Estlund on Epistocracy: A Critique

Abstract

An influential anti-democratic argument says: ‘(1) Answers to political questions are truth-apt. (2) A small elite only—the epistocrats—knows these truths. (3) If answers to political questions are truth-apt, then those with this knowledge about these matters should rule. (4) Thus, epistocrats should rule.’ Many democrats have responded by denying (1), arguing that, say, answers to political questions are a matter of sheer personal preference. Others have rejected (2), contending that knowledge of the true answers to political questions is evenly distributed. David Estlund finds neither of these replies conclusive. Instead, he attacks (3) arguing that there can be no agreement between qualified people as to who the epistocrats are and that people are not subject to being ruled by experts, whose status as such they can reasonably dispute. Critically, I argue that this argument does not block all forms of epistocratic argument and that Estlund fails to consider the full range of plausible epistocratic views. More constructively, I offer a modest argument for why greater expertise does not necessarily warrant greater political authority. Presumably, the set of feasible options might differ, depending on what procedure is used, and a sub-optimal choice by nonepistocrats from a better set might be superior to the optimal choice by epistocrats from a worse set. In such cases, the mere fact of greater expertise does not warrant political authority, i.e., (3) is false.

Introduction

Epistocracy has prominent proponents. In The Republic Plato famously argued against democracy. According to Plato, there are true answers to questions about how a state should be ruled. These answers, however, are hard to come by and people in general do not have, nor can acquire, the required wisdom. Moreover, they cannot be trusted not to follow leaders who might be good at manipulation, but who have no insight into the idea of the good (e.g., Plato 1983, pp. 282–283). Hence, in Plato’s ideal state political authority belongs not to everyone but to the philosophers. While for perfectionist reasons John Stuart Mill thought it important that everyone had a share in political authority, he concurred with Plato’s scepticism about ordinary people’s ability to grasp political issues.¹ Hence, Mill proposed a scheme according to which people with higher education, whose political views presumably are better than average, epistemically speaking, have more votes than citizens without. While less extreme than Plato’s republic, this unequal, moderately epistocratic distribution of political authority is seen by many as undemocratic and for that reason problematic.

A crucial ambition in David Estlund’s important book Democratic Authority (2008) is to defeat epistocracy and defend democratic authority, while conceding to Plato and Mill the proposition that knowledge of political issues is possible and that, in all likelihood, it is unequally distributed. Estlund’s key premise is that for any purported political expert some people might reasonably resist the suggestion that he or she is a political expert, and that people should not be subjected to the political authority of people whose expertise they can question on qualified grounds. Having set out the epistocratic argument and Estlund’s response to it, I explore the sense in which the former argument is epistocratic as well as the range of possible epistocratic views. In these first three sections I do not directly challenge Estlund’s rejection of epistocracy, but paint a broader and less elitist picture of epistocracy than that conveyed in Estlund’s (as well as many other contemporary) discussions. The next sections, I challenge Estlund’s response to the epistocratic argument. In a nutshell, I argue that the general acceptability requirement does not rule out plural voting or, more generally, unequal political authority. Accordingly, I offer a different argument for why greater expertise does not warrant greater political authority. My argument appeals to the fact that the set of feasible options might differ, depending on what procedure is used, and that a sub-optimal choice by non-epistocrats from a better set might be superior to the optimal choice by epistocrats from a worse set. This possibility invalidates the direct inference from expertise to authority.

The Epistocratic Argument

Any epistocratic political arrangement faces the formidable pragmatic challenge that, over time, absolute power corrupts absolutely. However, it would be surprising if our only objection to epistocracy is pragmatic. Hence, for the sake of argument Estlund assumes that epistocrats are incorruptible and that they will do what they know to be right. I make the same assumption. I set aside what many readers will see as the most powerful and pragmatic objection to epistocracy, not because I think it is bad, but because Estlund’s and my focus lies elsewhere.

Suppose that there are true answers to questions about which political decisions ought to be made where the truth of these answers is not determined by the outcome of a democratic procedure. In such a case, it may seem as if democracy is justified, only if democracy is the best way to ensure that political decisions reflect these truths. Since democracy implies that people influence which political decisions are made regardless of whether they have knowledge of the relevant domain of truths, it is difficult to justify democracy over the rule of those who have knowledge of the relevant truths, the epistocrats. Estlund presents the epistocratic argument roughly as follows (Estlund 2008, p. 30):

• (1) There are true, procedure-independent normative standards by which political decisions ought to be judged. (The truth claim).

• (2) For any demos, it is true that there is a small group of people—the epistocrats—who know those normative standards better than others and, thus, know better what the decisions that conform to those standards are. (The privileged knowledge status claim)²

• (3) For any demos, if it is true that the epistocrats know those standards better than others, then these people should have political authority over others. (The authority claim)

• (4) Thus, for any given demos, epistocrats should have political authority over others. (The epistocratic conclusion)

The conclusion of the epistocratic argument is generally considered anti-democratic and, as it is presented here, the argument appears valid. Democrats will either have to argue that one of the premises is false, or they must argue that, on reflection, the conclusion can be seen not to be anti-democratic after all.

One common response among democrats has been to deny the truth claim, i.e., (1). But for reasons brought out by Estlund the response is problematic. Essentially, there are different notions of truth and a simple disquotational account of truth will do for the purposes of the epistocratic argument. In this view, the statement that, say, ‘Uttering political dissent ought, morally speaking, to be legal’ is true if, and only if, uttering political dissent ought to be legal, morally speaking. Even non-cognitivists about morality should allow for truth in that minimal sense. To deny it, one would have to embrace the thoroughly unattractive idea that ‘there are no appropriate standards (not even minimally true ones) by which to judge political decisions’ (Estlund 2008, p. 25). Indeed, in the present dialectical setting it is precisely to rescue democratic normative standards that we attack the premises of the epistocratic argument, but that cannot be done by appealing to a premise that does away with normative standards across the board.

Another response is to deny that there is a small group of people who know better than others what the true, political answers are, i.e., the privileged knowledge status claim. However, in the light of how weak (2) is—for instance, it does not involve an assertion that ‘some elite has privileged infallible access to a Platonic realm of absolute truths’ (Estlund 2008, p. 31)—and the fact that some of the main reasons for resisting (2) are flawed—e.g., appeals to the fundamental moral equality of people (Estlund 2008, p. 32)—Estlund grants the epistocrat this premise (Estlund 2008, p. 33). I shall follow him in doing so and add that (2) is quite a reasonable assumption to make, too. Claims about certain factors, e.g., literacy, being such that they boost political knowledge are plausibly and commonly made. Such claims seem to commit one to rank the average political knowledge of one group higher than that of another when these factors are present to different degrees and there are no other relevant differences between the two groups.

Estlund resists the epistocratic argument by rejecting the authority claim, i.e., (3). Crucial here is the so-called ‘expert/boss fallacy’, that is, the inference from the fact that some people know better than others what ought to be done to the claim that this warrants that those people should have political authority over those with lesser knowledge.³ Why can we not infer authority from better knowledge then? Here, Estlund’s argument rests crucially on:

The general acceptability criterion: ‘(N)o one has authority or legitimate coercive power over another without a justification that could be accepted by all qualified points of view’ (Estlund 2008, p. 33).

This criterion blocks the inference from greater knowledge to authority because those who have less knowledge than others, but hold qualified points of view, might reasonably deny that those who as a matter of fact know better do know better. Hence, even though you know what the true religion is, assuming there is one, and I do not and that makes you give all the right answers to political questions and makes me give all the wrong ones, it could still be the case that you could give me no justification for your views on what political decisions should be made that I have reason to accept given my false, but qualified, point of view. As Estlund puts it: ‘any particular person or group who might be put forward as such an expert would be subject to controversy, and qualified controversy in particular’ (Estlund 2008, p. 36). From this claim and the general acceptability criterion Estlund infers:

The thesis of no invidious comparisons: ‘No invidious comparisons among citizens with respect to their normative political wisdom can pass the appropriate general acceptability criterion (yet to be specified) of political legitimacy’ (Estlund 2008, p. 36).⁴

One could accept the general acceptability criterion and yet deny the thesis of no invidious comparisons, for example because one thought that certain claims about how some citizens are superior to other citizens with respect to their normative political wisdom can pass the appropriate general acceptability criterion.

In effect, Estlund considers such a challenge to the thesis of no invidious comparisons and rejects it in connection with his discussion of a Millean differentiated voting system where people who are better educated in ways that make them more likely to hold more correct views on political matters get more votes than others. He points out that it is hard to deny that education has this effect, once we favour better education for the population as a whole, because we think that this will improve the epistemic qualities of political decisions within that community.

In response to this challenge, Estlund contends that we might agree that education boosts people’s knowledge of the right political decisions and yet deny that those people who in fact have these epistemic quality enhancing features, such as a higher education, belong to the group of people who collectively knows best. The reason is that the features might come together with other features that function as epistemic quality spoilers. So people who are better educated might systematically be people who belong to the wealthiest parts of society, as is in fact the case, and this might bias their views such that even if they are more politically knowledgeable as a result of their higher education, a group of highly educated people, who are thus more wealthy than people are on average, would collectively know less well what the true answers to political questions are (because of the presence of counterbalancing biases) than would a group that consists not wholly of people with a higher education. Estlund calls this kind of argument the demographic argument. The defeat of Mill’s plural voting scheme shows that ‘the prospects for any form of epistocracy is in very serious doubt’ (Estlund 2008, p. 222).

This completes my presentation of Estlund’s anti-epistocratic argument.

An Epistocratic Argument?

Before I proceed with my assessment of Estlund’s response to the epistocratic argument, I want to consider the extent to which the impulse underlying the epistocratic argument really is elitist. Estlund’s discussion conveys an almost passionate, but seriously misleading, image of epistocracy as a matter of a tiny group of guardians ruling in a rather autocratic manner over everyone else: ‘Epistocracy is authoritarian, not metaphorically but literally, in advocating a form of elite rule. Advocates of democracy, and other enemies of despotism, will want to resist the case for epistocracy’ (Estlund 2008, p. 31). In this section and the next one, I show that there is a broad spectrum of epistocratic positions and that there is nothing inherently elitist or despotic about epistocracy. This point is important, because it locates the discussion of the epistocratic in its proper argumentative setting, which is not the friends of epistocracy versus ‘enemies of despotism’-setting suggested by the Estlund passage just quoted.

Consider again the authority claim. There are various ways in which one might defend it. For instance, one might think it is motivated by the putative fact that an individual’s higher political competence gives this person a claim right to have political authority over others. Yet, this would not be the most natural way to justify the authority claim. Rather, the most attractive line of argument is one that holds that political authority should be distributed in such a way that political decisions converge as much as possible with the correct political decisions and that in a situation of unequally distributed political expertise, the best way to achieve this is to give political authority to those who individually know better than others what these decisions are, i.e., the epistocrats. But, if so, everything hangs on the partly empirical assumption that those who individually know better together form a group of people which collectively knows better than all other possible groups of people what the true answers are. This comes out in Estlund’s discussion of Aristotle’s response to Plato’s point: namely, that the group of, say, the wisest and the next wisest may well know better than the smaller group consisting only of the wisest what the true, political answers are. Indeed, there is nothing, logically speaking, that prevents it from being the case that the group of people that collectively knows best what the right decisions are has none of those individuals who, individually speaking, knows best, i.e., no epistocrats, as its members. This might be the case, if, say, those individuals, who have less knowledge, are more inclined to listen to and learn from others than those individuals, who, considered by themselves, know best, are inclined to do. Accordingly, the premise motivated by the underlying impulse of the argument is:

(2*) For any demos, it is true that there is some group of people (a subset of the demos, a group of people from some other demos, or a mixture of the two) which is different, extensionally speaking, from the demos and which collectively knows those normative standards better than other groups and, thus, knows better what the decisions that conform to those standards are. (The revised privileged knowledge status claim)⁵

(3*) For any demos, if it is true that there is some group of people which is different, extensionally speaking, from the demos and which collectively knows those normative standards better than all other groups, then this group should have political authority over others. (The revised authority claim)

This is significant. It brings out that the basic thought underlying the epistocratic argument is not epistocratic in a pejorative sense as it might justify non-elitist forms of government. Insofar as those who are individually most unknowledgeable about true answers to political questions are those who, when put together, will make decisions that best correspond to these answers they should rule (Hong and Page 2004; Zollman 2010).

The Full Range of Epistocratic Arguments

As Estlund presents the epistocratic argument it is portrayed as one that motivates an extreme degree of inequality in political authority (Estlund 2008, p. 31). All political authority rests with the epistocrats, who form a tiny elitist group (whether it consists of those few individual knowers who know best or not). The alternative is thought to be a democratic and equal distribution of political authority. But in between these two views there is a range of intermediate positions that favour moderately unequal distributions of political authority. These I label (somewhat tendentiously) moderate epistocratic views and each of these might be defended through a refined version of the epistocratic argument. I now want to consider some of those moderate epistocratic views. First, one might refine the privileged knowledge claim to accommodate the fact that which group knows best might vary across different political issues:

(2**) For any demos and any political question, it is true that there is some group of people which is different, extensionally speaking, from the demos and which knows the normative standards pertaining to this issue better than others and, thus, knows better what the decisions that conform to those standards are. (The issue-sensitive privileged knowledge status claim)

This view allows that it is true of all members of a community that, for some political issues, they belong to the group of individuals who know the true answers to the relevant question best. If so and if we implement an issue-focused political decision procedure, it is true of anyone that there are some issues on which she belongs to a group of people who have political authority over others and other issues on which she belongs to a group of people over which another group has political authority with regard to the relevant political issues.⁶ One might imagine a weighted-voting scheme in relation to referenda where, on a case by case basis, elected representatives decide that the votes of different groups of individuals carry more weight when there is reason to believe that they have better knowledge of the pertinent issue, thus the votes of judges might count for more in relation to criminal justice proposals and the votes of teachers might count for more in relation to proposals on educational policies.

Second, the privileged knowledge claim might be formulated in such a way that it accommodates the fact that which group knows best varies across time:

(2) For any demos and any given time, it is true that there is some group of people which is different, extensionally speaking, from the demos and which knows the normative standards better than others and, thus, knows better what the decisions that conform to those standards are. (The time-sensitive privileged knowledge status claim)

This view allows that it is true of all of us that we, at some point in our lives, are subject to the political authority of others. I suspect that most people (including Estlund) would endorse some time-sensitive privileged knowledge status claim as reflected in the fact that most people think that children should be subject to the political authority of adults. Incidentally, this case shows that many people endorse the present moderate form of epistocracy even if they do not think of themselves as epistocrats.

Third, rather than making a claim about which groups have a claim to a privileged knowledge status, one could make a claim about small groups that have an underprivileged knowledge status, i.e.,:

(2****) For any demos, it is true that there is some large group of people (a subset of the demos, a group of people from some other demos, or a mixture of the two) which collectively knows those normative standards better than the demos as such and, thus, knows better what the decisions that conform to those standards are. (The underprivileged knowledge status claim)

This view is worth considering, because even if one could make a case for denying that, for any demos, there is a tiny group of people who have better knowledge of normative standards than any other group, it still could be the case that, plausibly, there are groups slightly smaller than the entire demos, e.g., the demos as a whole except for the Nazis, that know better than the demos as such what the relevant answers are. This might be used as a premise in an argument for why most, but not all, people should have equal political authority.

There might be several reasons why the underprivileged knowledge status claim could be true. First, it might be that some individual members of the demos hold epistemically deficient views on politics, for example they are uninformed about relevant empirical matters. As already indicated, some such view seems to underpin the widespread view that children should not have the right to vote. Second, it might be the case that some people, even if their credentials as individual knowers are no worse than average, perhaps even better, spoil the epistemic division of labour within a group such that the group will know better without these individuals.

I have presented three dimensions along which one might imagine moderate epistocratic positions.⁷ The general upshot is the following: since each of these positions may motivate a modest denial of the democratic view that political authority should be equally distributed and since each of these positions seems much more plausible than the extreme Platonic view that all political authority should belong to a tiny elite, the democratic view of political authority seems harder to justify than otherwise (that is, than if we simply contrast it with the Platonic view). Note, in particular, that the first two views are compatible with an equal distribution of political authority, globally speaking, even if they support unequal distributions, locally speaking, e.g., across specific issues and across age. To favour one of the moderate forms of epistocracy expounded here is not to reveal oneself to be a friend of ‘despotism’, as Estlund’s framing of the issue of epistocracy suggests. With these points in mind, I return to Estlund’s rebuttal of the epistocratic argument, as it was initially presented, offering two reasons for thinking that it does not undermine any possible epistocratic position.

Denying the Demographic Argument

Consider the demographic argument again. According to this argument, differential voting rights and, thus, differential political authority favouring the literate or those with a higher education should be resisted on the ground that while these features promote political knowledge, they go together with other features, for example high income status or being a member of a particular racially biased group, such that differential political authority will generate worse political decision, epistemically speaking.

No doubt, the kind of correlations that Estlund points to might well obtain and when they do, they tend to lead to worse political decisions and, thus, undermine the case for epistocracy. So to that extent I accept the demographic argument. But there is a flipside to the claim that certain correlations undermine the quality of political decisions. For just as higher education might be correlated with features that spoil the epistemic quality of a decision process involving differential political authority, it might also be correlated with epistemic quality boosters. For instance, suppose that a racially unbiased minority tends to be overrepresented among people who obtain a higher education, or suppose that members of the racially biased majority who have a higher education tend to be less racially biased than other members of that majority. Under these circumstances Estlund’s appeal to the demographic consideration actually strengthens the case for the kind of epistocracy which giving more votes to people with higher education involves. My main concern here is to make the conceptual point that the sort of consideration to which Estlund appeals might in some possible cases support, rather than undermine, Mill’s plural voting scheme. Hence, I have no stake in asserting that education decreases racial prejudice, nor in denying that the literacy condition on voting in some southern US states was bad democratically speaking (as well as in other ways).⁸ Whatever the merits of this particular case, it is very unlikely that all epistemic quality boosters, e.g., higher education, will always be correlated with a set of features that, all things considered, counterbalance the positive effects of the quality booster.

One might appeal to the notion of qualified points of view and claim that any suggestion as to what counts as epistemic quality boosters is bound to be rejectable from some qualified points of view, while no such claim is true of quality inhibitors. This response, however, seems flawed for the simple reason that if something boosts the epistemic qualities of a political decision process, for example literacy, then the absence of the booster, that is, illiteracy, is a quality spoiler. Accordingly, if some claim about quality boosters is not rejectable from any qualified point of view, then an analogous claim applies to the corresponding quality inhibitor.

Another line of reply appeals to the view that ‘all Estlund needs to defeat epistocracy is that someone can qualifiedly deny that the educated are as a group more likely to vote correctly than the population as a whole, in which case the extra authority of the educated over the uneducated cannot be justified to that person, and we must default to whatever we take to be the most accurate equal authority scheme.’⁹ However, this reply does not defeat the present objection. It might be clear that there is more reason to think, all things considered, that the educated minority is likely to vote correctly in view of its being much less biased than the uneducated majority—this assessment might, say, be supported through psychological surveys mapping the distribution of biases across groups individuated on the basis on education—in which case citizens cannot raise a qualified objection to differential political authority when informed about the relevant facts.

A third reply on Estlund’s behalf consists in saying that all that the demographic argument is intended to show is that ‘there may often be no adequate grounds for disqualifying’ an objection to a scholocratic voting system where voting rights favour those with a higher education—specifically, the argument is not intended to defend an equal voting system—and for there to be such an objection, it is enough that ‘there may remain important sample errors [even after we have tried to correct for the way in which higher education or literacy goes together with other features] of which we are unaware’ (Estlund 2003, pp. 64–65 my emphasis; cf. Estlund 2008, pp. 217–219). Following Estlund, I shall refer to this objection as the appeal to conjectural features. This objection comes in two versions. In its pure version, it simply appeals to the mere logical possibility of unknown, important sampling errors. In its impure version, it appeals to the fact that on previous occasions it turned out that people were unaware of unknown, important sampling errors and the fact that we are unable to show that our present epistemic situation is relevantly different from those previous ones. Both versions of the appeal to conjectural features differ from the demographic argument in that they do not provide any evidence for higher education being correlated with any particular factors that undermine the quality of political decisions made in a way that gives more weight to the votes of people with higher education. The pure version simply points to the sheer logical possibility that some such correlation may exist. The impure version points to our rather thin evidence for asserting its existence that derives from the fact that in the past such correlations existed without people having evidence that they did.

The appeal to conjectural features in neither its pure, nor its impure, version will do, however. I shall concentrate on the impure version for three reasons. First, this is the version Estlund canvasses (Estlund 2008, p. 218). Second, if the appeal to conjectural features in its impure version fails, then, a fortiori, so does it in its pure version. Third, the mere fact that it is logically possible that a scheme is flawed is not enough to show that, as a matter of fact, it is flawed. The objector to epistocracy needs some reasonable reasons for doubting that, actually, the epistocratic system is epistemically superior to democracy, and the conjectural claim in its pure version provides no such reason.

To explain why the appeal to conjectural features in its impure version fails, I need to say something about qualified objections. In my view, an objection to a certain scheme, p, is not qualified if the person who puts it forward concedes (or ought to concede) that (i) it has to be the case that p or q and (ii) there is no reason to think that an objection analogous to the one that applies to p does not apply to q.¹⁰ Call this the coherence requirement on qualified objections. The argument for the coherence requirement is that if there is a requirement that political authority is justifiable to the ruled, it must be possible for a scheme of political authority to satisfy this requirement and this is impossible in cases where an objection applies to any scheme of political authority that the objector favours or that is possible.

Applying the coherence requirement to the impure appeal to conjectural features disqualifies Estlund’s objection to a differentiated voting scheme based on conjectural features. Thus, for example, that even though, all other things being equal, literacy promotes epistemic qualities of political decisions and even though we have neutralized any source of reduction of epistemic quality due to the fact that literacy is correlated with such features, we cannot rule out the possibility that it is correlated with epistemic quality reducers of which we are presently unaware.¹¹ The coherence requirement rules out disqualification on this ground, because anyone who concedes that there must be some sort of voting scheme might with equal reason reject an equal voting scheme. Even if we have no reason to affirm that the majority is biased against a particular minority (or just some minority or other) in a certain way relative to the relevant smaller group of epistocrats, we cannot rule out that they are so and must concede that there have been cases in the past where the demos in general turned out to have been more biased than an educated minority. But then an equal voting scheme is vulnerable to a qualified objection (from members of the relevant minority, from people who entertain the thought that they belong to some minority which the majority is biased against, or simply anyone) if an unequal voting scheme is vulnerable to qualified objection on grounds of impure conjectural features. Since voting schemes must be either equal or unequal, it follows that the appeal to conjectural features in its impure version (unlike the demographic argument) violates the coherence requirement on qualified objections.

Finally, by way of further support for my scepticism about the strength of Estlund’s anti-scholocratic argument, note that age is certainly correlated with factors that might lead one to be sceptical about the epistemic qualities of a political process that excludes a certain age group.¹² Disenfranchising non-adults, for instance, might lead to the overfunding of healthcare for Alzheimer’s and similar diseases that primarily strike in old age, lax speed limits for cars, alcohol policies that do not adequately address violence against and abuse of children, child-unfriendly work regulations, and underfunding of kindergartens and schools. Hence, if the demographic objection applies anywhere, it also applies here. Yet, many, including Estlund it seems, would not see this as sufficient to undermine any epistocratic argument against extending voting rights to non-adults.¹³ But then the question arises as to why this argument should be thought to be undermined in other cases by the sort of considerations invoked by the demographic argument. Note also that to defeat the present Estlundian objection to disenfranchising children, it is not enough to point to the various reasons we have for believing that adults are better judges than children of what is in the latter’s best interest. The existence of these reasons is compatible with there being a qualified, impure conjectural objection on Estlund’s account to the unequal voting rights of adults and children. We know that in the past people have held there to be reasons why some individuals are better judges of what is in the best interests of others—to witness, men have been thought to be in such a position relative to women—and this view is now known to have reflected various biases rather than the facts (and we cannot point to any relevant epistemic differences between our present situation and the one that obtained in the past).

The Limits of Estlund’s Rejection of the Authority Claim

In this section I want to show that given Estlund’s commitments, he must allow that the epistocratic argument is successful in some cases at least. Consider the notion of qualified points of view. Estlund does not offer anything other than a formal characterization of such views: people who hold qualified points of views hold views such that, even though these might be false, they are not ones that disqualify their holders from only being permissibly coerced politically when that arrangement can be justified in terms that do not contradict their views. Thus, he ‘leaves it [largely] open who should count as qualified’ (Estlund 2008, pp. 40–63). However, Estlund does give some examples. His argument against Mill’s plural voting system is based on the substantive assertion that objections to this scheme based on conjectural biases are not disqualified (Estlund 2008, pp. 217–218). Also, he thinks that it would be an odd amoral view to require that justifications of political authority are ‘acceptable to everyone’, since it would imply, for instance, that ‘otherwise sensible lines of justification are unavailable if they are not acceptable to Nazis’ (Estlund 2008, p. 4). This indicates that Estlund thinks that a Nazi point of view is not qualified.

Suppose that, regrettably, we live in a state where almost everyone is a Nazi (or almost everyone holds a relevantly similar unqualified point of view) and a few enlightened people are not. The latter correctly believe themselves to have better knowledge of political answers than the majority of their fellow citizens. The Nazi-majority, of course, rejects this claim, yet it seems that with Estlund we can disregard this, since this particular invidious comparison among citizens disfavouring the normative political wisdom of Nazis pass the appropriate general acceptability criterion as Estlund himself understands it. Ex hypothesis, the general acceptability criterion does not imply that no one can have authority over Nazi citizens without this being justifiable from a Nazi point of view and, accordingly, we cannot appeal to the general acceptability criterion to rule out epistocracy here.¹⁴ My point here is not that it speaks against the general acceptability criterion that it fails to rule out political authority of the tiny minority with a qualified view in this case. My point is that Estlund’s criterion does not rule out epistocracy across the board.

In the scenario just described, the epistocrats might not be particularly knowledgeable relative to normal standards, but they are relative to their Nazi co-citizens. Incidentally, this case raises an interesting issue, which Estlund does not address.¹⁵ This issue is whether one should see the acceptability criterion as a non-comparative one—such that a view is qualified regardless of which other views are present within the relevant demos—or as a comparative one—such that views are not qualified per se but are qualified relative to one set of points of view, but perhaps unqualified relative to another set of points of view. A somewhat artificially precise illustration of the latter view would be a criterion according to which one holds a qualified view if it is more than half as qualified as the averagely qualified view. It seems to me that both possibilities spell problems for the anti-epistocratic employment of the general acceptability criterion and that together the two options exhaust the space of possibilities.

Suppose first that we understand the acceptability criterion as being non-comparative. In that case, it is possible for all views in a political community to be unqualified, since none passes the non-comparative threshold of having a qualified view. Still, this would not rule out the possibility that among those unqualified views, some are much more unqualified than others. Epistocrats would plausibly want those with the least crazy unqualified views to rule, and Estlund’s general acceptability criterion read as a non-comparative criterion would not rule out epistocracy in such a case.

Suppose next that we understand the acceptability criterion as comparative. In that case it might block justification of epistocracy in a case, where half the people are Nazis and the other half hold a view that is slightly more unqualified. But now a different problem arises. For, presumably, we could imagine cases where the views present in a community are all what we would normally think of as acceptable, non-comparatively speaking, but one view is a lot more qualified than the others, comparatively speaking, thus, thereby rendering the other views unqualified according to the comparative conception. If so, again the general acceptability criterion read as a comparative criterion would not rule out epistocracy in this scenario either.

I conclude this section by noticing that appeals to the general acceptability criterion are not as effective in blocking the epistocratic argument as Estlund wants them to be. Estlund might respond that in the extreme sort of cases, to which I appeal, where the majority has an unqualified view, he does not want the epistocratic argument to be blocked and has no interest in defending democratic authority. Such a concession comes with a price, however. First, it means that Estlund is an epistocrat of a sort such that his disagreement with other epistocrats really is a family disagreement. Second, to settle it in his favour Estlund would then need to say much more about why other well-known members of the epistocratic family have a much too restrictive view of which individuals hold a qualified point of view. At the moment, he says almost nothing substantive about this issue. Finally, it would be even harder to see how he can resist some of the less well-known epistocratic theories which I presented earlier. For instance, it is hard to see how he can deny the rather modest issue-sensitive privileged knowledge status claim ascribing a qualified point of view to people who, say, lack any empirical information on the relevant issue, when for some political majorities he asserts, more radically, that the majority is unqualified.

An Alternative Reason to Resist the Authority Claim

Having challenged Estlund’s depiction of epistocracy and his anti-epistocratic argument, I now sketch an alternative and modest response to the epistocratic argument. Unlike Estlund’s argument, this argument is supposed to block the inference from superior knowledge to authority even under circumstances in which some people are identifiable as political experts from all qualified points of view. Accordingly, while the present line of argument is different from Estlund’s, it is not incompatible with his overall line of argument. Also, the argument does not speak to the issue of how often democratic authority is justified, even if we grant the first two premises in the epistocratic argument. Finally, the argument allows that there might be other conclusive anti-epistocratic arguments, e.g., the pragmatic one set aside earlier.

The first premise of the argument is that a situation in which epistocrats rule and make the best political decisions available of those available to them is inferior on the relevant evaluative dimension to a situation in which everyone rules and makes political decisions that are not the best among those available to them. This might be the case because the set of possible decisions available to epistocrats may differ from the set of possible political decisions available to a democratic people. For instance, if those people who know less well will resist the decisions of epistocrats but willingly accept being outvoted in a democratic decision procedure, that might eliminate from the feasible set of decisions and outcomes the one in which epistocrats make a certain decision which is unanimously complied with and implemented by both epistocrats and non-epistocrats alike. Similarly, it might be the case that people are more motivated to do their share in implementing a political decision, when they have participated in making it, or at least had the opportunity of doing so.¹⁶ This motivational constraint might not reflect any conscious judgments about or preferences over political decisions-procedures. Indeed, in principle this motivational constraint may obtain even if people prefer that those whom they recognize as epistocracts rule.¹⁷

The second premise in the argument submits that if the situation described in the previous paragraph is possible, then, possibly, it ought not to be the case that epistocrats rule. One should grant this premise, if one accepts the following two assumptions: (i) political authority should be distributed in such a way that the best political outcomes are realized and (ii) there is nothing better per se about fewer people rather than everyone ruling. The latter assumption is uncontroversial, and while one might deny the former on a number of grounds, for example a concern for political equality, it is acceptable to those who side with Estlund in wanting democratic authority to be grounded on a concern for ‘the quality of political decisions’ (Estlund 2008, p. 1).

Given the third and final premise that if, possibly, it ought not to be the case that epistocrats rule, then, possibly, epistocrats have no authority to rule, it follows that epistocrats may have no authority to rule. Friends of the epistocratic argument should grant the third premise. For presumably the reason that knowledge of the right decisions warrants political authority is that it is better if the right decisions are made than if they are not. If it were acknowledged that it would be worse that the right decisions, i.e., those that are best of those possible, were made, making the right decisions could hardly ground a claim to political authority. But then being an epistocrat does not warrant authority when it would be better for non-epistocrats to make political decisions.

This argument implies neither that epistocratic authority is never justified, nor that it is unlikely to be justified. My modest, but nevertheless significant, claim in this section is that we can accept that political authority should be based on the concern for the epistemic quality of political decisions accepting the existence of a group of people who form an epistocratic elite, and still consistently deny that epistocratic authority follows. But as my reference to participatory democracy above shows, I do not think the circumstances under which epistocracy would not be justified are particularly unlikely to obtain.

At this point it might be objected that since my argument relies on the claim that the best option available to epistocratic rulers may be worse than the sub-optimal option selected by non-epistocrats, it only shows that due to the non-compliance of some democracy is preferable as a second best. It would be better if non-epistocrats complied with the decisions of the epistocrats and were fully motivated to do their part in implementing it. I concede that this is indeed the case, but, first, it does not serve as an objection to my argument in the present dialectical setting, since Estlund must make a similar concession.¹⁸ In a sense it would be better if non-epistocrats adopted a view such that they would not object to the privileged knowledge position of the epistocrats, because then political decisions satisfying the general acceptability condition would be better. Hence, like me Estlund is engaged in theorizing about non-ideal circumstances.

Second, part of my point does not rest on people not cooperating to bring about the highest ranking outcome. To the extent that lack of motivation is the problem—people not involved in making political decisions just cannot, even with all the will-power they can muster, bring themselves to go to the same lengths in implementing them as they could had they played their part in making them—non-epistocrats may fully accept the political authority of epistocrats and want to comply with whatever they are required to do. It is just that they cannot bring themselves to make the effort they could, if they were somehow involved themselves in making the political decisions.¹⁹ It might be worth adding that the line of thought I suggest here—that participation increases engagement—is very influential among people who favour participatory democracy and has a respectable history that goes back at least to John Stuart Mill. By way of further support for the claim that the situation I appeal to need not involve non-compliance by anyone, consider the fact that it might simply involve a case of reasonable disagreement, i.e., one in which it is not unreasonable to deny the epistemic superiority of the epistocratic procedure. In this case, people might be doing their best to figure out what is justified, but reasonably disagree, and given this disagreement, on Estlund’s view procedures involving some having extra authority over others are not justified, because there is reasonable disagreement about whether these extra authority procedures are in fact epistemically superior.²⁰

Finally, some ascribe intrinsic value to people being subjected to laws that they have democratically decided in favour of (or some other relevantly similar form of collective autonomy of citizens, see Anderson 2008, p. 136). Epistocrats know that it would be best if certain laws were implemented. They also know that unless they and only they decide the matter, a different and suboptimal set of laws will be selected through the democratic election procedure. But since, ex hypothesis, it is intrinsic valuable that people are governed by laws democratically selected, epistocrats know that the choice is between democratically selected suboptimal laws and epistocratically selected optimal laws and that in that choice the former is better because the sub-optimality of the laws is counterbalanced by the value of democratic self-government. Consider the following analogy at the level of persons: many think that people should have authority over how they lead their own lives, even if they suspect that others could make decisions for them which would have a better epistemic quality. However, it has non-instrumental value that people run their own lives which may outweigh the loss of epistemic quality of the decisions they make. Similarly, one might think that it has non-instrumental value that present members of the demos run their own affairs such that even if the decisions reached by a previous generation of members of the relevant demos or present epistocrats belonging to other demoi have better epistemic qualities, all things considered, it is better that a demos runs its own affairs. Still, I cannot rest much on this last point. Estlund thinks democracy should be justified independently of procedure-based values and, in the present argumentative context I want to similarly restrict my justificatory resources. The interesting point is that even if we grant Estlund’s restriction, the epistocratic argument still is unsound, because the authority claim is false.

Conclusion

Estlund’s attempt to block the inference from knowledge to authority through the general acceptability criterion is limited in a number of ways. I have offered a way of blocking the inference that applies even to cases where the general acceptability criterion cannot block it. However, I concede that this argument does not block the inference in all cases—it does not block it in those cases where it actually would be best, if epistocrats ruled and made optimal decisions—but I am not sure this detracts from my argument rather than renders it more plausible. In short, epistocracy may be justified in a range of cases, even if the inference that I have discussed fails to justify it across the board. Whether it is, is something about which this article allows me to be neutral.