Epistemic Democracy: Making Pluralism Productive

Abstract

What, if anything, is the import of Hayek to epistemic democracy? Although Hayek is revered by epistemic democrats for his insights into the epistemic aspects of the market sphere, it is generally believed that his theory is moot with respect to democratic reason. This paper aims to challenge this verdict. I argue that a Hayekian analysis of inclusive public deliberation contributes at least three valuable lessons: (1) Hayek makes the case that under certain conditions even unbiased deliberators are permanently unable to converge on the best available policy option. Call this the problem of ‘persistent hidden policy champions’. (2) He demonstrates that to unlock hidden policy champions, reasonable minority factions need the opportunity to act on their own evidential standards. (3) He challenges epistemic democrats to think more carefully about how to design the “epistemic basic structure” (Kurtulmus and Irzik 2017) of society in order to account for persistent hidden policy champions.

[I]t should be realized that … introducing a new form of government is … unlikely to succeed. The reason is that all those who profit from the old order will be opposed to the innovator, whereas all those who might benefit from the new order are, at best, tepid supporters of him. This lukewarmness arises partly from fear … partly from the skeptical temper of men, who do not really believe in new things unless they have been seen to work well. (Machiavelli 1988: 20–1, emphasis added)

1. Introduction

A remarkable fact about the philosophical debate on epistemic democracy is that the Hayekian tradition seems to play no role whatsoever. This is surprising because scholars in the latter tradition have been concerned with the epistemic dimension of policymaking for decades. However, epistemic democrats do not ignore Hayekian thought. On the contrary, epistemic democrats discuss and applaud Hayek's insights into the price mechanism and therefore his epistemic analysis of the market sphere. In her important contribution Democratic Reason, Hélène Landemore perfectly summarizes the general position of epistemic democrats towards Hayekian thought. She writes: “Hayek's theory of how the dispersed and local knowledge of individuals aggregate through market mechanisms into accurate prices would seem to be an important part of the story of democratic reason. Upon closer examination, however, it is not” (Landemore 2013: 85). In a recent piece, Anderson (2006) introduced what might be thought of as a new philosophical discipline: institutional epistemology.¹ Institutional epistemology is about inquiring which set of institutions best aggregates and produces the information that is needed to solve the problems of specific subsystems of society. We might, for instance, ask what kind of institution is best suited to the goals of science or economics. In her discussion of Hayek she arrives at a similar conclusion: Hayek's insights, we learn, are an important contribution only to the economic subfield of institutional epistemology (Anderson 2006: 8–9).²

In this paper, I want to challenge this view. Epistemic democrats are correct insofar as Hayek's price theory – developed in “The Use of Knowledge in Society” (Hayek 1945) – indeed does not contribute much to democratic reason. However, it would be a mistake to conclude that the latter paper exhausts Hayek's inroads into institutional epistemology. If we want to understand his core contribution to epistemic democracy, I suggest, we should rather look at “Freedom, Reason and Tradition” published in Ethics (Hayek 1958). Here, as in ensuing work³, Hayek (1960 [2012]: 32) argues that for the epistemic progress of society “Freedom of action … is as important as freedom of thought.” Whereas “freedom of research and belief and the freedom of speech and discussion” are “widely understood”, he maintains, “we tend … to ignore the importance of the freedom of doing things” (Hayek 1960 [2012]: 30). These and similar paragraphs suggest that for Hayek freedom of action and freedom of thought play an equally important role for improving the epistemic progress of society in general. However, even if one is disposed to agree with Hayek that freedom of action is important for the epistemic progress of society as a whole, one might still doubt that freedom of action plays an important part in improving the accuracy of democratic decision-making in truth-apt problem-solving tasks. The main goal of this essay is then to explain the epistemic role of freedom of action in the context of improving the accuracy of democratic decision-making. Freedom of action, as we will see, plays an essential role in unlocking the gains from perspectival diversity.

Let me add here, that I do not claim that this is the only insight of Hayek (and Hayekian scholarship) that is relevant for epistemic democracy. For instance, scholars in the Hayekian tradition have looked at the question of what kinds of decisions should (from an epistemic vantage point) be left to the market and spontaneous social adaption processes on the one hand and which ones to democratic decision-making on the other (Hayek 1973, 1976, 1979 [2013]; DeCanio 2014; Tebble 2017a, 2017b; Boettke 2018). Moreover, Hayek and others have been concerned with the public choice question of how to make sure that democratic decision-making does not devolve into a ‘free for all’ that serves special interests instead of the common good (Hayek 1985: Ch. 5; Somin 2016). Whereas the latter literature is about pointing out contingent threats to successful deliberation (self-interest, biases, etc.), this essay is in part about reconstructing Hayek's analysis of the principled epistemic limits of public deliberation.

The paper is structured as follows: In the second part, I will give a brief overview of epistemic conceptions of democracy and argue that (at least some) epistemic democrats are committed to the claim that democratic decision-making procedures in principle are capable of making full use of society's perspectival diversity. Against this backdrop, the third part reconstructs a Hayekian worry about the capacity of rational collective decision-making to make full use of perspectival diversity. In particular, this reconstruction reveals that under certain conditions even unbiased deliberators are permanently unable to converge on the best available policy option. Call this the problem of ‘persistent hidden policy champions’. In section 4.1, I will discuss and model Hayek's solution to the problem of hidden policy champions. In a nutshell, Hayek argues that the same population of agents that gets stuck on a local optimum under a deliberative regime will be able to find the optimal solution under a regime that permits them to act on their own interpretation of the available evidence and learn from the experience of others. In section 4.2, I will briefly discuss some empirical results that support the Hayekian solution. The final section 4.3, draws together various threads of the paper in order to articulate Hayek's central contribution to the question of how to improve the truth-tracking capacity of democratic decision-making.

2. Epistemic Democracy

Epistemic democrats argue that democratic decisions, under favourable conditions, have the tendency to get things right (Estlund 2009: 175–6). The epistemic approach to democracy claims that democratic decisions tend to be truth-tracking with respect to both means and goals. At least some theorists favouring conceptions of epistemic democracy argue that the democratic decision-making procedure is good at truth-tracking because it is able to “tap into the intelligence of the people as a group” (Landemore 2013: 1) and in that is capable of making full use of cognitive diversity (Landemore 2013: Ch. 4; Landemore and Page 2014: 234–5; Misak and Talisse 2014).

If democratic decisions indeed possess this epistemic virtue, it seems quite plausible to claim that this virtue contributes to the legitimacy of democratic systems and the authority of democratic decisions. Contrast this with the deliberative approach to democracy. According to the deliberative approach, what makes democratic decisions justified or legitimate is the inclusive process of collective deliberation. The rightness of democratic decisions is a function of the process that gave rise to the decision. Whether a certain decision is suitable to meet the expectations to fulfil some collectively desirable goal, however, is not immediately relevant for the justificatory story of deliberative democrats. The core difference between epistemic and deliberative approaches to democracy is that the former values democracy, at least in part, as an instrument to arrive at some independent truth.

What, then, are the mechanisms of democracy that ensure that democratic decisions tend to hit the right goals and correct means? There are two main mechanisms. The first mechanism is democratic voting. According to the Condorcet Jury Theorem (Goodin and Spiekermann 2018), voting processes are truth tracking, given that a great number of people participate in voting, that the choice-set is binary, that the issue that is voted on is truth-apt, and some other conditions about the voters are met. In this essay, however, for reasons of space, I will concern myself only with the second mechanism.⁴ The second mechanism that is supposed to ensure that democratic decision-making converges on the correct decision is inclusive deliberation.⁵ The epistemic virtue of deliberation can be summarized in the proverb two heads are better than one. In this essay, I will have much to say about the epistemic virtues and limitations of deliberation. For this reason, I will confine myself here to pointing out some of the more obvious reasons that explain why deliberation does indeed often lead to correct solutions. The epistemic virtue of deliberation is predicated on the epistemic nature of human beings. Human beings usually possess only bits and pieces of the knowledge relevant to a particular question. Moreover, people tend to be biased in all kinds of ways and are prone to make basic errors in thinking through issues. In such a world, deliberation is an important tool to compensate for the limited knowledge and biases of individual agents, and to correct for errors in argumentation.

3. The Epistemic Limits of Deliberation

In this section, my primary goal is to reconstruct Hayek's worry about the capacity of democratic decision-making to make full use of perspectival diversity. As such, the worry to be articulated lies at the very bedrock of Hayek's thinking. Hayek's work both in economics and social theory revolves around the question of how society can leverage the knowledge of its members for solving its social and economic problems. For that reason, a key theme that runs through Hayek's work consists in comparing procedures of rational collective decision-making to procedures of free competition from an epistemic vantage point. Like most theorists working on democracy, Hayek (1960 [2012], 1973, 1976, 1979 [2013]) was concerned about lobbying, the protection of minorities, as well as about the biases that besiege voters and politicians. His original contribution to the discussion of epistemic democracy, however, consists in his analysis of competition as a discovery procedure. Here, Hayek points to a specific principled, epistemic drawback of collective decision-making. Articulating this drawback is the goal of this section.

3.1. Hidden Champions

Let us start, then, by looking at one of the central passages in which Hayek talks about the limits of deliberation in general:

It is worth our while to consider for a moment what would happen if only what was agreed to be the best available knowledge were to be used in all action. If all attempts that seemed wasteful in the light of generally accepted knowledge were prohibited and only such questions asked, or such experiments tried, as seemed significant in the light of ruling opinion …We might conceive of a civilization coming to a standstill, not because the possibilities of further growth had been exhausted, but because man had succeeded in so completely subjecting all his actions and his immediate surroundings to his existing state of knowledge that there would be no occasion for new knowledge to appear. (Hayek 1960 [2012]: 34, emphasis added)

In the passage, Hayek argues that if society limits itself to undertake projects that are chosen in accordance with “what was agreed to be the best available knowledge”, then the progress of society will come to a halt or at least will be severely hampered. Note that Hayek isn't concerned here with biases, ignorance, or malevolence, but with some kind of epistemic drawback. However, it is not quite clear from his description how to conceptualize this worry about collective decision-making. The goal of sections 3.1 and 3.2 is then to reconstruct Hayek's worry precisely.

I will start by reconstructing Hayek's worry within Landemore's definition of rational deliberation. According to Landemore (2013: 91–4), rational deliberation has three constitutive features. The first feature is that deliberation is aimed at a binding decision. The rationality requirement has two parts; the first part demands that deliberators aim at aggregating, critically examining, and systematizing the available evidence. The second part of the rationality requirement demands that we choose the alternative from the set of alternatives that is best backed up by evidence.

Rational Deliberation: Rational deliberation is, at its core, a process that aims at aggregating and systematizing the available evidence for the purposes of choice. The rationality requirement, inter alia, requires that we choose the alternative from the set of alternatives that is best backed up by the available evidence.

This definition has the advantage that it entails the exact conditions under which rational deliberation will pick out the objectively best alternative in an option set.⁶ Rational deliberation will always pick out the objectively best alternative in a given choice set if the objectively best alternative is the one that is best backed up by the evidence. To put it differently, we can state that rational deliberation is particularly well suited to pick out salient champions.

Salient Champion: A proposal is a salient champion if and only if it is the objectively best proposal within a choice set and it is the proposal that is best corroborated by the available evidence.

The first point I take Hayek to make is that choosing alternatives in accordance with the demands of rational deliberation cannot ensure that we pick out the objectively best solution in a given choice set. The objectively best alternative being simply the solution that an agent would choose, if the agent had perfect knowledge. What a Hayekian analysis alerts us to is that sometimes the most justified and the objectively best choice do not coincide but diverge. Sometimes it is the case that – even if we flawlessly aggregate, critically examine, and systematize the evidence – the processed evidence will fail to point us to the best solution. In such cases, a choice set features what I call a hidden champion.⁷

Hidden Champion: A proposal is a hidden champion if, and only if, it is the objectively best proposal within a choice set, but there is at least one proposal in the choice set that is better corroborated by the available evidence.

The notion of rational deliberation, as well as the notion of hidden champions, presupposes both an evidential and an evaluative standard. An evidential standard, among other things, determines what counts as evidence and how to weigh different kinds of evidence. For the purposes of this essay, I will assume that there are always a number of reasonable evidential standards.⁸ This entails that two agents employing different evidential standards might draw different conclusions from the same bundle of evidence. As introduced, the concept of a hidden champion presupposes a problem-solving context. In a problem-solving context, the task of the deliberators is to find the optimal means to a shared common goal. Hence, the models employed throughout this essay assume that the agents share a common goal, i.e., an evaluative standard.

To illustrate the notion of hidden vis-à-vis salient champion, we might think of a public moral problem. A deliberating group sharing one evidential standard wants to decide which of the three policies, A, B, or C, best solves the problem at hand. The task of the deliberators is then simply to create a ranking of the policies on the basis of the available evidence against the background of a single evidential standard. Such a ranking I will also call ‘a perspective’.

Now, consider the case in which A enjoys high evidential support, B enjoys medium evidential support, and C enjoys low evidential support. Applying the concept of salient and hidden champions, it follows that if Policy A is objectively the best solution, then Policy A is a salient champion. On the other hand, if Policy C is objectively the best solution even though it is not corroborated by the evidence, then Policy C is a hidden champion.

Building on both the definition of rational deliberation and the concept of hidden champions, we can delineate the conditions under which rational deliberation will fail to converge on the best solution in a given choice set.

The Epistemic Limits of Deliberation: A group of deliberators sharing one evidential standard, which deliberates in accordance with the standards of rational deliberation, will always fail to pick out the best solution in a given choice set if the best solution is a hidden champion.

The argument here is straightforward: The rationality constraint requires deliberators to pick the solution that is best recommended by the evidence. Sometimes the best evidence, however, points in the wrong direction. Whenever this is the case, rational deliberation will fail to pick out the best proposal.

Up until this point, we have assumed that deliberators argue on the basis of a shared evidential standard. Next, I want to discuss the notion of salient and hidden champions in the context of reasonable pluralism. The reasonable pluralism that I am concerned with here is a reasonable pluralism of evidential standards (and thus perspectives.) Assume that there are three reasonable factions deliberating about how to achieve a certain shared goal. Let us assume further that the factions are of different size: The Majority Faction has 65% of the votes, the Big Minority Faction has 25%, and the Small Minority has 10%. Each faction reasons about the question of which policy proposal (A–C) is most likely to result in the desired outcome on its unique evidential standard. Assume this leads to the ranking of policy alternatives shown in Figure 1.

Figure 1. Reasonable Disagreement.

To complete the picture, we stipulate that Policy C is the objectively best policy proposal in the option space. Putting the thought experiment in a matrix form gives prominence to a further important detail of the hidden champion concept. A hidden champion, as mentioned, is always defined against a specific evidential standard. In the thought experiment at hand, Policy C is thus a hidden champion given the specific ranking (or perspective) of the Majority and the Big Minority. However, in the perspective of the Small Minority, Policy C is a salient champion.⁹ At a certain time, t₁, the debate about the policy comes to a conclusion and ballots are cast. The majority wins the vote and Policy A is introduced. Imagine further that Policy A, as the majority expected, ameliorates the problem at hand somewhat but doesn't solve it as well as Policy C would.

This brings an important question to the forefront that so far has received little attention. To rationally convince the majority, the minority faction needs to acquire a certain new bundle of evidence, E, which it does not possess in t₁. The question then becomes: what mechanism within democratic society ensures (or makes it likely) that the minority – at least in principle – will be able to produce the evidence it needs to convince the majority of a hidden champion at a later point?

3.2. Persistent Hidden Champions

In Habermas's 1989 article “Volkssouveränität als Verfahren”, we find an illuminating response to the problem of hidden policy champions.¹⁰ In the piece, he interprets democracy as an iterative process that is truth-oriented. Adopting a view by Justus Fröbel, Habermas (1989: 468) suggests that majority rule only demands that the minority renounces its will to govern in favour of the majority until the minority is able to rationally convince the majority of the soundness of its views.¹¹

Habermas thus concedes that agents who argue for and defend (what can ex-post be identified as) hidden champions might be unable to convince other deliberators in a rational deliberation at a certain moment in time. This view is common to a whole range of democratic theorists. The underlying idea here seems to be something like this: It must be conceded that minority factions, putting forward hidden champion proposals, might not be able to convince the majority at a certain point of time.¹² Deliberation itself, however, should be viewed as an open process that extends over time. Nevertheless, Habermas neither analyses the problem of hidden champions nor presents an answer to the question of what mechanism within democratic society ensures that the minority will be at least in principle in a position to produce the set of novel evidence it needs to convince the majority that its own favoured proposal is indeed a hidden champion.¹³

Historically, as Popper has noted, philosophers such as John Stuart Mill and August Comte subscribed to what he calls the unconditional theory of epistemic progress. However, the idea that epistemic progress is unconditional is rather implausible. As Popper (2002) argues, epistemic progress in any area is, for the most part, a function of the rules of the game. To see this, we only need to consider what levers we would need to pull to slow down or arrest scientific progress, such as decreasing public funding of the sciences or reducing academic freedom. If Popper is right, we cannot simply assume that minorities are (usually) in the position to generate the evidence that is needed to convince the majority. However, it is not only Habermas who fails to present an answer to the question. Epistemic conceptions of democracy so far have neglected the issue of knowledge generation almost entirely (Müller 2018), and hence have failed to give the issue of hidden policy champions serious attention.

That leading epistemic democrats are insufficiently aware of the issues pertaining to knowledge production is also reflected in the recent discussions about the Diversity Trumps Ability (DTA) theorem, spearheaded by Landemore (2013). The theorem states that under certain conditions, a group of randomly selected diverse problem solvers beats a group of experts in problem-solving tasks. Take for instance a case that Landemore discusses in a co-authored paper (Landemore and Page 2014: 235). In the case discussed, a deliberative body of locals is attempting to solve “a recurrent safety issue on a dark bridge that separates it from the city's downtown”. Initially everybody in the community has some sub-optimal proposal for solving the issue, but by building on each other's proposals, they eventually find the best one. The DTA model employed by Landemore and Page, however, is based on two important assumptions. The first assumption is that deliberators share an evaluative standard or as the co-authors put it a “common objective” (2014: 235). The second assumption is that the deliberators can readily agree on whether a proposed solution will solve the problem at hand. Landemore and Page (2014: 234) write “in pure problem-solving context[s], we implicitly assume the existence of an oracle, namely a machine, person, or internal intuition that can reveal the correct ranking of any proposed solutions”. What I want to draw attention to is that the employed DTA model essentially assumes away the problem of hidden champions by introducing the idea of an oracle. If an oracle is at hand, the problem that a participant to a deliberation cannot prove the superiority of her own proposal because she cannot field sufficient evidence is simply ruled out. Henceforth, the distinction between using knowledge via deliberation and creating knowledge via experimentation does not surface as a distinct problem because the model assumes – via the introduction of an oracle – that the knowledge needed to solve the problem at hand is always readily available. The problem of knowledge production is thus sidelined by the very assumptions of the model.¹⁴

That epistemic democrats are insufficiently aware of the problems associated with knowledge production might also explain (in part) why epistemic democrats employ rather minimalist definitions of democracy. Landemore (2013: 10), for instance, defines democracy as “a procedure for collective choice decisions” rather than, “a set of political institutions”. Estlund (2009: 65) similarly employs a rather thin notion and understands democracy primarily as the “collective authorization of the laws by voting”.

The current discussion in epistemic democracy then does not present any answer to the question of what mechanism ensures (or makes it likely) that hidden champions will – at least over time – be discovered. Indeed, theorists seem to be mostly unaware of the problem. To think through the issue of hidden champions, the concept of an “epistemic basic structure” (Kurtulmus and Irzik 2017) is helpful. The concept of the epistemic basic structure of a well-ordered democracy denotes the institutions of democracy that are tasked with creating and disseminating knowledge. The authors develop the concept within a Rawlsian framework, but it fits the current issue well enough. The branches of the basic epistemic structure that are tasked with knowledge creation, according to Kurtulmus and Irzik (2017: 129), are the “institutions of science and … those government agencies and offices that carry out research or publish basic statistics”. However, the creation of new knowledge is, at least to some extent, dependent on the creation of new data. One of the main sources of data creation is the democratic political process itself (Müller 2018). By introducing new laws or discarding old ones, the government changes the institutions of society. This, in turn, produces new data on the effects of policies. The question is then whether these listed mechanisms can ensure that hidden champions will usually be discovered. In what follows, I will argue that if the institutional arrangements of democracy are not purposefully designed with an eye to the problem of hidden champions,¹⁵ there is reason for doubt.

The problem that Hayek points to is that at times the proponent of a hidden champion will not be able to produce the required evidence if she is not permitted to put her idea into practice. Hayek (1960 [2012]: 96, emphasis added) explains: “Though discussion is essential, it is not the main process by which people learn. Their views and desires are formed by individuals acting according to their own designs; and they profit from what others have learned in their individual experience.” He continues: “it is always from a minority acting in ways different from what the majority would prescribe that the majority in the end learns to do better.” However, this means that to the extent that minorities are not able to act according to their own evidential standards, they might be precluded from generating the evidence that is needed to convince the majority. According to this reconstruction, reasonable minorities face some sort of catch-22 in democratic settings that are not purposefully designed around the problem of hidden champions. To convince their rational peers that their solution is a hidden champion, they need to provide new evidence. However, in order to provide the missing pieces of evidence, they need to have the permission of the majority to try in the first place (which they do not get, because they lack the appropriate evidence).

Hayek's argument then seems to revolve around the claim that in some instances, the evidence that doing X has the desired consequences Y cannot be sufficiently established for agent A (who deliberates on the basis of evidential standard A) in the absence of doing X. Call this the demonstration claim. The demonstration claim rests on the assumption that sometimes the ‘proof is in the pudding’. Occasionally a social or public innovator has no way to convince her rational and benevolent peers that her solution for a given problem is the best if she is not permitted to demonstrate her solution. To put it in the words of Machiavelli quoted in the epigraph, convincing the majority of new forms of government, or innovative policy solutions for that matter, is difficult partly because citizens “do not really believe in new things unless they have been seen to work well”. The innovation diffusion literature, that studies how innovations across domains (innovative social rules, policies, products, etc.) diffuse through society, adds additional support to the demonstration claim. Within innovation diffusion studies, it is well established that the trialability and demonstrability of an innovation of an ex-post successful innovation is positively correlated with its rate of adoption (Rogers 2003). Novels and movies like to portray the stories of scientists who developed new approaches to specific problems even though the deck was stacked against them. In every particular case, one might ex-post quibble over whether a certain result couldn't have been predicted and whether the experts doubting the ex-post successful innovator were really justified in discarding a certain alternative proposal. In every particular case, thus, it might be hard to judge whether the reason why a certain proposal had not gained sufficient uptake by the majority was because of psychological bias or genuine epistemic inability to recognize the true value of a proposal. Nevertheless, it is undoubtedly true that there are cases in which well-meaning experts are swayed by the demonstration of results they would never have expected (Zollman 2010).¹⁶

If the proponents of hidden champions are permanently prohibited from producing the required evidence, this generates, what I want to call the problem of persistent hidden champions: A proposal is a persistent hidden champion if it is (i) a hidden champion, (ii) there is an entrenched majority that prevents proponents from establishing the required evidence and (iii) the demonstration claim applies. The issue of persistent hidden champions is, of course, a general one that applies across domains of collective choice.

In section 3.1, we started out with Hayek's (1960 [2012]: 34) worry that social progress might come to a hold if “only what was agreed to be the best available knowledge were to be used in all action”. We are now able to put this worry more precisely. What the reconstruction of Hayek's inroad into political institutional epistemology suggests is the following conditional: If the institutional arrangements of democracy are not purposefully designed with an eye to the problem of persistent hidden champions, democratic reason will often be unable to converge on the best policy solution over time because the advocates of the best alternative will be permanently prevented from generating the required evidence that is needed to convince the majority in rational discourse.¹⁷

Hayek worries thus that rational collective deliberation under certain conditions might be unable to make full use of reasonable perspectival diversity. As a result, society might get permanently trapped in a local optimum. In a nutshell, this is the problem of persistent hidden policy champions.

In the next section, I will demonstrate that the same group of rational deliberators that gets stuck on a local optimum will discover its hidden champion if different factions are free to act on their own evidential standards.

4. Making Pluralism Productive

Naturally, society has a great interest in finding increasingly better solutions to its problems. This means that society has a great interest in discovering (persistent) hidden champions across domains. In this section, I will focus only on hidden champions in the realm of public policy. However, the insights developed here can be generalized much more broadly.

4.1. Unlocking Hidden Champions

How can we discover hidden champions if our prime method of discovering the best solutions – rational deliberation – fails to deliver? Hayek did not only pose this question but also proposed a solution. For society to uncover (persistent) hidden champions, he argued, it is paramount that new ideas have space to demonstrate their value. Let us look at three key passages in Hayek that give us a first hint of how we might be able to discover persistent hidden champions:

Though we must always strive for the achievement of our present aims, we must also leave room for new experiences and future events to decide which of these aims will be achieved. (Hayek 1960 [2012]: 22–3)

[N]obody can know who knows best and … the only way by which we can find out is through a social process in which everybody is allowed to try and see what he can do. (Hayek 1985: 15)

It is because every individual knows so little and … because we rarely know which of us knows best that we trust the independent and competitive efforts of many to induce the emergence of what we shall want when we see it. (Hayek 1960 [2012]: 27)

The crucial element to unlocking hidden champions, according to Hayek, is granting agents the “room” to act on their own evidential standards and “try what he [or she] can do”. Hayek believes that only through such competition, we can identify hidden and persistent hidden champions.¹⁸

Next, I want to reconstruct the mechanism that explains how a freedom to act on one's views can unlock (persistent) hidden champions. In Constitution of Liberty, as pointed out earlier, Hayek (1960 [2012]: 96) writes: “It is always from a minority acting in ways different from what the majority would prescribe that the majority in the end learns to do better.” Even though it seems highly questionable whether the majority only learns by observing what the minority does, this statement, nevertheless, provides the key to understanding of how democratic societies can become better at unlocking hidden champions.

To make the Hayekian point as lucid as possible, I want to return to our earlier example (Figure 1), featuring three factions evaluating three policies (A, B, C) according to their own evidential standards. Assume that the Small Minority over many years has tried to convince the majority of the desirable effects of Policy C, the hidden champion, but hasn't been successful. As a result, our stylized polity got stuck at a local optimum (Policy A).

Now, what is the Hayekian solution to this problem? The Hayekian solution to the problem of persistent hidden champions consists in granting factions, within reasonable limits, an opportunity to act on their own evidential standard. For the present case, this means that all three factions are permitted to implement their favourite policy, say, in a separate location.¹⁹ Assume, then, that every faction implements its favourite policy and by the time of t₂, Policy C has produced certain desirable effects. Let us further imagine that Policy B is not making much of a dint. Having arrived in t₂ and evaluating the policy effects in t₁, the Big Minority is confronted with a new set of evidence in favour of Policy C and is hence forced to re-evaluate its policy ranking. Given the new evidence, the Big Minority judges that the total evidence (given their evidential standard) now speaks in favour of Policy C. The Big Minority, therefore, adopts Policy C in t₂.

Assume, further, that the Majority Faction is still not convinced. The reason being that there are reasonable worries that Policy C's positive effects might be not sustainable over the long term. However, a couple of years later, well into t₃, there are still no signs that the worries by the Majority might come to fruition. Thus, re-evaluating its policy, adding the new evidence to the existing stock, the Majority finally decides to adopt Policy C as well, since the total body of evidence (applying the majority's evidential standard) now speaks in favour of Policy C. Figure 2 depicts the policy diffusion just described.

Figure 2. Hayekian Cascade Model.

The surprising result of the cascade model is this: The same population of agents that gets permanently stuck on a local optimum under a deliberative regime is able to find the optimal solution if each faction is permitted to act on its own evaluative standard.²⁰

4.2. Hayek's Cascade Model: empirical evidence

Before I go on to discuss the upshot of this model for current political theorizing, let me comment on the question of whether there is any empirical evidence that corroborates the Hayekian Cascade Model. Is political learning really, at least at times, working like this? Are political units taking note of the policy experimentation of surrounding political units? The short answer to this question is yes. Political scholars for decades have worked on the mechanisms that underly policy diffusion. One of the central mechanisms that explains policy diffusion is policy learning. Mitchell and Petray (2016: 286) explain, policy learning is understood as a process in which “policymakers observing the successes and consequences of a policy in other jurisdictions, assessing the outcomes of the policies, and then deciding on whether or not to adopt the policy”. Over the last few decades, scholars have gathered an impressive number of studies documenting that political learning takes place horizontally (between political units of similar size) as well as vertically (between political units of different sizes and authority). The literature, moreover, is in consensus that local politicians in their innovation decisions often take into account the “experience of their more adventurous state counterparts” (Hollander and Patapan 2017: 4) and check specifically whether the “dire consequences predicted may have … materialized”. This, however, does not mean that policy learning is the only mechanism that leads to policy diffusion. Indeed, scholars have identified at least three other mechanisms that drive policy diffusion: coercion by other governments, pure imitation of successful first movers and competitive pressures (Shipan and Volden 2008).

That political units take into account the experience of similar or even quite different political units to aid them in their decision process for or against a certain policy should be quite uncontroversial to begin with, since learning from the experience of others seems to be a simple demand of rationality.

4.3. Hayek's import for epistemic conceptions of democracy

What lessons for epistemic democracy can we draw then from the Hayekian analysis of the principled limits of rational deliberation? There are three important takeaways. First, Hayek makes a convincing case that even under favourable conditions, sometimes inclusive public deliberation will be unable to converge on the best available policy option over time because it is unable to make full use of perspectival diversity. The Hayekian analysis of democratic choice alerts us thus to the problem of persistent hidden policy champions and in that expands our conceptual toolbox for thinking about epistemic progress in collective choice settings.

Secondly, advocates of democracy have rightly emphasized the epistemic value of freedom of speech and conscience. Misak and Talisse write:

[T]he processes by which people reason together must be formally secured – there must be free speech, free association, freedom of conscience, as well as various protections for dissent, disagreement, and protest. We … contend that the social-epistemic environment requisite for proper believing is best secured under democracy. Every believer thus has compelling epistemological reasons to embrace democracy. (Misak and Talisse 2014: 368–9)

Putnam puts the epistemic argument for democracy in even more striking terms:

Democracy is not just one form of social life among other workable forms of social life; it is the precondition for the full application of intelligence to the solution of social problems. (Putnam 1992: 180)

Hayek does not deny the epistemic value of free speech and concomitantly the value of epistemic deliberation, but he points out that this is not the whole epistemic story. The second lesson that emerges from his analysis is then that in order for democracy to create the “social-epistemic environment requisite for proper believing” and ensure “the full application of intelligence to the solution of social problems”, certain institutional background conditions need to be in place that facilitate the emergence of politically relevant knowledge. In particular, the Hayekian analysis points out that in order to make full use of the perspectival diversity, it is not sufficient to guarantee that reasonable minority factions have a say in collective democratic decision-making but have a live option to create evidence for their views. Freedom of action, Hayek argues, is as important as freedom of thought and speech for reaping the epistemic benefits of cognitive diversity. Against this background, we are finally in a position to understand more clearly what Hayek meant in his concluding remarks to “Freedom, Reason, and Tradition”:

None of the conclusions are arguments against the use of [democratic] reason but only arguments against such uses as require any exclusive and coercive powers of government; not arguments against experimentation as such, but arguments against all exclusive, monopolistic power to experiment in a particular field – power which brooks no alternative and is in its essence based on a claim to the possession of superior wisdom – and against the consequent right to preclude the emergence of better solutions than the ones to which those in power have committed themselves. (Hayek 1958: 242, emphasis added)

What is important to note here is that Hayek is not simply concerned with increasing the rate of experimentation. He is concerned with the epistemic gains that society forgoes, if reasonable minority factions are prevented to act on their own evidential standards.

It should not go unnoticed that this result also has important implications for the debate about exit and voice. Ever since the publication of Exit, Voice and Loyalty by Hirschman (1970), exit was predominantly viewed as a potential threat to democratic values. It has only been recently that scholars have rediscovered the old insight by political economists (Vaubel 2008) that exit options are tremendously important to protect minority groups from domination by persistent majorities (Warren 2011; Taylor 2017). Exit emerges from the current literature as an important democratic tool to empower minorities and to facilitate the important value of non-domination in the public and private sphere. In a ground breaking paper, Warren has recently introduced the distinction between enabled exit and institutionalized exit. Whereas enabled exit is mainly about “situations in which states … structure social and economic relations by underwriting individual choice” (Warren 2011: 690), institutionalized exit is about designing institutions with an eye towards forestalling political domination. For the purposes of this paper, the latter half of the distinction is of particular interest. As Warren (2011: 692) explains: “Institutionalized exit refers to situations in which exit is designed into an institution” “such that individuals can chose (or exit) providers of services, voice, or representation”. As the present analysis shows, institutionalized exit options do not only forestall political domination, they can also facilitate epistemic progress by enabling minority factions to produce the evidence that is needed to unlock persistent hidden policy champions.

Finally, the Hayekian analysis of persistent hidden champions raises the following question: What institutional scheme is most suitable to empower reasonable minority factions to produce the evidence that is needed to unlock persistent hidden champions? I cannot engage with this question in this essay, nor do I believe that this question can be adequately settled in a single dedicated paper. A full treatment of this question will need to engage with the normative question of the optimal trade-off between epistemic gains (and hence the potential future moral gains resulting from unlocking hidden champions) and various other democratic values that attach to installing effective exit options within a certain democratic community.²¹ To put it differently: Because creating institutionalized exit in some cases is infeasible or undesirable all-things-considered, creating institutionalized exit schemes is probably an issue that needs to be decided on a case-by-case basis. This is not to deny the fact that certain grand institutional schemes are better than others in unlocking hidden champions. Undoubtedly, distributed forms of government, such a polycentric structures of various kinds (Ostrom 2005; Kukathas 2007; Aligica 2014; Müller 2019; Gaus 2021) provide much more opportunity for minority factions to innovate than highly centralized forms of government. Hayek's third input then consists in a challenge: He challenges epistemic democrats – and political theorists more generally – to think more carefully about how to design the epistemic basic structure of democracy such that it provides sufficient opportunity for minority factions to contribute to the epistemic progress of society. To put it in other words, he challenges us to inquire about the institutional preconditions that need to be in place to make pluralism productive.²²