Estlund Reading Group Chapter 11

In this chapter Estlund asks the question whether an ‘epistocracy of the educated’ — whether, as J.S. Mill recommends, the educated should receive more votes than the uneducated — could satisfy the ‘qualified acceptability requirement’, that is, be a political principle to which no qualified objection could be levelled. Most epistocratic proposals are defeated because they could not satisfy the qualified acceptability requirement, as there exists qualified disagreement in pluralist societies over who counts as ‘wise’ with respect to political matters. Thus epistemic proceduralism rules out ‘invidious comparisons’ amongst citizens with respect to their normative political wisdom (as explained in chapter II). However, given the widespread view that a ‘good political education’ promotes good political decision-making, and that under Mill’s proposal all citizens would have at least one vote, can the Millian proposal for additional votes for the educated satisfy the qualified acceptability requirement?

Estlund begins the chapter by explaining why Mill’s proposal for additional votes for the educated, and thus an epistocracy of the educated, is a view that epistemic proceduralism needs to take seriously. First, he summarises Aristotle’s criticism of Plato’s proposed epistocracy. Aristotle’s argument, roughly, is that even if one individual, or a few individuals, are wiser than others, decision-making can nonetheless (under most circumstances) be improved by including larger numbers of individuals (even if they are less wise) in the decision-making process. So restricting political decision-making to only the wise can have epistemic costs. However, while this argument might be successful against Plato’s epistocratic proposal, it is unsuccessful against the Millian proposal, since everyone would have at least one vote. The Millian proposal is that the educated should receive more votes, not that political decision-making should be limited to some subset of the population (‘the wise’).

This leads to a dilemma. Most epistocratic proposals can be rejected because there exists qualified objections to claims about who ought be counted as ‘the wise’ (the principle against invidious comparisons). On the hand, the following proposition appears to be extremely plausible:

The political value of education: A well-educated population will, other things equal, tend to rule more wisely (p. 211).”

The proposition is very plausible because it is hard to imagine that any qualified objections could be raised against the idea that persons who are literate, are knowledgeable about how their government works, are knowledgeable of history, and so forth, will (ceteris paribus) tend to make better political decisions.

From the ‘Political Value of Education’ there appears to be a small step to:

The epistocracy of the educated thesis: Where some are well educated and others are not, the polity would (other things equal) be better ruled by giving the well educated more votes (p. 212).”

Estlund then considers two objections to block the move from ‘The Political Value of Education’ to ‘The Epistocracy of the Educated Thesis’: (1) The Deference Objection (which Estlund rejects), and (2) The Demographic Objection (which Estlund endorses).

The Deference Objection

This objection is based on the premise that it is always a qualified objection to refuse to defer to some other agency (in this case, the educated) authority over substantial moral matters. The reason why this objection fails, if I understand Estlund’s discussion correctly, is that if it were to succeed it would require denying that the well educated are wiser with respect to political decision making. But this looks absurd, since if it is correct, why care about the education of the population as a whole (at least with respect to political decision making)? Furthermore, since “the epistemic value of any arrangement at all would fit into the same template, not just arrangements involving education,” granting this objection potentially defeats “the whole epistemic dimension of political argumentation” (p. 213). (I should note that I had some trouble following Estlund’s discussion on p. 214, so I may not be presenting accurately his overall reply to the deference objection.)

The Demographic Objection

This objection holds that limiting the political decision-making process to only the educated can have epistemic costs that outweigh any potential epistemic benefits. More specifically:

The demographic objection: The educated portion of the populace may disproportionately have epistemically damaging features that countervail the admitted epistemic benefits of education (p. 215).”

A common objection to Mill’s plural voting proposal is that since certain groups in existing western societies have superior access to education (based on gender, race, class, and so forth), certain inevitable biases will affect the political decision-making processes if the educated are given additional votes (even if we assume that that the educated do their best to act in good faith). This objection could be met, perhaps, by demographically correcting the group given extra votes (i.e., correcting for under-representation of certain races, classes, and so forth, amongst the educated population). There is a practical worry, though, that it may be unfeasible to demographically ensure that all relevant groups are represented (if we correct for gender, race, and class, what about religion? Sexual orientation? Vegetarianism? Etc.).

An even deeper worry is that there may be what Estlund calls ‘empirically latent’ features — features that cannot (realistically) be tested for, but which nonetheless may be biases — that might disproportionately be found amongst the educated (the examples of racism and sexism are given). Objections based on such empirically latent features to the Millian plural voting proposal are qualified objections. Likewise, one might make a more general objection to the Millian plural voting proposal, namely, that giving additional votes to the educated might involve some possible epistemically damaging demographic distortion(s). Such appeals to what Estlund calls ‘conjectural features’ also count as qualified objections. Consequently, because of qualified objections based on ‘empirically latent’ features and ‘conjectual features’ that might have epistemic costs, and might disproportionately be found amongst the educated, the plural voting proposal cannot satisfy the qualified acceptability requirement.

Estlund then notes that any argument in favour of either equal or unequal votes based on the special epistemic insights of a particular sub-group (e.g., an argument in favour of equal votes for rural voters based on their special insight into rural matters, or an argument in favour of additional votes for victims of injustice based on their special insight into matters of justice) involves invidious comparisons amongst citizens. Consequently, all such arguments are ruled out.

Estlund concludes the chapter by noting that none of his arguments in this chapter (or, for that matter, this book) should be understood as defending “formally equal voting power for each citizen”, as such a defence would require “some non-epistemic consideration such as some suitable conception of procedural fairness of equal respect” (p. 221). Estlund’s aim has been only to show that Mill’s argument in favour of additional votes for the educated — which Estlund takes to be the strongest argument in favour of a kind of epistocracy — fails.


My main concern with chapter XI is a concern that I’ve had since chapter III, namely, the unspecified nature of the ‘qualified acceptance requirement’. In chapter III Estlund presents the general idea of the qualified acceptance requirement (of which Rawls’s ‘Liberal Principle of Legitimacy’ is a specific version) and defends that general idea against some common objections (the overinclusive and underinclusive objections). In chapter III, Estlund was reluctant to provide a specific account of the qualified acceptance requirement, including what kinds of arguments or objections could count as ‘qualified’, in order to defend the plausibility of the idea itself, and not get hung up on details or confusions over, for example, the idea of the ‘reasonable’ (as used by Rawls). However, there seem to be costs in not providing a more specific account of the qualified acceptance requirement — its grounds and its boundaries — and these costs, I think, come out in this chapter.

The first cost is an appearance of a kind of ‘ad hoc-ism’ with respect to the arguments presented. That is, at different points in Estlund’s argument, the idea of the qualified acceptance requirement fails to rule out some kinds of objections (some objections are unqualified, e.g., the objection based on deference does not defeat the Millian proposal) and rule in other kinds of objections (some objections are qualified, e.g., objections based on conjectures about possible biases amongst the class of the educated defeat the Millian proposal). However, unless some more detailed account of the qualified acceptance requirement is provided — specifically, what its basis and boundaries consist in — invoking it in order to rule in and rule out different kinds of objections looks ad hoc. (To be clear: I do not in fact think that Estlund is using the qualified acceptance requirement in this way. However, by leaving it so underspecified, there is, I think, an unfortunate appearance of ‘ad hoc-ism’ in some of his arguments.)

The second cost is that so long as there is no positive argument for the truth of the qualified acceptability requirement (as opposed to the arguments against objections to the idea, as presented in chapter III), I suspect that whatever positive argument there is available for the qualified acceptability requirement, that argument may suffice also to justify (at least some form of) democracy (independent of epistemic considerations). The qualified acceptability requirement is a moral doctrine, not an epistemic one. It constrains the range of acceptable political doctrines such that if a qualified objection can be levelled against a political proposal (say, a proposal in favour of a particular kind of epistocracy), that objection defeats that proposal, even if that proposal would have resulted in an epistemically superior political institution or policy. Consequently the justification for the qualified acceptability requirement has to be some kind of moral principle. But is there any moral principle that could justify the qualified acceptability requirement that could not also justify (some form of) democratic government? I’m not sure that there is.

Let me unpack this a bit. My own understanding of political liberalism has been greatly influenced by Estlund’s article, ‘The Insularity of the Reasonable’ (1999). In that article (in which can be found the ancestor of the overall view presented in chapter III), Estlund made the argument that political liberalism could not be indifferent to the truth about everything. Instead, in order to avoid a vicious circularity, political liberalism had at least to think that its criterion of justification — the Liberal Principle of Legitimacy (or a suitably modified version of it) — was true. However, this was not a significant concession for political liberalism (contra Raz), since it only had to assert that this one principle was true, and that non-true political conceptions of justice could satisfy it (i.e., be ‘successful’ in Estlund’s terms).

I found this argument quite compelling with respect to political liberalism. But it naturally raised the question: why think that the Liberal Principle of Legitimacy is true (the correct principle to govern coercive political arrangements)? Drawing on articles by Thomas Hill (1994) and Charles Larmore (1998), the most plausible answer, it seems to me, is that the Liberal Principle of Legitimacy expresses a principle of equal respect for persons. It does so by trying to justify the exercise of coercive political power in ways acceptable to all ‘reasonable persons’. And the set of ‘reasonable persons’ consist of those persons committed to this very moral principle. (‘Reasonable persons’ are committed to the principle of equal respect for persons by being committed to the ‘criterion of reciprocity’ in their political relations with others, and by acknowledging the ‘fact of reasonable pluralism’. Acknowledging the fact of reasonable pluralism is a requirement of respect for persons if we think that reasonable pluralism is the natural consequence of the exercise of human reason under conditions of social freedom.) Unreasonable persons fail to recognise (or refuse to recognise) this true moral principle; consequently, justifications for the exercise of coercive political power need not endeavour to be justifiable to them. In short, the most plausible account of political liberalism (I think) is based on the principle of equal respect for persons, and this principle justifies maintaining that the Liberal Principle of Legitimacy is the true criterion for regulating the exercise of coercive political power.

However, if the justification for the truth of the Liberal Principle of Legitimacy is the principle of equal respect for persons, could the principle of equal respect for persons not also justify equal voting rights for all citizens? I think that it clearly could, although I will not develop such an argument here (doing so would take too long, and would involve repeating things previously already said by Quong and others; I nonetheless hope that the basic idea is clear enough to most readers).

Estlund himself suggests that the principle of equal respect for persons might support equal voting rights on p. 221, but disavows any commitment to that principle. However, this brings me back to my worry: if Estlund is not committed to some version of the principle of equal respect for persons in order to justify the qualified acceptability requirement, what positive justification exists for the qualified acceptability requirement? The only plausible justification that I can think of, equal respect for persons, also justifies (some form of) democratic government, independent of epistemic considerations. I would be very interested to learn of some other positive justification for the qualified acceptability requirement other than a principle of respect for persons.

In short, my overall concern is that so long as the qualified acceptability requirement remains unspecified — so long as its grounds and boundaries are not explained — a key premise in the overall argument for epistemic proceduralism is missing, and, moreover, that specifying the qualified acceptability requirement may turn out, at the end of the day, to be sufficient to justify (some form of) democratic government (independent of, although obviously not incompatible with, epistemic considerations).

Be Sociable, Share!
This entry was posted in Posts, Reading Group. Bookmark the permalink.

7 Responses to Estlund Reading Group Chapter 11

  1. Blain Neufeld says:

    First of all, my apologies to David, Jonathan, Simon, and everyone else involved in this reading group for being so late with my post! I’ve had to do some unexpected traveling over the past week (which is still ongoing, and may prevent me from commenting until Tuesday). I very much regret any inconvenience that I may have caused.

    Second, my apologies for the big gaps between the paragraphs. I spent over an hour trying to figure out how to have ‘normal’ gaps, but to no avail. The only alternatives seemed to be cutting and pasting from my Word file with no modification (which produced the current ‘double gaps’ between paragraphs), or having no paragraph breaks at all (a far worse option). I’m sure that there’s some simple trick to fix this, but I was unable to figure it out. If someone knows what needs to be done, I’d be happy to edit my post into a more aesthetically pleasing form.

  2. Thanks Blain — I edited it for you. Others have had that paragraph spacing problem but I don’t know how to fix it, since everything works fine for me. If you were to look in the code tab, you might see too many paragraph break [p] tags (with < and > brackets instead of square brackets). One day, graduate programs in philosophy will provide actual technical computer skills so that people like me know what we are doing when we set up websites: “Hume, Reid, and php Coding,” etc.

  3. Ben Saunders says:

    So, if I understand correctly, this isn’t an argument for equality of votes (which Estlund now thinks isn’t really important), but simply an argument against one particular justification for inequality.

    One general worry I have (which doesn’t directly relate to the arguments of this chapter) is about the extent of inequality that may be acceptable. If we allow certain inequalities, it might be that some people are effectively excluded altogether. One example might be the weighted voting rules adopted by the EU (or whatever it was then) in the 1950s: if I recall correctly, Germany, France and Italy each had 4 votes, Belgium and Netherlands 2 and Luxembourg 1. A ‘super-majority’ of (I think) 12 votes was necessary to pass motions.

    Now this isn’t really the context that David is talking about, but presumably if we extend his argument there’s no fairness-based objection to such inequalities; they can only be rejected on epistemic grounds. The problem with this case of weighted voting is that Luxembourg effectively had no power, because it was never pivotal to a winning coalition (any group that won with Luxembourg would also win without). That, to me, looks like an objection to this distribution of votes. I’m not sure whether David can give some epistemic-based reason for this (perhaps along the lines that there’s a loss in excluding a certain distinct perspective?).

    As I say, that’s a bit of a digression, but perhaps it’s worth thinking about as I’m aware of a few recent proposals of weighted voting – by Brighouse and Fleurbaey, and Heyd and Segall – based on differential interests at stake rather than epistemic considerations. I’m not quite sure whether David’s argument is intended to exclude those too. That is, does establishing a qualified objection on epistemic grounds mean we can also (with slight modifications to the argument) block other arguments for weighted voting, or is it simply a rejection of one (epistemic) argument for weighted voting that allows other (non-epistemic) arguments for such?

    Finally, I share Blain’s worry that things seem a bit ad hoc. Presumably, given that Plato would say it’s unjust to treat unequals equally, he’d claim there’s some sort of qualified objection to giving the uneducated equal votes (which they don’t really deserve). After all, the uneducated will presumably also be biased in certain ways (e.g. disproportionately poor), so couldn’t we simply end up facing conflicting (possibly qualified) objections, unless what it is to be ‘qualified’ is spelled out in more detail?

  4. Hi Blain,

    Thanks for the great post. I want to raise a worry I have about this chapter, a worry that relates to a long discussion we had regarding chapter 2. David claims that objections based on latent or conjectural features of an educated class are qualified objections, and thus sufficient to defeat proposals for an epistocracy of the educated. An epistocracy of the educated is ruled out because it is vulnerable to qualified objection. On pp. 218-219 David then reminds us of a passage in chapter 2, where he considered the possibility that universal suffrage might be vulnerable to qualified objections, and thus would be ruled out as well. David’s response to that objection was to declare that departures from universal suffrage introduce an extra element or degree of authority, and that additional increment ‘is subject to the qualified acceptability requirement, whereas its absence is not’ (p. 219). Universal suffrage is thus the default, from which departures need to be justified in a manner that is immune to qualified objection.

    I think this passage poses problems. I believe that objections based on latent or conjectural features are roughly as plausible when applied to the populace as a whole, as they are when applied to a sub-set of the populace like the educated. That is, it is roughly just as plausible to suppose that the potential biases and epistemic flaws of the general public will outweigh whatever epistemic benefits universal suffrage brings, as it is to worry about this with regard to some sub-set of the population like the educated. If objections based on latent/conjectural features can be used to defeat the argument for epistocracy, why can’t they be used to defeat the argument for universal suffrage? Why can’t they be used as qualified objections to the claim that democracy will do better than random procedures? David’s position, maybe, is that universal suffrage is a default, from which only departures need to be justified in a manner that is not vulnerable to qualified objection. If this is his position, then this means even though there are qualified objections against the epistemic argument for democracy, they can be ignored because universal suffrage is the default from which only departures need to be justified. But this makes David’s epistemic arguments in favour of democracy unecessary – all the work is done by the assumption that universal suffrage is the default and the further arguments showing departures cannot avoid qualified objection – we are left with no independent argument explaining why universal suffrage is the default.

    Since I think this is not the view David wants to take, perhaps he would not agree with my claim that the latent/conjectural features objection is qualified when applied to universal suffrage. But that response doesn’t strike me as plausible, plus it has the added problem of ad-hocery that Blain mentions. I must have misunderstood something, so hopefully David will point out where I’ve gone wrong.

  5. Ben Saunders says:

    I’m not sure I’ve quite understood Jonathan’s point. In one sense, the whole electorate can’t be biased as they must be a representative sample (of the whole electorate). In another sense, of course, they can have various (cognitive) biases. That just leads me to think, though, would David support compulsory voting, as a solution to biases of the first sort?

    If people are given freedom whether to vote, then predictably some won’t and abstention is unlikely to be evenly distributed (e.g. the young, old and poor may vote less). It seems that this, like weighted voting, effectively gives certain social groups more power. If that’s objectionable, then perhaps that’s reason to compel everyone to vote.

  6. Ben,

    Just to clarify my point. The argument for an epistocracy of the educated is analagous to David’s argument for democracy in the sense they both rely on epistemic considerations. The former argument claims there are epistemic benefits (relative to universal suffrage) to be had by giving the educated more voting power. The latter argument claims there are epistemic benefits to be had (relative to some randomized procedure) by having a democratic system of decision-making. If the conjectural/latent features objection applies to the epistocracy argument, I think it applies to David’s argument for democracy as well, albeit not in precisely the same way. They are, after all, both arguments which favour giving political power to a certain group of people on the basis of alleged epistemic benefits. An opponent of democracy only needs to say something like this:

    ‘There is reason to believe (or to worry) that people, in general, are vulnerable to certain biases (like racism) and cognitive errors (like regret-aversion or confirmatory bias). We thus have reason to believe that even though there are epistemic benefits to the practice of democracy (e.g. deliberation etc…), those benefits are outweighed by the probability of biases and cognitive errors that people generally make. This is a qualified reason to believe that randomized decision procedures (or an epistocracy of people who have been selected by virtue of being less prone to such biases) will do better in epistemic terms than univeral suffrage’.

    My claim is that if the conjectural/latent features objection is qualified when applied to the epistocracy argument, I think this modified version of it is qualified as applied to democracy. More generally, I think the conjectural/latent feature objection to the epistocracy argument is fairly weak, and thus if it is a qualified way of objecting to an epistemic argument, it seems hard to believe there can be no qualified objections to the epistemic argument in favour of democracy that David advances.

  7. I’ve now posted some remarks on the comments up to this point.

Leave a Reply