, he believes that we also need to acknowledge the intrinsic value of political equality in order to defend democracy. The reason is that under the plausible assumption that some citizens are wiser than others, it is difficult not to think that the epistemic potential of deliberative procedures could be improved by restricting them to the wiser (Martí, 2006: 48).

Similarly, David Estlund (2008: 37–38) adds to the epistemic argument in favour of democracy a ‘general acceptability requirement’: not only should our political institutions be judged on the quality of their decisions, they should also be acceptable to all qualified points of view. This precludes, for example, excluding some adult citizens from the decision-making process, or giving more importance to some voices, and hence protects democracy against epistocratic challenges – yet not necessarily against the epistocratic council challenge.

In contrast with these modest epistemic defences of democracy, Hélène Landemore’s more recent contribution stands out by its ambition, arguing that:

there are good reasons to think that for most political problems and under conditions conducive to proper deliberation and proper use of majority rule, a democratic decision procedure is likely to be a better decision procedure than any nondemocratic decision procedure, such as a council of experts or a benevolent dictator. (Landemore, 2013: 3)

The general argument of her book relies on the epistemic benefits of cognitive diversity in the resolution of problems. This refers to a diversity of ways of representing and interpreting a problem, of generating solutions to it, and inferring cause and effect (Landemore, 2013: 102). It seems that in the resolution of complex problems, the benefits of such diversity often outweigh the benefits of ability or expertise (see also Page, 2007). This empirical finding, she suggests, could speak in favour of maximizing the number of decision makers. In addition to that, she argues, it is difficult to identify in advance the cognitive qualities that we will need in the resolution of the diversity of cognitive problems faced by political decision makers. Therefore, maximizing diversity is preferable to trying to identify a subset of the population combining ability and diversity.

Landemore’s stimulating epistemic argument in favour of democracy has faced a plurality of objections. Some of them concern the mathematical foundations of the ‘Diversity Trumps Ability Theorem’ she relies upon (Houlou-Garcia, 2017) – which I lack the ability to assess. Others argue that the main problem lies in the generalization from the theorem, the latter providing only pro tanto reasons to enlarge the group of decision makers – if all other things remain equal, which is implausible when we move from problem solving in laboratory conditions to real decision-making in large and complex societies. Finally, even if the ‘Diversity Trumps Ability Theorem’ was correct and generalizable, it would still be difficult to exclude the possibility of performing better without universal inclusion. If what matters is competence and diversity, we can imagine maximizing them while excluding some citizens whose inclusion would add nothing in terms of diversity or competence (Martí, 2013). The theorem provides reasons to be inclusive but not necessarily to include everyone or to give equal formal political power to all.

Another possible objection to epistocracy relies on the Condorcet Jury Theorem. The latter is composed of two claims:

First, the majority vote among a group of (independent, competent, sincere) voters, each of whom is more likely to be right than not, is itself more likely to be right than are individual voters separately. Second, as the number of such voters approaches infinity, the probability that the majority among them is correct approaches one. (Goodin and Spiekermann, 2018: 5)

It is often objected to this theorem that although its mathematical foundations are firm, it does not apply to real-world circumstances where votes are not sufficiently independent from one another and voters’ competence is dubious (Brennan, 2016: 179–180; Estlund, 2008: ch. 12). One of the merits of Goodin and Spiekermann’s (2018) recent contribution to the debate is to show how much the conditions necessary for the theorem to work can be relaxed, why it is not far-fetched to believe that people might be on average – and not all – better than random (second condition) and how this makes its application to the real world more plausible. Still, it seems to me that the Condorcet argument for democracy relies too heavily on a condition – average voters being better than random – that is extremely difficult to assess. Furthermore, it focuses too much on the instrumental-cognitive dimension of what counts as quality decisions, whereas their moral rightness seems central to me.

There are thus reasons to think that another argument is needed to strengthen the case for egalitarian inclusion and accountability in political decision-making processes. Must this additional argument be procedural, that is, refer to the intrinsic value of democracy, not its results, as argued by Martí and Estlund? Not necessarily. I want to argue that taking seriously the importance of moral impartiality buttresses the general epistemic justification of democracy. Put simply, the reason is that egalitarian inclusion and institutionalized accountability minimize the risks of morally biased decisions, as I will argue in the next section, assessing the epistemic prospects of an epistocratic council.

Would an epistocratic council perform better than democracy?

Let us recall the basic features of such council: ‘Membership in the epistocratic council is potentially open to all members of society. Citizens may join the epistocratic council only by passing rigorous competency exams’ (Brennan, 2016: 216). And the required competencies could be decided through a popular consultation.

One obvious difficulty is the selection of those in charge of setting up the test and assessing its results. I will leave this issue aside here, assuming that the risks of abuse can be limited. More generally, in order to draw a fair comparison between democracy and epistocracy, I will try to compare plausible standards, each with their qualities and limits. The standard of democratic institutions includes free and fair elections, that is, non-manipulated elections ³ with a plurality of competing platforms or candidates, in a context of freedom of press and expression. The standard for the epistocratic council includes the same criteria of freedom and fairness for the collective vote on the required competencies, and a reliable competency test. However, for both democracy and epistocracy, I take it for granted that rulers are intellectually fallible and morally imperfect, and thus cannot be trusted to spontaneously and continuously behave impartially, which is a reasonable assumption.

Taking the idea of an epistocratic council beyond Brennan’s concrete proposal, we could imagine it with executive, legislative, or controlling power. Brennan argues in favour of the latter: a power to veto any legislation. Let me however leave this question open for a while. I will first examine the general proposal, and then see whether some specific arrangements are more plausible or attractive than others.

The important question, for the purpose of this article, is to imagine the consequences of adopting an epistocratic institution. What would it change regarding the average quality of the decisions produced? Would there be a gain in instrumental rationality? Three conditions would therefore need to be met:

Whether these conditions would be met is debatable. The first is problematic, because it may be difficult to circumscribe precisely what competencies will be necessary to solve complex and heterogeneous political problems (Landemore, 2013: 110–113). The second and third are also problematic because any selection relying on knowledge is likely to exclude some voices and perspectives already disadvantaged (Estlund, 2008: 215–219). And it is hard to ensure that this epistemic loss is compensated. Let us nonetheless assume, for the sake of the argument, that these conditions can be met, and hence that an epistocratic council would offer better prospects of instrumental rationality. It is not sufficient to build a case in its favour, as we also need to take into account moral rightness, previously defined as impartiality.

Do we really need everyone?

There are thus important epistemic reasons to mistrust epistocracy – at least in the form of an epistocratic council. And I believe that they apply to most decision-making procedures privileging some citizens’ perspectives over others’. Are they nonetheless sufficient to ground the case for universal inclusion? To come back to a point made earlier, it seems that depriving one person or very few people from the right to vote would not by itself affect the impartiality of the decisions produced. The risks of biases come from the political exclusion or marginalization of some groups. Even if we believe in a certain dispersion of knowledge and in the epistemic importance of including all possible standpoints, it is hard to believe that, within a large electorate, each and every individual perspective will be an epistemic added value – an input that others cannot offer. And this is particularly the case when votes are the only means of political expression for most citizens and rulers are selected through majority rule. Thus, it is hard to deny that we could include everyone in public deliberations but exclude a few from suffrage without affecting the epistemic potential of the process.

There are two ways of responding to this challenge. First, one can point out that the exclusion of very few individuals from suffrage is an unlikely scenario. In real-world politics, it is always groups that have been excluded, inevitably generating a bias against them. What is more, opening the door to disenfranchisement would more probably lead to the arbitrary exclusion of marginalized or vulnerable groups than to net epistemic gains. As highlighted by Estlund, those with the most epistemic deficiencies are also likely to be groups already suffering other disadvantages (Estlund, 2008: 215–219). Yet if these groups already suffer disadvantages with the right to vote, it is hard to imagine them better-off without it. This could happen only if they appear to be the main responsible party for their current disadvantages, which is difficult to believe. Granted, the most disadvantaged – when they vote – do not always make the best electoral choices in light of what could be perceived as their objective interests, but they are also mostly suffering from the lack of concern – if not frank hostility – from other voters. Thus, even if disadvantaged people are poor guardians of their interests, even if they participate less in politics, it would be implausible in the real world to trust others to take better care of these interests.

There are thus good political reasons to opt for universal inclusion. These can be considered as secondary reasons. Our ultimate concern is epistemic and provides us primary reasons to include a wide diversity of perspectives. Some individual votes might not add anything, yet political or strategic considerations provide us secondary reasons to enfranchise all adult citizens.

We can nonetheless imagine an implausible scenario where the disenfranchisement of very few citizens would have no negative epistemic impact. Hence, the case for democracy might need additional philosophical grounding. There is arguably a somewhat intellectually lazy move from the consideration that we need an important variety of perspectives to the decision to enfranchise everyone. Even if it is hard to imagine what criterion would allow us to disenfranchise some without increased risks of biased decisions, this does not mean that it is impossible.

Hence, an additional – neither epistemic nor strategic – argument might be necessary to make a decisive case in favour of democracy. This is the second line of argument. We could say, for example, following Brian Barry and Thomas Christiano, that justice must not only be done but also seen to be done (Barry, 1995: 111; Christiano, 2008: 8), and hence equality or impartiality should not only be pursued but also publicly affirmed in order to reduce public suspicion. Or, following Estlund (2008), we could argue for a general acceptability requirement. Nevertheless, these procedural arguments should not be considered as sufficient to defend democracy. They should be conceived as supplementary arguments that may help moving from wide inclusion to universal suffrage in a philosophical discussion. However, in debates about democracy’s value, it seems to me that the emphasis should be laid on the epistemic benefits of democracy as minimizing the risks of biases. Otherwise, we send the message that we are committed to democracy however bad it is, that we would defend it even if alternative regimes offered better prospects of social justice. Yet we cannot just build ‘bulwarks’ protecting political equality (Saffon and Urbinati, 2013); we need to be able to justify this choice to those who heavily suffer from current democratic decisions. And in such argumentation, as Christiano (2017) recently recognized, ‘the expressive value piggy-backs on instrumental value’.

Conclusion

In this article, I have defended the view that standard democratic institutions provide better securities against misrule than any form of epistocratic council with decision-making or vetoing power, thanks to egalitarian inclusion and accountability, which respectively reduce the risks of biases and minimally check the use of power. Even in the case where such epistocratic institution would increase the instrumental rationality of political decisions, and even under the strong assumption that the selection process would not be manipulated, it is hard to believe that it would increase substantive impartiality. Assessed against this widely shared standard of social justice, an epistocratic council is thus not appealing. However, although the epistemic argument combined with political prudence seems sufficient to defend democracy, I conceded that we might need to call upon an additional – procedural – argument in order to philosophically reject politically implausible hypothetical examples.

A diversity of other epistocratic rules or institutions have been imagined. John Stuart Mill defended literacy tests and plural voting – the latter being still advocated today (Mulligan, 2018). Beside his epistocratic council, Brennan (2016) also imagines a system (inspired by Scott Althaus and Bryan Caplan) of votes weighted based on people’s objective political knowledge, possibly ‘controlling for the influence of race, income, sex, and/or other demographic factors’ (2016: 15), and he supports Claudio Lopez-Guerra’s (2011) idea of an enfranchisement lottery, where a set of voters is randomly selected among the population and has the opportunity to inform and deliberate properly before electing representatives for the whole. It seems to me that the argument defended in this article applies to these alternatives as well but they also raise specific questions that would require a more careful discussion. For this reason, this article has not offered a full epistemic case in favour of standard democratic institutions, but suggested and tested a way of epistemically assessing competing decision-making systems.

Before finishing, I would like to stress that asserting the epistemic superiority of standard democratic institutions vis-à-vis their plausible alternatives is fully compatible with the conviction that existing democratic institutions are unsatisfying and should be reformed with the aim of improving their epistemic potential. All the reforms that could increase citizens’ social and political knowledge and reduce the risks of biases are welcome (see Goodin and Spiekermann, 2018: ch. 6–9). The epistemic justification of democracy should thus not be interpreted as an ideological justification of the current state of affairs. It is rather grounded in a historically informed and epistemologically based mistrust against alternative regimes. As David Runciman (2018) puts it, ‘sometimes it is more important to avoid the worst’.