Reply to Cyril Hédoin’s “The ‘Epistemic Critique’ of Epistocracy and Its Inadequacy,” Samuel Bagg

The core of Cyril Hédoin’s (2021) essay is a critique of a recent article by Julian Reiss (2019), which claims to offer an “epistemic critique” of epistocracy.[1] As Hédoin explains in §I, in short, Reiss offers two reasons that epistocracy is unjustifiable:

(i) There is no such thing as superior political judgment and;

(ii) There is no such thing as uncontroversial social scientific knowledge … [please read below the rest of the article].

Image credit: Tim Dennell via Flickr / Creative Commons

Article Citation:

Bagg, Samuel. 2021. “Reply to Cyril Hédoin’s ‘The “Epistemic Critique” of Epistocracy and Its Inadequacy’.” Social Epistemology Review and Reply Collective 10 (4): 31-36.

🔹 The PDF of the article gives specific page numbers.

This article replies to:

❧ Hédoin, Cyril. 2021. “The ‘Epistemic Critique’ of Epistocracy and Its Inadequacy.” Social Epistemology 1-13. doi: 10.1080/02691728.2021.1882609.

Articles in this dialogue:

❦ Reiss, Julian. 2019. “Expertise, Agreement, and the Nature of Social Scientific Facts or: Against Epistocracy.” Social Epistemology 33 (2): 183-192.

Yet as Hédoin argues in §II, both propositions are true only in a relatively weak sense, and on this weaker interpretation, neither actually undermines the possibility of a justified epistocracy.

First, Hédoin insists, political judgments often depend on technical judgments. Thus, while there may be no such thing as a best political judgment, there may still be better and worse political judgments, based on the quality of the technical assumptions embedded in those judgments. Second, he argues, social science is certainly controversial, but only moderately so. More specifically, at least some social scientific claims can still merit enough confidence to justify using them for purposes of what Hédoin (somewhat obliquely) calls “political governance.”

In §III, finally, Hédoin addresses another argument against epistocracy, which claims that epistocracies have epistemic costs. While he admits that this argument has some truth to it, he essentially turns the tables on those who advance it, claiming that in order to rule out epistocracy, this “epistemic costs” argument would have to show not only that epistocracy has costs, but that those costs actually outweigh the potential benefits of epistocracy. No one, Hédoin claims, has done so.

This latter point is well taken, though it must be said that one can only accomplish so much by shifting the burden of proof in this way. What Hédoin gives us in the paper is not a defense of epistocracy, after all, but a critique of two different critiques of epistocracy. And in both cases, he relies extensively on this sort of burden-shifting. In other words, he argues that his opponents cannot establish what they want to establish, without actually defending the opposite claim in any substantive way. As he acknowledges in the conclusion, meanwhile, there are a number of other critiques of epistocracy which appear likely to be more successful. (I have argued, for instance, that the most promising critique of epistocracy is pragmatic: in short, any epistocratic institutions are liable to be manipulated, and are thus unlikely to succeed in actually producing an epistocracy—see Bagg 2018).

As it turns out, in other words, Hédoin admits that the normative and practical conclusion his opponents seek to establish is probably true: his argument is concerned only to cast some doubt on whether their method of reaching it is fully reliable. Nevertheless, Hédoin’s point in §III strikes me as perfectly sound on its own terms. In this reply, then, I focus on what I see as the more central piece of Hédoin’s paper: his argument against Reiss’ critique.

In that respect, it seems to me that what Hédoin has shown is that Reiss’ argument is underspecified and/or poorly expressed, not that it actually fails in any deeper sense. On my view, indeed, a sufficiently clarified and expanded version of the “epistemic argument against epistocracy” remains quite plausible. In forcing its advocates to clarify and justify it more explicitly, then, Hédoin has certainly done a service for the literature on epistemic arguments about democracy and epistocracy. However, I do not think he is justified in concluding that he has entirely refuted Reiss’ argument—or at least its more sophisticated relatives.

In short, the core of Hédoin’s objection is that certain of Reiss’ claims are too categorical: “there is no such thing as superior political judgment,” and “there is no such thing as uncontroversial social scientific knowledge” (183). As Hédoin rightly points out, these categorical formulations are surely too strong, and a more textured, scalar account would be more appropriate. There may be no such thing as the best political judgment, but there are still gradations of better and worse. Similarly, there may be no such thing as un-controversial social scientific knowledge, but there are again gradations of controversiality. In my view, however, Reiss (or another supporter of this argument) could acknowledge this and still reformulate the epistemic critique of epistocracy in a way that accounts for this nuance.

The Role of Social Scientific Knowledge

The key claim that must be established by an epistemic argument against epistocracy, along the lines of the one Reiss (2019) suggests, is that we should not disenfranchise anyone on the basis of their failure to possess certain social scientific knowledge. As Hédoin (2021, 5) spells out, an advocate of epistocratic exclusions on political power:

[…] should be able to answer positively to the following requirements: are there some social scientific propositions (i) in which we have a sufficient degree of confidence to warrant the partial or complete disenfranchisement of a part of the population, (ii) that are broadly relevant and important enough to serve as a basis for such a disenfranchisement, (iii) that are not trivial or universally accepted?

These three requirements are apt, but I do not think that Hédoin’s argument actually shows that they can be met. In other words, he does not adequately show, either through argumentation or by providing examples of such claims, that there is any social scientific knowledge which meets conditions (i), (ii), and (iii) simultaneously. And if no knowledge meets those criteria, then there is no plausibly justifiable epistemic reason for excluding people who lack certain social scientific knowledge, and thus no plausible reason for implementing what is commonly called epistocracy. If no knowledge meets these criteria, that is, the epistemic argument against epistocracy goes through. And since Hédoin fails to demonstrate that any social scientific knowledge meets these three criteria, it seems to me, he fails to undermine the epistemic argument against epistocracy.

Much of his argument in §II is an attempt to show that some social scientific knowledge is relatively better established than Reiss seems to claim. This seems right, perhaps even inarguable: different forms of knowledge surely merit varying degrees of confidence, and some knowledge may warrant quite a bit of confidence indeed. Such relatively well-established knowledge, he infers, ought to be usable in “political governance”—and this inference seems quite plausible as well. In politics, we must often make judgments about things that require technical knowledge, and presumably some of that technical knowledge is well-established enough to warrant those judgments in some adequately justified sense. For instance, we do not have absolute certainty in our knowledge of climate science or epidemiology, but this knowledge does seem well-grounded enough to justify reducing carbon emissions or implementing mask mandates in the face of a global pandemic.

However, the fact that some bit of knowledge is well-established enough to be justifiably usable in the vaguely-defined task of “political governance” does not imply that those who lack it can properly be excluded from political power. That is a much more specific, and serious, practical inference, and it does not seem justified by the arguments or examples that Hédoin offers. The plausibility of his argument therefore seems to depend on this ambiguity between knowledge being usable for political governance and knowledge being usable for political exclusion.

For one, we might think that the level of certainty required to justify substantive decisions is lower than the level of certainty required to justify political exclusions. After all, substantive decisions have to be made somehow, even in the face of great uncertainty: we must use whatever knowledge we have access to. By contrast, we do not have to exclude anyone from political power, and it seems quite plausible that we would want to have a higher threshold for this rather drastic action. This concern is entirely internal to criterion (i): i.e., about the level of confidence in a proposition that is required to warrant disenfranchisement.

Even more importantly, then, we must also account for criterion (ii)—and that makes the task of justifying political exclusions far more difficult still. To illustrate this point, consider the following “relatively uncontroversial examples” offered by Hédoin, which are meant as instances of social scientific knowledge that is both well-grounded and important enough to potentially warrant the exclusion of those who do not possess it (2021, 5):

the working and likely effects of short-term economic policies, the understanding of the effects of different schemes of taxations (especially regarding incentive-compatibility considerations), the understanding of the different mechanisms underlying various forms of discrimination, the effects of various voting schemes on political stability …

None of these examples will be remotely convincing to the skeptic of epistocracy. Perhaps it is uncontroversial that if “some persons are completely ignorant about [these issues], letting them make decisions about them would be collectively detrimental.” However, voting on political candidates / parties (call it task V) is just not the same thing as making decisions about those issues (call it task D). Voting for candidates in an election, which is how the vast majority of people exercise political power, is a completely different task from making substantive political decisions, and people undertake it in a very broad variety of ways, on the basis of a very broad variety of reasons. And in particular, even those with high political knowledge tend to vote on the basis of their social identity – see Lodge and Taber 2013; Achen and Bartels 2018. The cognitive content of vote choice and the overall consequences of voter ignorance are controversial topics within political science, but the bottom line is that one’s level of explicit political knowledge is almost certainly not the most important factor in determining the quality of one’s vote choice.[2]

Even if it were, however, it is unclear that any particular piece of knowledge (or set of knowledges) is crucial enough to justify excluding those who lack it. To illustrate: let KD stand for the set of social science knowledge relevant to a particular substantive task D (such as “making decisions about economic policy”) and let KV stand for the set of social science knowledge relevant to task V (i.e., choosing between candidates or parties in an election). Since task V involves selecting people who will ultimately undertake task D, KD is relevant to task V, and is therefore a subset of KV. However, since task D is only a very small part of what the people who are elected through task V will have to do, KD is only a very tiny subset of KV. Now consider that nobody possesses all the knowledge in KV. And people who lack KD may possess other important subsets of KV, which people who possess KD may lack.

So, it may be uncontroversial that we should consider barring people from task D on the basis of their lacking KD. (This is what is implied by the inference that certain social scientific knowledge is well-established enough to warrant using it in “political governance.”) But it does not follow that we should bar people from task V on the basis of their lacking KD (i.e., the claim required for epistocracy, as it is commonly defined in the literature). Since nobody possesses the full knowledge set KV, we have to acknowledge that all voters will lack some subsets of KV. Many of these subsets (KA, KB, KC, etc…) will be about tasks A, B, and C, which are also important—perhaps even more important than task D. The fact that D is an important task, in absolute terms, is thus not a sufficient reason to think that lacking KD is grounds for exclusion. We have to consider whether it is really so important, in relative terms, that lacking it is grounds for exclusion. As far as I can tell, Hédoin does not even try to establish this claim of relative importance. As such, he shows only that lacking KD is grounds for barring someone from task D, not that lacking KD is grounds for barring someone from task V.

To bring it back to the particular examples that Hédoin actually raises: we should probably try to ensure that people with no knowledge about the effects of short-term economic policies are not given the power to decide which economic policies we adopt. However, I do not find it remotely plausible that we should consider disenfranchising people just because they lack this knowledge. The latter claim just does not follow from the former, unless we make a massive number of other additional and unstated assumptions about the cognitive content of voting, the relative importance of economic policy knowledge, and so on.

These assumptions are likely to be far more controversial than the claim that people who make economic policy should probably know something about economic policy. Yet they are required in order to establish the conclusion that an epistocratic exclusion based on the lack of knowledge about economic policy might be justified. So what Hédoin paints as an “uncontroversial” example of social scientific knowledge that meets all three conditions—and might therefore justify a turn to epistocracy—turns out not to be uncontroversial at all. And the same can be said, mutatis mutandis, for the rest of his examples as well.  Given this, we have reason to believe that the epistemic critique of epistocracy succeeds after all. Or at the very least, Hédoin’s argument against that critique fails.

Framing the Argument

In the end, Hédoin wants to avoid talking about what might actually ground exclusion from voting, because he is framing the article as a response to the particular argument made by Reiss, and not as an overall defense of epistocracy. (Indeed, as I have mentioned, he acknowledges that epistocracy is probably not justified, overall). However, even a relatively limited argument against the “epistemic critique of epistocracy” cannot be established without addressing this question. It may work against an uncharitably dense and limited version of Reiss’ argument, which relies on the sort of exaggerated categorical claims that Reiss apparently makes, but it does not work against a version of that argument which adopts more nuanced parameters, such as the one I have suggested in this reply.

On this more robust version of the epistemic critique of epistocracy, in short, the claim is that no social scientific knowledge meets conditions (i), (ii), and (iii) simultaneously. In other words, the claim is that no social scientific knowledge is both secure and important enough to justify excluding those from political power who lack it, yet also not simply trivial or universally accepted. Unless Hédoin can point to knowledge that meets these conditions, his critique of that critique will fail.

Perhaps this seems like one more exercise in burden-shifting. So long as we grant some presumption of value to democracy, however, this worry does not stand up to scrutiny. If our default assumption is that epistocracy is justified, then the burden of proof would be on its critics to demonstrate conclusively that no social scientific knowledge could ever meet all three criteria. I think such an argument is likely possible, and Reiss gets us some of the way there. As Hédoin (2021), shows he does not get us all of the way there, and I certainly have not taken us the rest of the way in the space of this reply.

However, I do not think that making such an argument is necessary, because there are many reasons not to take the justifiability of epistocracy as our default position. On the contrary, we have many reasons to take democracy as our default. And if we do, this puts the burden of proof on advocates of epistocracy to show that some deviation from this democratic baseline can in some cases be justified. Hédoin successfully shows that certain of the apparent obstacles to such an epistocratic argument are less serious than critics like Reiss may claim, but he does not show that those obstacles are entirely illusory—much less that they can be overcome.

Author Information:

Samuel Bagg,, Postdoctoral Prize Research Fellow in Politics, Nuffield College, Oxford University.


Achen, Christopher and Larry Bartels, Democracy for Realists. Princeton: Princeton University Press.

Bagg, Samuel, 2018. “The Power of the Multitude: Answering Epistemic Challenges to Democracy.” American Political Science Review 112 (4): 891-904.

Hédoin, Cyril. 2021. “The ‘Epistemic Critique’ of Epistocracy and Its Inadequacy.” Social Epistemology 1-13. doi: 10.1080/02691728.2021.1882609.

Lepoutre, Maxime. 2020. “Democratic Group Cognition.” Philosophy and Public Affairs 48 (1): 40-78

Lodge, Milton and Charles S. Taber. 2013. The Rationalizing Voter. Cambridge: Cambridge University Press

Reiss, Julian. 2019. “Expertise, Agreement, and the Nature of Social Scientific Facts or: Against Epistocracy.” Social Epistemology 33 (2): 183-192.

[1] Hédoin provides a definition of epistocracy as “government by the wise” (1).

[2] Again, see Bagg 2018 and Lepoutre 2020.

Categories: Critical Replies

Tags: , , , , , , , ,

Leave a Reply