Abstract
Recently, epistocrats have challenged the value of democracy by claiming that policy outcomes can be improved if the electorate were narrowed to empower only those with sufficient knowledge to inform competent policy decisions. I argue that by centering on contesting how well regimes employ extant knowledge in decision-making, this conversation has neglected to consider how regimes influence the production of knowledge over time. Science and technology studies scholars have long recognized that political systems impact the productivity of expert research. I argue that in order to evaluate which regime is “smarter,” we must consider not only how well they employ existing knowledge in decision-making, but we must also assess how those regimes influence the ongoing production of policy-relevant knowledge. Thus, I offer an instrumental defense of democracy based on its capacity to encourage a superior pattern and quality of expert research to inform policy decisions over time. Epistocracy may be effective at employing extant knowledge in the short run, but in the long run, democracy is a superior environment for producing knowledge to inform policy decisions.
Contemporary democracies are experiencing a prolonged period of threats from populist antisystem forces (e.g., Connolly 2017; Levitsky and Ziblatt 2018). Mounk (2018) likewise flags survey evidence of a multidecade decline in support for belief in the essential value of living in a democracy (105–10). As democracies degrade and publics grow skeptical of their value, epistocrats such as Brennan (2017), Somin (2016), Bell (2015), and Caplan (2011) argue that the best response is to sacrifice democracy in order to save and fortify the authority of experts over policy decisions. They indite the mass publics in democracies as incorrigibly ignorant, irrational, and tribalistic. In contrast, epistemic democrats such as Landemore (2012, 2020), Estlund (2009; Estlund and Landemore 2018), Surowiecki (2005), and Goodin and Spiekermann (2018) argue that rule by expert groupthink is prone to failures of myopic vision. Instead, they argue that democratic decision-making possesses emergent epistemic properties that epistocrats wrongly discount. Other committed democrats, such as Bagg (2018) and Achen and Bartels (2017), lament that the mechanisms for democracy’s alleged emergent epistemic properties often appear to be fragile beyond restrictive conditions. They contend that a more realistic and robust defense of democracy is possible, though that defense centers on valuing democracy narrowly as an anticorruption mechanism after accepting the bulk of epistocratic critiques of voter competence. Ultimately, I side with the democrats here and offer an epistemic defense of popular rule. Unlike the existing defenses, however, I do not locate democracy’s advantages in the emergent properties of collective decision-making or anticorruption mechanisms. Instead, I argue that democracy is better than epistocracy at producing socially useful patterns of expert research.
Advocates for competing regimes agree that we ought to evaluate regimes on the exclusive basis of their instrumental merits. To wit, we ought to value regimes based on their ability to solve social problems. Brennan (2017) claims “the only reason to favor democracy [or epistocracy] over any other political system is that it is more effective at producing just results” (11). I reformulate the instrumental criteria of solving social problems here as the question of which regime is more capable of intelligent action. Intelligent action, whether it be performed by individuals, political assemblies, AIs, animals, or regimes, requires the actor(s) to engage with the world with intention and success. Those actions require (1) sufficiently accurate knowledge, data, and interpretations of the world and (2) an adequate decision-making function that transforms that knowledge/data as inputs into successful actions as outputs. Intelligent action is possible when an actor satisfies these requirements through both effective knowledge-making and decision-making procedures. Knowledge-making includes generating, verifying, and communicating knowledge claims. In this article, we will be primarily, though not exclusively, concerned with expert research as a form of knowledge-making. Decision-making entails the application of those knowledge claims to the world through policies. By extension this also concerns the procedures for deciding who is empowered to decide for others as representatives. In the example of COVID-19, knowledge-making included experts identifying the transmission methods of the virus and decision-making involved applying that knowledge in the form of masking and social distancing requirements.
While knowledge- and decision-making are both necessary for intelligent action to solve social problems, the debate thus far has centered near exclusively on the (in)competencies of the people who are assembled in decision-making rooms. With the notable exception of Ober’s (2008) study of innovation in classical Athens, the central point of contention in this conversation has been questioning which regime best composes that group of deciders to apply extant knowledge to policy decisions. This approach omits entirely how that knowledge is produced. Gunn (2014) notes that the origins of knowledge often do not enter into conversations among democrats. Landemore (2012, 240) highlights the question of regimes knowledge-making over time as an unexplored frontier. Friedman (2019) has similarly criticized this conversation for taking for granted the existence of adequate knowledge to ground the competing projects. This lacuna in the conversation concerning the production of knowledge is notable in light of the fact that scholars such as Brown (2009, 15) and Collins and Evans (2017, 149) have identified intuitive affinities between scientific knowledge-making and democracy. Those affinities, however, have not been developed into an epistemic defense of democracy that analyses how electoral mechanisms interact with science as a social process.
Recently some political theorists have turned to take an interest in the role of science in a democratic society (Brown 2009; Fischer 2000, 2009; Moore 2017; Pamuk 2021; Turner 2003). These contributions frequently, though not always, adopt an interdisciplinary approach to bridging political theory and science and technology studies (STS) scholarship. As of yet, however, these scholars tend to not speak extensively and directly to the epistocracy versus democracy debate. To take a recent example, Pamuk (2021) explicitly declines to weigh in on the comparative merits of democracy versus epistocracy. Instead, she takes democracy’s desirability as a given and focuses on exploring institutional innovations for democratizing science. Similarly, Moore (2017) is centrally concerned with structuring public communications of science in deliberative systems in order to preserve the possibilities for active judgments on the part of non-expert citizens. Hence, while progress has recently been made to turn the attention of political theory toward the role of science in a democratic society, the utility of that turn has not yet been applied to reevaluating the debate between epistocrats and democrats. More generally, Moore (2017) observes that “there has been relatively little systemic engagement between science studies and political theory” (15).
Here I address this lacuna in the conversation by bringing knowledge-making as it operates within existing democracies into center focus. This shift enables an indirect flanking approach to the entrenched lines of contention in the debate between epistocrats and epistemic democrats focused exclusively on voter competence. This new vector reveals that democracy holds the advantage of fostering superior expert knowledge-making that in turn contributes to superior intelligent action. Additionally, by centering my analysis on existing democracies I craft an approach here that is intended to be resilient to the many epistocratic arguments concerning voter competence that are well established in empirical political science scholarship. Potential institutional innovations to democratize expertise (e.g., Pamuk 2021) may certainly enhance the effects I highlight here; in this sense, my argument ought to be understood as establishing a floor, not a ceiling, to democracy’s epistemic potential. That said, my argument here avoids appeals to hypothetical institutions.
The possibilities of democracies holding advantages in knowledge-making have remained largely invisible because political theory, again with notable exceptions, has largely failed to engage with STS scholarship. The result of this neglect in the epistocracy versus democracy discussion is that the present conversation defaults to assuming the linear model of science and politics (“linear model” for short). That linear model includes two primary features. First, there is a strict sequencing of knowledge- and decision-making, with knowledge-making occurring first upstream in the sequence. Secondly, knowledge-making, especially scientific knowledge-making, is taken to be a strictly autonomous realm independent from political decision-making (e.g., Guston 2007, 70; Pielke 2007, 80). Following from those two points, there is the implication that knowledge enters into decision-making; however, political decisions ought not to influence knowledge-making. Thus, knowledge-making has been assumed to be strictly prior to and independent of decision-making, and that assumption leads this conversation to partition knowledge-making outside of consideration.
The linear model, however, is simply mistaken. STS scholars have long argued that knowledge- and decision-making exist in a bidirectional relationship (Miller and Munoz-Erickson 2018). Political decisions do indeed influence how knowledge-making occurs. Following from this feedback between knowledge- and decision-making is the conclusion that selecting between democracy and epistocracy is not merely a question of systems for decision-making. The choice also implicates differences in expert knowledge-making across the regimes. The endogeneity between knowledge- and decision-making is worth emphasis and analysis because it reveals that the democracy vs. epistocracy debate has been founded on the mistaken assumption that these processes were strictly partitioned when in actuality they are interwoven. That assumption has produced a peculiarly odd gap in this conversation in which scholars discuss the proper respective roles of democratic publics and expertise in society, while explicitly declining to directly examine how scientific expertise interacts with politics. On the rare occasions scientific experts are acknowledged, they are relegated to advisory roles outside of politics (e.g., Goodin and Spiekermann 2018, 82; Landemore 2012, 13; 2020, 192).
Here I expand on the feedback between knowledge- and decision-making to craft an epistemic defense of democracy. I center on the role politicians play to influence expert knowledge-making in contemporary societies. That role can be understood by examining the (1) means, (2) motives, and (3) opportunity structures politicians encounter across democracies and epistocracies. I contend that politicians are motivated primarily by the desire to gain (re)election. They pursue that goal by appealing to salient identity groups within the electorate. Importantly, politicians understand expert knowledge is a strategic resource for political communication. As one Republican political strategist advised candidates, “‘be even more active in recruiting experts who are sympathetic to your view and much more active in making them part of your message’ because ‘people are more willing to trust scientists than politicians’” (emphasis added; quoted in Lee 2003). Following this advice, politicians employ what I describe as an epistemic toolkit of formal and informal powers as the means by which they encourage advantageous expert knowledge-making. Further, democracy and epistocracy provide two distinct opportunity structures to inform how politicians employ their epistemic toolkits. Within democracies, elites are confronted with heterogeneous electorates, and they are expected to correspondingly diversify how they deploy their epistemic toolkits. Epistocracies, in contrast, present politicians with homogenous electorates. Politicians operating in that context are expected to concentrate their toolkits on attending to the enfranchised few.
After accounting for the means, motives, and opportunities that confront elites across democracies and epistocracies, I argue that democracies hold an advantage in knowledge-making. Specifically, democracies generate a favorable alignment between citizens, politicians, and experts. That alignment emerges when the electorate roughly approximates the diversity of social identities among the public, and politicians encourage a correspondingly diverse pattern of expert knowledge-making in society. To wit, democracy improves the pattern of knowledge production that informs intelligent action to solve social problems. Further, those advantages in knowledge-making are significant enough to enable us to prefer democracy over epistocracy.
In section one, I recap the current discussion while emphasizing how it has centered on decision-making to the exclusion of knowledge-making. In section two, I draw upon STS scholarship to describe how knowledge- and decision-making exist in a bidirectional relationship. In this section, I also establish the means politicians employ to influence knowledge-making by inventorying the contents of their epistemic toolkits. In the third section, I describe the motives of political elites and contrast the opportunity structures created by epistocracy and democracy. Within section four, I compare the effects of epistocracy and democracy on expert knowledge-making. Here I contend that democracy is more effective at fostering a beneficial pattern of knowledge-making that aligns with the plural group identities within the public. Finally, in section five, I consider capitalism and liberalism as two potential alternative “epistemic engines” for society, and I argue that democracy encourages more and different knowledge-making than either alternative.
Decision-Making in Democracy and Epistocracy
Intelligent action to solve social problems requires both quality decisions and quality knowledge. Excellent decision-making procedures will often fall prey to “garbage-in, garbage-out” if the knowledge claims informing them are illusory. Knowledge, however, has been taken for granted in this conversation. Fitting within the logic of the linear model, it operates under the assumption that knowledge exists prior to and independent of political decision-making. Thus, knowledge has been wrongly interpreted as a fixed resource in the background. As a result, this discussion has been dominated by contests over how to compose the correct assembly of people to competently apply that assumed fixed knowledge.
Epistocrats and democrats concur in accepting expertise as a valuable input into decision-making. While championing the epistemic benefits of democratic over exclusively expert decision-making, Landemore (2020, 192) and Goodin and Spiekermann (2018, 82) reserve roles for experts to assist the demos by providing technical information. Democrats may express skepticism about experts, but they are not anti-experts. Epistocrats, predictably, prioritize expertise in decision-making. They hold that while experts are not infallible, they are worth listening to unless a compelling reason exists to discount their claims. Caplan instructs, “we all share a presumption that when an expert disagrees with a nonexpert, the expert is right. This holds in math, science, history, and car repair” (emphasis added; ibid. 81–82). In politics too, we are advised that the public would profit by deferring to experts.
The points of departure in the argument regard (1) the broader distribution of knowledge in society and (2) how easily that knowledge can be identified. For epistocrats, policy-relevant knowledge is concentrated in experts, and they are easily identified in advance of decision-making (or at least could be if the demos would listen). An analogy here for the epistocratic case is Plato’s tale of the true ship’s captain and the democratic crew. Despite the presence of a true captain aboard with the knowledge to safely steer that vessel home, it is doomed by a recalcitrant democratic crew who will not listen to expertise (Plato 1992, 488b–e). In politics, just as on Plato’s ship, including the broader public in decision-making allegedly adds nothing of epistemic value. Brennan (2017, 53) compares the votes of the demos to a polluting corruption that putrefies the contributions of the knowledgeable few. The mass public allegedly exerts this toxicity because they are incorrigibly ignorant (Achen and Bartels 2017, 12; Converse 2006; Kinder and Kalmoe 2017), irrational (Bartels 2005; Gaines et al. 2007; Taber and Lodge 2006), and motivated by tribalistic attachments to their parties and social identities (Iyengar et al. 2019; Mason 2018). Brennan (2018) argues that we would rightly object to an ignorant and irrational jury voting to determine our fates at trial (86), and voting in elections ought to be held to the same standard. The moral failing of democracy as a decision-making function is that it includes so much epistemic pollution that it cannot solve social problems, despite experts standing by with the answers ready in hand.
The epistocratic solution to the alleged failings of democracy is simply to filter away the “voting pollution” until the electorate is reduced to the knowledgeable few who are competent to make decisions. Brennan (2017) envisions a regime in which, “by default, no one is entitled or permitted to exercise any degree of political power . . . an elite electoral system requires citizens to earn a license to vote” (211). Qualification exams ought to filter for “social scientific knowledge such as economics, sociology, or political science—the knowledge needed to form sound policy judgements” (Brennan 2018, 164). This entails not directly empowering scientific experts. Instead, the filtering empowers a class of quasi-experts. These quasi-experts know much but do not produce knowledge. Here the knowledge to solve social problems already exists, and those who hold that knowledge are identifiable. Creating a decision-making procedure that can solve social problems simplifies to the task of aligning the distribution of power with that of extant knowledge just as it did for Plato (1992, 473d).
Democrats respond point by point to this indictment against the mass public. Voter ignorance may be overcome by informational heuristics (Lupia 2016; Lupia and McCubbins 2000), irrationalities at the individual level can resolve into rationality at the level of the collective public (Page and Shapiro 1992), and cross-cutting community attachments can counteract tribalistic impulses (Mason 2018). Additionally, democrats submit that aggregating the public’s votes often enables superior decisions than expert decisions taken in isolation. Thus, democrats contend that expert knowledges do not exhaust the relevant extant knowledge in communities, and it is impractical to identify all useful existing knowledges prior to decision-making. In their own nautical analogy, democrats, such as Landemore (2012, 20–21), point to Hutchins’s (1995) study of how navigation is a distributed cognitive process across a ship’s crew rather than the task of an individual expert/captain’s mind. They also charge that relying exclusively on experts has often been a fatal flaw in policy interventions (Levy and Peart 2016; Scott 1998). Fischer (2000) suggests that expertise alone is rarely sufficient and most frequently “there is no alternative to relying significantly on ordinary local knowledge” (212). Far from being a form of epistemic pollution, lay public knowledge is often a critical component in successful decision-making. Further, Landemore (2012) instructs that “because of the unpredictable and ever-changing nature of the problems that the community will have to deal with, the relevant knowledge, perspective, or information are simply unknown” (13). Taken together, democrats claim that we do not know what knowledges are useful in advance, but we do know that expert knowledges alone are not sufficient. The moral failing of epistocracy as a decision-making function is that it processes the existing knowledge of a community wrongly because it filters away distributed knowledge without accounting for its value.
If it is impossible to specify in advance where all the useful knowledge resides in a community, then a reasonable strategy is to incorporate as many perspectives as possible. This leads to the epistemic advantages of maximizing cognitive diversity. Ober (2008), for example, attributes the prosperity of classical Athenian democracy to its institutional capacity to aggregate the distributed knowledge of the demos. Landemore (2012) similarly contends that “it is often better to have a group of cognitively diverse people than a group of very smart people [experts] who think alike” (103). She illustrates this in her own juridical example of the film Twelve Angry Men. “Only by harnessing the intelligence and cognitive diversity of the other members, . . . does the group ultimately reach the truth. The contributions of each jury member vary and complement each other . . . deliberation enlarged the pool of information and ideas for all jurors, bringing to the surface knowledge” (Landemore 2012, 98–99). Contrary to epistocratic fears, a jury whose members are individually ignorant and irrational may still function competently as a collective intelligence. Importantly, though, that collective intelligence operates by assembling a broader scope of existing knowledge. Epistocrats and democrats disagree about the distribution of knowledge in society, though they are in alignment in taking knowledge-making for granted as an upstream process. As a result, both exclusively concentrate their efforts on contesting the proper means for composing a competent decision-making room.
Lingering Democratic Vulnerabilities
Some democrats, such as Bagg (2018) and Achen and Bartels (2017), lament that the above defenses of democracy are not wholly convincing. Bagg, for example, argues that the mechanisms supposedly producing democracy’s emergent intelligent properties appear to be fragile. The democratic deliberations that center in Landemore’s (2012) Democratic Reason, for example, require restrictive conditions such as participants agreeing on the nature of problems and pursuing solutions cooperatively (Brennan 2017, 180–85). In comparison, Goodin and Spiekermann (2018) offer a relatively resilient account of collective intelligence through aggregation. Alas, they also note that aggregation breaks down when opinion leaders’ messages are highly correlated with each other (ibid., 169), which indeed appears to be the case within the American political right’s “propaganda feedback loop” (Benkler, Faris, and Roberts 2018, 79). Bagg (2018) and Achen and Bartels (2017) offer what they describe as more realistic and robust defenses of democracy by first accepting the bulk of epistocratic criticisms. They contend that we may still prefer democracy because it provides an anticorruption mechanism. Yet even this defense is not unassailable, as Bell (2015, 117) argues that it is plausible for an epistocratic regime to also develop anticorruption mechanisms. On balance, we have many defenses of democracy, though none offer a “knockdown” blow to epistocracy. Therefore, it is reasonable to search for a defense of democracy that is simultaneously robust while also offering advantages beyond anticorruption mechanisms. One pathway forward to a new defense of democracy, as I argue here, is to change how we ask the democracy versus epistocracy question.
Learning versus Knowing
I contend the current way scholars have asked the democracy versus epistocracy question will not lead us to an answer because the frame is incomplete. The conversation has been fixated on how to compose decision-making bodies (electorates and juries), yet good decision-making capacities alone are insufficient to produce quality outcomes. Intelligent action requires strong knowledge- and decision-making. Advantages in knowledge-making can compensate for disadvantages in decision-making faculties. For example, at the level of individuals, we may reasonably expect a student with average cognitive faculties who studied to gain requisite knowledge for an exam will perform better than a cognitively gifted classmate who neglected to do the readings. Knowledge is power and can more than compensate for moderate deficiencies in decision-making faculties.
This effect is also true in politics and public policy. As I detail in the following section, there are substantial bidirectional feedback loops between political decision-making and expert knowledge-making. Given that endogeneity, our current analyses that centers exclusively on decision-making can lead to erroneous conclusions. An epistocrat, for example, may persuasively argue that a regime of quasi-experts is better equipped for decision-making. Yet, if knowledge-making effects from those decisions tend to censor or distort the data necessary for future policy decisions, then epistocracy will be an unintelligent regime in the long run. We may fully grant the epistocratic set of claims to be superior decision-making mechanisms for processing existing knowledge into policies without being forced to conclude that they are also a superior intelligent action system. The key point here is that assessing a regime’s capacity for intelligent action requires considering knowledge- and decision-making together. It is fully possible that a regime appears superior when considering decision-making in isolation, yet when we account for interactions with knowledge-making it performs worse than the alternatives. Thus, to adequately resolve the democracy versus epistocracy question, we must re-ask the question in a novel way that incorporates knowledge-making into consideration.
How Political Elites Shape Knowledge Systems
STS scholars have long recognized that knowledge- and decision-making exist in a feedback loop with each mutually influencing the other. Miller and Munoz-Erickson (2018) demonstrate this in their work on knowledge systems analyses. Presently, I draw on their work and contributions from other STS scholars to demonstrate, first, the bidirectional relationship between knowledge- and decision-making. Second, I inventory the formal and informal powers politicians possess to influence knowledge-making. Those powers compose their epistemic toolkits—the means—that politicians employ to strategically cater to their electoral coalitions.
Expert knowledge-making, in Miller and Munoz-Erickson’s (2018) work, can be divided into three core tasks: (1) generating new claims, (2) sorting claims based on quality, and (3) communicating claims to appropriate audiences (53). Generating claims involves processes such as data collection, data analysis, crafting theories, and composing publishable contributions (ibid., 54). Sorting claims for quality “consists of the practices, processes, and routines by which knowledge claims are subject to review, critique, assessment, check, and so forth” (ibid., 54). An immediate example of sorting claims for quality is peer review in academic publications. Communicating claims entails employing various forms of media to circulate knowledge to the people who can make use of it. This includes publishing governmental reports, contributing to popular scientific media, publishing scholarly articles, etc. The response by biomedical scientists to COVID-19 offers a concise example of these three knowledge-making tasks in action. During the pandemic, experts created new claims by identifying potential vaccine candidates. The FDA sorted those claims for quality through clinical trials to verify the safety and effectiveness of vaccine candidates. The development of those vaccines was then communicated to the public and elected officials. This knowledge-making process also equipped decision-makers with the knowledge necessary for intelligent action regarding public investments in vaccine production, distribution, and mandates.
As mentioned previously, STS scholars have consistently demonstrated that knowledge- and decision-making mutually influence each other. That is, “the relationship between knowledge and [political] power is bi-directional: (1) power shapes the organization and control of knowledge systems, their design, and their operations; and (2) knowledge systems and their knowledge claims shape the making and justification of decisions to use power” (Miller and Munoz-Erickson 2018, 99). Indeed, feedback between knowledge- and decision-making is common, in part, because they occur even without the intention of decision-makers. Policy decisions demarcating the size and scope of the welfare state, for instance, embed unintended knowledge-making effects in the world. Eubanks (2018) documents these effects with a case study of the implementation of a predictive algorithm by the Office of Children, Youth and Families (CYF) in Pennsylvania to guide interventions against suspected child maltreatment. That algorithm relied on all available government records on citizens. On subsequent analysis, it was discovered that the CYF algorithm was biased. It over-targeted the poor while neglecting child maltreatment by the affluent. This bias resulted because of previous decisions to means-test various government programs that generated the administrative data on which the CYF algorithm relied. To wit, by establishing government programs with aggressive means testing, legislators had also inadvertently means-tested the data generation process that would inform future policies. This demonstrates how decision and knowledge-making can unintentionally feed back upon each other.
The Politician’s Epistemic Toolkit
These bidirectional feedbacks between knowledge- and decision-making are also brought about intentionally. Politicians possess an epistemic toolkit at their disposal to influence knowledge-making, and they employ their toolkits to cater to the groups comprising their electoral coalitions. STS scholars have documented a wide variety of these mechanisms employed by politicians. I divide these powers into two broad categories. First, there are formal powers over knowledge-making that politicians possess as a result of holding office. The ability of Congress to initiate inquiries is one example of a formal knowledge-making power. Secondly, informal powers, such as directing funding to friendly think tanks, are available to politicians independent of holding office. The various powers inventoried here compose an epistemic toolkit that politicians employ to shape knowledge-making.
Formal Powers
First among the formal powers politicians possess to influence knowledge-making is their ability to mandate or forbid the production of specific knowledges. Fisher (2009) writes that when elected officials encounter unanswered questions, then “experts can be commissioned to supply [missing knowledge], a relatively common practice on the part of politicians” [emphasis added] (202). Indeed, Hess (2016, 33) elaborates on this point by highlighting how politicians are favorably empowered to shape knowledge-making because they can direct the state resources needed to meet the often-high costs of expert research. Sarewitz (2013b) contrasts the administrations of Presidents Bush and Obama to exemplify how elected officials wield their powers of office to structure research. Bush, for instance, banned most forms of embryonic stem cell research to satisfy religious conservatives in August 2001. Obama, in turn, reversed that policy two months into his first term as part of his campaign pledge to “restore science to its rightful place.”
Elected officials can also insert themselves into the gears of knowledge-making. Porter (1996) describes, for example, how Congress mandates specifically quantitative knowledge generation in the form of cost-benefit analyses in regulatory agencies. Jasanoff (1987) discusses how Congress has negotiated settlements over which evidence counts as properly “scientific” for regulatory decisions. Further, politicians also influence knowledge-making by creating and managing various bureaucratic institutions. Guston (2013) describes how these institutions “sit astride the boundary between politics and science and involve the participation of nonscientists as well as scientists in the creation of mutually beneficial outputs”. For example, Congress created the Office of Technology Transfer (OTT) to manage knowledge-making with priority placed on incentivizing commercially profitable research. The lesson here, as Miller and Munoz-Erickson (2018) attest, is that “legal and political work is an essential part of knowledge generation. Congress must write laws governing knowledge generation . . . and government agencies must, in turn, develop regulatory rule-making processes that determine exactly which data to collect and which methods to use” (emphasis added; 61).
Informal Powers
Politicians also retain informal powers to influence knowledge-making independent of holding office. Specifically, elites informally foster knowledge-making both by direct investments in research at think tanks and by raising issues to salience in the public sphere. Regarding the former, elites expend money on think tanks with an eye toward cultivating expert knowledge as an electoral resource. Rich (2004) quotes a think tank founder on this point thusly, “I just viewed [knowledge-making investments] . . . as a remarkable, highly leveraged way to participate in the national policy debate” (emphasis added; 55). Miller and Munoz-Erickson (2018) offer a similar observation, “the significance of knowledge in justifying the exercise of power can create incentives for multiple, diverse groups in society to invest heavily in their own knowledge systems . . . especially in political cultures like the United States that foster and catalyze political competition” (emphasis added; 100).
Elites also indirectly guide the flow of expert knowledge-making into electorally beneficial channels by raising issues for public attention. While experts are commonly stereotyped as the cloistered residents of the ivory tower, in practice scholars are frequently trend-following creatures. Trend-following behaviors flow naturally from the fact that experts, as Levy and Peart (2016, 199–200) remind, possess the same motivational drives as all other humans for relevance, social approbation, influence, and career advancement. Those motives draw them to research agendas that are salient in the public sphere as “science, after all, takes its problems from the society of which it is itself a part” (Fischer 2009, 203). Those natural trend-following behaviors are reinforced by institutional requirements, such as grant applications and peer review, that require scholars to give accounts of the potential social relevance and impact of their scholarship. Taken all together, politicians raise issues to salience in the public sphere based on expected electoral advantageous, and experts operate under the influence of that same public sphere. Scholarly research is certainly not totally determined by these influences, though, Rich (2004) finds that, “experts exist in greatest number in issue areas where there is support and demand” (140). In sum, politicians possess epistemic toolkits for influencing expert knowledge-making. Moreover, they use these tools as components in electoral strategies because politicians understand that “people are more willing to trust scientists than politicians” (quoted in Lee 2003).
Elite Motives & Opportunities Structures
To grasp how transitioning between democracy and epistocracy will influence knowledge-making, we need to unpack the key differences between those systems and how they interact with the motives of politicians. The central actors in this analysis are elected political officials. These elites are understood to be primarily motivated by the drive to gain power through electoral mechanisms in both epistocracy and democracy. To that end, they adopt strategies that are conditioned by the electoral context they operate within. Of interest to us is that those variable electoral strategies will include actions to influence the knowledge-making activities of experts. Simply put, elites can be expected to employ their epistemic toolkits to encourage knowledge-making that serves their ultimate purpose of winning elections. Political elites exist in a critical role as the mediators in the middle between the electorate and knowledge-making experts. Centering on elected officials as the key actors allows us to trace how changes in electoral environments alter the strategic considerations of politicians. In turn, we can further trace how changing contexts lead politicians to alter where/when they employ their epistemic toolkits.
Expert Knowledge as a Strategic Electoral Resource
Before comparing models of democracy and epistocracy, we need to consider why expert knowledge claims are a valuable strategic resource for politicians. For those elites, expert knowledge is first and foremost a tool for political communications. For instance, we must note the powerful modern trend of political actors enlisting the support of scientific forms of knowledge as resources in political competitions. As Pielke (2007) puts the matter, in contemporary politics “groups with otherwise conflicting interests each look to science to enhance their political standing. The result is that political battles are played out in the language of science” (10).
Expert knowledge confers multiple political benefits. When politicians appeal to identity groups in the electorate, for example, expert knowledge allows them to signal that our team is correct and the out-group is wrong because “science” agrees with us. Expertise here provides a mechanism for elites to validate partisan identities akin to other forms of partisan affirmation around policy conflicts (e.g., Mason 2018, 49). Expertise also allows politicians to impose closure on policy conflicts by declaring a question “solved” by science (Hess 2016, 60; Pielke 2007, 43). Further, placing communications in the rhetoric of science cloaks politicians in the cultural authority of science with its attendant connotations of rationality, logic, and empiricism. Thus, expert scientific knowledge is a highly valuable strategic resource in modern electoral competitions. As STS scholars observe, politicians understand that value and act accordingly (Miller and Munoz-Erickson 2018, 100).
Opportunity Structures: Democracy versus Epistocracy
Democracies are understood here as systems of competitive elitism with universal adult enfranchisement. In particular, I take Achen and Bartels’s (2017) Democracy for Realists as the basic blueprint for contemporary democracies. The mass public is understood to be largely politically ignorant, irrational, tribalistic, and only episodically politically active in the form of casting votes for competing elites. “Partisan loyalties and voting behavior,” as Achen and Bartels contend, are primarily driven by “social identities, group attachments, and myopic retrospections, not policy preferences or ideological principles” (emphasis added; ibid., 267). Social identity groups, here, are taken to be the core mechanism for explaining mass political opinions and behaviors (e.g., Klien 2020; Mason 2018; Miller and Conover 2015).
Importantly, the electoral strategies of politicians also accordingly work through those social identity groups. Achen and Bartels recount, “when political candidates court the support of groups, they are judged in part on whether they can ‘speak our language.’ Small-business owners, union members, evangelical Christians, international corporations – each of these has a set of ongoing concerns and challenges,” and politicians speak to these many identity groups to assemble winning electoral coalitions (ibid., 309). Among the ways that politicians speak to these social identities is by employing their epistemic toolkits to attend to specific groups. That is, our expectations for the epistemic toolkits are merely a special instance of our general expectations about the group-based nature of politics.
For the model of Epistocracy, I will center on one of Jason Brennan’s proposals. 1 I set aside non-electoral epistocracies such as Bell’s (2015) China Model to limit the scope of this discussion. In Brennan’s (2017) model of Epistocracy, we see that “the major difference between epistocracy and democracy is that people do not, by default, have an equal right to vote” (208). Epistocracy consists of the mass disenfranchisement of all citizens who do not pass a qualifying exam. The rump electorate retained after that winnowing will be composed of a small set of quasi-experts who hold sufficient knowledge of relevant social sciences. They are not, however, full scientific experts capable of contributing new knowledges. All other background institutions are anticipated to remain constant unless changes in them can be clearly linked to limiting the vote to quasi-experts. This means that (1) the public at large, though not the epistocratic rump electoral, remains socially diverse, (2) elites retain their epistemic toolkits, and (3) competitive electoral strategies must speak to the groups who are retained in the rump electorate.
We can identify two effects of epistocratic filtering that will alter the opportunity structures politicians navigate. These changes encourage them to alter how they deploy their epistemic toolkits. The first effect concerns Brennan’s (2017) prediction that winnowing down the electorate to an epistocratic rump will produce a concomitant contraction of the set of social problems that hold electoral value for elites to encourage knowledge-making to address (242). As Miller and Munoz-Erickson (2018) attest (100), political conflicts provide strong motives for parties to invest in generating useful expertise. To wit, politicians arm themselves with expert knowledge to win contests. Hence, simply reducing the frequency of conflicts should be expected to reduce the incentives politicians experience to use their toolkits.
The second and more critical effect concerns the demographic objection to epistocracy. That objection charges that tests to filter for social scientific knowledge or other markers of “competence” will tend in practice to select for privileged groups. Estlund (2009) provides the example that “having a degree [as a filtering criteria for competence] is disproportionately the privilege of members of certain races, classes, and (formerly) genders,” and this is problematic as “people are inevitably biased based by their race, class, and gender” (215). Epistocratic filtering homogenizes the electorate by overrepresenting privileged groups while underrepresenting minoritized groups. A politician seeking election before an epistocratic electorate is incentivized to prioritize encouraging knowledge-making that appeals to those privileged and enfranchised identities. Traditionally, this demographic objection has been raised to emphasize the threat of epistocratic biases in decision-making. I contend that this also biases expert knowledge-making. To recall the jury analogy, the traditional demographic objection focuses on how decisions are biased by who is in the room, and my addendum is that there are associated biases based on what evidence is created for the jury’s consideration.
Comparing Learning Regimes
Comparing epistocracy and democracy necessarily requires a degree of speculation owing to the simple fact that we lack extant electoral epistocracies for study. That said, in the preceding sections I have labored to compose an extensive assembly of political science and STS scholarship to reduce that final speculative component to a minimum. The epistemic toolkits elites possess as their means to influence expert knowledge-making are founded in the long-standing consensus of STS scholars concerning the bidirectional relationship between politics and science. Likewise, the motivations of politicians to gain elected office, their strategies for appealing to politically salient group identities, and the broader opportunity structure offered by democracy are grounded in empirical political science. Similarly, the opportunity structure presented by epistocracy follows directly from epistocrats’ own models.
The extrapolation I advance from this foundation is simply that politicians, who retain the same means and motives, will alter their strategic behaviors in response to confronting a homogenous electorate under epistocracy. While there remains a speculative element here, the leap involved is well-founded on existing literature. I would briefly invite readers to also consider two counterfactual possibilities here. Would we expect politicians to not tailor their strategies to a new electorate? Or for them to abandon the epistemic toolkits they currently employ in democratic settings? The answer to both, I expect, is no. Thus, with means, motives, and opportunities all in place, we can assess the epistemic consequences of transitioning between democracy and epistocracy.
Elite strategies are adopted as responses to specific electoral environments, and we ought to expect that altering the electorate will alter elite strategies and expert knowledge-making. From the perspective of elites, altering who can vote alters whose research questions are worth investigating. Altering what questions are worth investigating produces a change in the pattern of knowledge-making that occurs in society. Ultimately, I claim that the pattern of knowledge-making encouraged within democracies is more effective for guiding intelligent action to solve social problems. Democracy, by composing an electorate that (roughly) reflects the social diversity in the population, incentivizes political elites in the aggregate to foster a correspondingly diversified expert research agenda.
The Patterns of Knowledge-Making
Whose research questions receive attention or neglect, as Sarewitz (2013a) observes, is critically important in determining the social utility of scientific expertise. Alas, all questions cannot be investigated simultaneously, and therefore there is always a pattern to whose problems receive attention or neglect. Hess articulates in Undone Science that by default knowledge-making favors the concerns of privileged members of society. This occurs in part because expert knowledge-making is often prohibitively expensive for nonstate and noncorporate actors (Hess 2016, 33). This is readily apparent for grand projects such as large particle colliders, though even comparatively smaller-scale social science research can be costly; a module on the Cooperative Election Study (CES) can cost thousands. These pecuniary barriers produce epistemic injustices when the elites who command resources lack incentives to sponsor knowledge-making that addresses the concerns of minoritized groups. For example, throughout much of the twentieth-century, testing for automotive safety features routinely neglected to check their impacts on women’s bodies under the assumption that women were essentially “smaller men” and hence undeserving of additional research expenses (Perez 2019, 186).
This default toward epistemic injustices, however, is not destiny. Hess (2016), for instance, advises that egalitarian social movements and institutions can counteract that default. Democracies, by universally enfranchising diverse publics, can resist this default on the margins. This occurs, first, because a heterogeneous democratic electorate pressures politicians to spread their attention across many groups. We can see this drive to distribute attention across constituencies at the level of individual politicians in Fenno’s (1977) homestyle study. He captures this in his description of politicians who perceive themselves as confronting multiple plural constituencies in their quest for (re)election. Those politicians must divide their attention to ensure that no constituency necessary for (re)election feels persistently neglected. Politicians express a desire for the plural group identities in their districts to identify with the politician, to trust the politician is at least “on their team” or better yet “one of them.” The homestyle strategies they adopt to build that trust and identification vary across districts and politicians. A uniting feature, however, is that politicians in heterogeneous electorates experience an imperative to present themselves as caring and attending to many plural groups. This drive to diffuse attention across groups in democracies also plays out at the level of party competition as well. Parties in modern democracies are bundles of varied identities; they form to allow diverse coalitions to coordinate and share power over time. Interparty competition produces a division of labor as politicians specialize their efforts to attend to their coalitions. In aggregate, this means that politicians spread their attention across the groups that compose the overall electorate.
Critically, the necessity for politicians to diffuse their attention across the groups composing their coalitions applies to how they wield their epistemic toolkits. To wit, from the perspective of politicians, these toolkits are mechanisms for winning elections through political communications. We should expect politicians to strategically employ these tools to cater to group identities as they do with their other persuasive techniques. The empirical evidence assembled by Rich (2004, 140) concerning think tanks supports this interpretation. Think tanks are not distributed across topic areas at random. Instead, most possess identifiable partisan affiliations and are specialized for research on narrow topic areas that concern particular groups.
Democracy, then, requires politicians to diffuse their attention across the plural identities necessary to form winning coalitions. This diffusion fosters a concomitant epistemic division of labor among experts as politicians employ their epistemic toolkits to bend knowledge-making into the service of electoral ends. Moreover, the diffusion of attention allows democracies, on the margins, to resist the default toward epistemic injustices. This is because democracy provides an incentive structure for politicians who can direct the resources necessary to foster expert knowledge-making to topics concerning the groups who compose their constituencies. Hence, the aggregate national research agendas in democracies bend toward reflecting the diversity of groups in the electorate, and by imperfect extension the diversity of the population. In this sense, we can say that democracy aligns politicians, experts, and the public. The alignment is certainly imperfect, as politics is one of a plethora of influences on expertise. Yet, we can say democracy is superior to what we can expect from epistocracy.
Within an Epistocracy, we ought to expect this effect to be reversed. There, the electorate is filtered down to a rump of quasi-experts who are homogenous, privileged, and trusting of the status quo. More precisely, we can expect the enfranchised rump electorates within epistocracies will be far more homogeneous than the population as a whole both in terms of identity groups (Estlund 2009, 215) and political opinions (Althaus 1998). Put differently, an epistocratic electorate is composed very differently from the whole public by design.
Evaluating epistocracy, we can expect that the same basic rules of politics apply for epistocracy as they did for democracy. Again, the only variable we are changing is who can vote. Epistocratic politicians must appeal to the groups who remain well represented in the rump electorate by strategically catering to them with the tools at hand for political communications, which includes their epistemic toolkits. The expectation here will be for the elites to concentrate on deploying their toolkits to foster expert knowledge-making agendas that present clear electoral payoffs in terms of resonating with the quasi-experts who remain in the electorate. This does not, of course, mean that all knowledge-making to address issues of concern for the disenfranchised will evaporate into nothingness; epistocratic politicians may also be motivated by benevolence or curiosity. What matters, however, is that under epistocracy, those politicians lose their electoral incentive to mobilize the resources necessary to encourage expert knowledge-making for the disenfranchised. Thus, the distribution of epistemic labor that roughly corresponded to the public at large under democracy gives way, under epistocracy, to a narrow and homogeneous pattern of knowledge-making. The epistocratic pattern diverges significantly from various concerns of the public because the epistocratic politicians are aiming at a new target (the rump-electorate) that is distant from the public.
One item to emphasize here is that a diffused and diversified pattern of knowledge-making in a democracy offers greater social utility. The public has many varied social problems worthy of attention. Across those many problems, we can say that they will overwhelmingly tend to not be remedied when they are ignored. Recall that Sarewitz (2013a) contends that the social utility of research is significantly determined by whose questions receive attention or neglect. Put differently, extremely smart folks, like those in an epistocratic rump-electorate, cannot solve social problems they cannot see. Epistocracy, by removing the electoral incentives of politicians to diversify knowledge-making, constructs ignorance about the social problems for the disenfranchised. We can grant that an epistocracy may exploit existing knowledge more effectively, but in the long run it will degrade the pattern of knowledge-making necessary to solve social problems.
Alternative Epistemic Engines
Liberalism and capitalism are two additional mechanisms driving knowledge-making in contemporary societies. Is it possible that one or both are accomplishing all the epistemic work here and democracy is superfluous? I contend that no democracy is not epistemically superfluous. Instead, we can identify how democracy provides distinct contributions to knowledge-making. Additionally, the differences flagged here can be attributed to the varied incentive structures liberalism, capitalism, and democracy create, just as is the comparison between democracy and epistocracy.
Beginning with liberalism, the suite of individual liberties it affords citizens, especially freedom of thought and expression, is indeed conducive to producing knowledge. Open societies permit experts the freedom to explore the frontiers of knowledge to produce novel research. That negative liberty in an open society, however, is not a sufficient condition to generate a pattern of knowledge that addresses the social problems encountered by contemporary publics. This occurs because, again, expert research is commonly prohibitively expensive. Hess (2016, 33) argues that owing to these financial barriers, politicians and corporations are uniquely equipped to foster expert knowledge-making. What liberalism lacks that democracy supplies is an incentive structure for directing resources to research projects that address the varied interests of the public. Democracy provides that direction on the margin by affording political elites the electoral motivation to employ their epistemic toolkits toward aligning expert research with public concerns.
Capitalism provides a second epistemic engine for society. Corporate firms certainly possess the resources to subsidize expert knowledge-making, and profit-seeking furnishes firms with a motivational structure to guide research. Indeed, the output of firms in terms of innovative new technologies, such as AI systems, attests to the productivity of experts under capitalism. That said, we can observe again that democracy provides more and different knowledge-making contributions beyond what capitalism affords. That is, democracy and capitalism diverge in what knowledge-making they foster because their incentive structures are not identical. The research agendas that address “What is a profitable product?” for firms and “What is an electable message?” for politicians will often diverge. This divergence in motivations manifests in firms (1) neglecting questions that democracy attends to and (2) firms developing research agendas that contest the research of democratic institutions.
Firms acting under the competitive pressures of capitalism are selectively attentive to knowledge-making projects that promise potential profits and avoid projects that pose risks to the bottom line. Pamuk (2021, 37–38) provides a useful illustration of this tendency by recounting how corporations are selectively attentive to sponsoring safety research on genetically modified organisms (GMOs) that promise favorable conclusions, while strenuously avoiding research that may uncover new risks. Hence, when solving social problems cannot be readily monetized by corporations, then democracy provides an alternative way to foster research into questions that firms neglect.
To take one recent illustrative example of research that politicians foster that firms neglect is the Department of Justice investigating allegations of systemic racism within the Minneapolis Police Department (MPD) (U. S. Department of Justice, Office of Public Affairs 2023). This is politically useful research as it allows the Biden administration to signal to Black Americans who were over-targeted by the MPD that the Democratic Party is attentive to their group concerns. This politically useful research does not present an obvious opportunity for monetization that could attract the interest of profit-seeking firms. Elected officials ordering research into police misconduct also highlights that politicians and firms are differently empowered to conduct knowledge-making. For good or ill, state officials have a degree of coercive capacity in their epistemic toolkits that firms lack. The general lesson here is that democracy and capitalism produce different patterns of knowledge-making because they possess different incentives and powers. In particular, research is expensive, and firms that are under competitive pressure to be profitable will neglect socially important knowledge-making opportunities that lack (or endanger) opportunities for profits.
We can also observe that democracy and capitalism are contributing different forms of epistemic work by the pattern of firms and democratic institutions frequently advancing opposing research agendas. Michaels (2020) documents how a range of topics (diesel, plastics, carbon fuels, social media, tobacco, sugar, alcohol, opioids, etc.) have all been subject to vigorous contestation between competing slates of experts. The pattern here is that experts in state institutions perform research that addresses concerns raised by various subsets of the public, and in response, firms pay for expert knowledge-making to defend their products and practices. The point here is that the divergent incentive structures in democracy and capitalism produce correspondingly diverging patterns of knowledge-making. We can observe this clearly both in how the two engines will be selectively attentive to different topics and by how they often sponsor opposing research agendas when they attend to the same topics. In sum, liberalism and capitalism are powerful engines for knowledge-making in society. Yet, democratic knowledge-making mechanisms provide more and different incentives to structure knowledge-making than either liberalism or capitalism provides.
Conclusion: Intelligent Action in Democracy
Democracy possesses a clear advantage in its capacities for intelligent action to solve social problems. It holds that advantage because of how it aligns the interests of citizens, politicians, and experts to encourage a superior pattern of knowledge-making. Put another way, democracy shifts what questions experts investigate to address social problems impacting more consistently on a diverse range of citizen interests. Democracy is epistemically desirable, ceteris paribus, because it encourages experts to ask socially useful questions that correspond to the many plural identities and interests among the public. The long-run net effect is that democracies are more likely to have the necessary knowledge on hand for intelligent action to solve social problems.
A point to emphasize here is that advantages of democracy that I describe are located in an unexpected place. The existing discussion has operated under the assumption that if we simply composed the correct assembly of people in the decision-making room, then social problems will be effectively resolved. As a result, previous contributions have all orbited around the question of voter (in)competence. Here I move beyond the contest over voter competence to search for advantages in expert knowledge-making. I have attempted to flip the old discussion around. Instead of asking “What can expertise do to solve the problems of democracy,” I have inquired as to “What can democracy do for expert knowledge-making?” Democracy viewed from this new perspective is superior because it induces a beneficial alignment between citizens, elites, and experts.
Footnotes
Acknowledgements
I would like to thank Dr. William Scheuerman, Dr. Hamid Ekbia, Dr. Jeffrey Isaac, Dr. Volker Schmitz, Dr. Vivian Ferrillo, Dr. Simon Luo, and Erika Coe for their invaluable feedback on evolving drafts of this paper.
Declaration of Conflicting Interests
The author declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author received no financial support for the research, authorship, and/or publication of this article.
