Foresight, Epistemic Reliability and the Systematic Underestimation of Risk more

SYMPOSIUM Foresight, Epistemic Reliability and the Systematic Underestimation of Risk Joshua A. Miller and Steven Douglas Maloney “Liberal governments cannot plan. ” —Theodore Lowi Good government must balance demands of legitimacy with those of accurate inquiry about the common good. David Estlund’s work on epistemic proceduralism, culminating (to this point) with Democratic Authority, demands theories of democratic legitimacy respect this balance. Estlund compels his reader to doubt that either correctness theories or pure procedural democracy adequately manage this balance. Estlund moves the conversation on deliberative democracy dramatically forward by pushing all parties away from either extreme. Still, we wonder about Estlund’s view that there exist no instances when a political regime might need to tilt heavily towards correctness at the expense of legitimacy. Perhaps some events are so cataclysmic that their avoidance must trump both legitimacy on the particular decision and general reliability on questions of less importance. Ultimately, Estlund’s claims to balance the demands of truthseeking and democratic justification raise many principled objections and empirical challenges to institutional design even as they lay others to rest. Estlund claims that public inquiry into moral and empirical facts is authoritative only when it is subordinated to a general acceptability requirement, “no one has the authority or legitimate coercive power over another without a justification that could be accepted by all qualified points of view.”1 The general acceptability requirement ideally harnesses the preference aggregating and information-assessing functions of democratic institutions like voting, deliberating, and protesting while curbing elitist (“epistocratic”) procedures. Moral experts are not, by virtue of this qualification, political authorities capable of making generally obligating policies. Since our moral intuitions and judgments are generally not biddable by unexplained appeals to expertise, democratic polities that defer to experts are, he claims, confusing correctness with authority. This caveat preserves the authority by which even dissenting citizens feel obliged to obey or face just coercion. 54 The Good Society, Volume 18, No. 2, 2009 Estlund’s view is laudable for trying to force formal political theorists to consider epistemic demands, but it ignores the growing requirements of predictive inquiry necessitated by the state’s exclusive claim to protect the health, safety, and welfare of its citizens. We agree that a fair procedure is insufficient if it is not also likely to produce good policy: “it must count in favor of a social decision procedure that it tends to produce the better decision.”2 Still, a procedure that does this must successfully collect epistemically reliable evaluations of facts and norms. A functioning democracy must not only be able to determine basic fact-finding (that a flood or financial collapse is occurring or has occurred) and norm-discovery (to proclaim in retrospect that it ought to have been prevented.) A state must also prevent and prepare for future risks, to preserve its citizens’ health, safety, and welfare. Even more difficult, it must handicap the likely risk of its occurrence against the expenditure of limited resources for prevention or ameliorative response. Strictly speaking, this is not normative inquiry. There are facts of the matter about future states of affairs, but they can only be measured in probabilities, cautiously and with constant corrections as new facts come to light. Nonetheless, future events have normative implications such that prediction and prescription are often conflated. We fear that predictive and preventative inquiry is a kind of empirical inquiry for which majoritarian institutions are ill suited. It appears, then, that Estlund cannot split the difference between authority and accuracy: for some decisions, at least, he must choose. A. The Challenge “What if democracy got most things right, but none of the most important things?”3 There are some decisions a state can make whose effects outweigh all others. The risks associated with failure in these decisions overshadow considerations of legitimacy, acceptability, and accuracy in lesser matters: getting those decisions right · Copyright © 2009 The Pennsylvania State University, University Park, PA GS 18-2_10.indd 54 11/2/09 4:11:15 PM B E YO N D ‘ U T O P O P H O B I A’ must be the first priority of any polity if it is to survive to make more banal choices. In this way, all regimes are truth-seeking, and all citizens should accept institutions that get these decisions right the most often rather than trading off the possibility of cataclysm for some other goods like fairness. When considering cataclysmic risks, Estlund focuses on a list of what he calls ‘primary bads’: “war, famine, economic collapse, political collapse, epidemic, and genocide.”4 Estlund defends inclusive and open democratic procedures from the potential epistemic privilege of, for instance, voting systems that prefer the educated to the uneducated, or deliberations that exclude some viewpoints. His defense fails—at least situationally—if epistemic proceduralism is less likely than some form of epistemic privilege to generate the best policy when it matters most. Estlund rejects correctness theories because they violate the axiomatic principle of political authority by subordinating the hypothetical consent of the governed to epistemic standards—even if the correctness theory improves the likelihood of reliably generating the best policies. While it is persuasive for lesser matters, Estlund relies on the axiomatic status of the general acceptability requirement when it comes to “primary bads.” We applaud his commitment to fairness, but we believe the general acceptability requirement reproduces too large a gap between legitimacy and accuracy. Questions of primary importance, traditionally called constitutional essentials, historically include the separation of powers among institutions, major budgetary initiatives, and war-making power, and more recently, the correct platform of personal rights. However, it has become clear in the last half century that the actual scope of questions of constitutional import is larger than traditionally imagined. It is tempting to infer that the best institutional design will avoid cataclysmic decisions as often as possible. Unfortunately, this theory of constitutional essentials ignores the emergent nature of potential cataclysms. There is no guarantee that the constitutional essentials of the last century will be the emergent threats of the next, and avoidance of a question is just as likely to be the cause of a cataclysm as it is to avoid it. Estlund advances the view that institutional arrangements need only supply “sufficiently good competence”5 when maximally good competence in the avoidance of cataclysm seems preferable. For example, one strategy to limit cataclysmic failure would be to prevent any institutional actor, including the public, from instigating crises unnecessarily. On this logic, the fewer truly important decisions we make, the less often we will risk a cataclysm. Elections are, in part, attractive because they supply majorities with opportunities to express their preferences while frustrating their unreasonable demands. Ideal design would limit making cataclysmic decisions to only when necessary. In the rare instances that cataclysmic decisions are forced upon a polity, we hope to have arranged both experts and laymen so as to maximize their chances of success. Like good gamblers, we would only risk everything when it is absolutely necessary. Consider another example: all citizens are affected by the Federal Reserve funds target rate (the rate that banks charge each other for overnight loans to cover capital reserve requirements) as it ultimately determines the availability of credit and thus the balance between economic growth, inflation, and unemployment. Most experts agree that the range of viable options for this rate is limited. Further, they agree that direct or representative democratic control of the rate would encourage non-optimal outcomes, including price bubbles that could lead to economic collapse. As a result, decisions on the target rate, which affect every citizen, are nonetheless denied to the public. Some citizens thus argue that the Federal Reserve ought then to be abolished as illegitimate. Just as Estlund predicts, these citizens charge that members of the Federal Reserve Board, who are drawn from the management of a few investment banks, allow systematic biases for their home institutions to color their decisions. Thus, insofar as it makes (1) findings of fact (2) in an exclusive and closed manner that (3) have coercive effects on citizens because (4) democratic decision-making would lead tocataclysmic primary bads, it seems that Estlund would have to admit that the Federal Reserve Board does not meet the demands of the general acceptability requirement. Perhaps he would embrace this conclusion, as he seems to do with small state preference built into the US Senate’s disproportionate representation. However, we hope to give him reasons to avoid this admission, though we fear he must re-evaluate his theory of epistemic proceduralism as a result. To be specific, we propose that he consider the possibility that reasonable citizens must sometimes accept that decisions that may lead to primary bads be made in a manner that excludes them if they are to qualify as rational. B. The Emergence of Cataclysmic Threats to Constitutional Essentials When policy-makers underestimate systemic risks, and these risks explode, they can potentially damage or destroy the traditional rights-granting and public-goods distributing roles of the state. Call this the ‘emergence theory of cataclysm threats’: the institutions that constitute a regime may face as much danger from the concatenation of apparently superficial events and decisions as they do from fundamental design flaws. Regimes can be seriously undermined by apparently super-structural or ephemeral political decisions when those decisions ‘add up’ or exacerbate systemic risks to the breaking point. The phrase “constitutional essentials” derives from John Rawls, for whom it includes both “basic rights and liberties” and the “fundamental principles that specify the general structure Volume 18, Number 2, 2009 55 GS 18-2_10.indd 55 11/2/09 4:11:15 PM SYMPOSIUM of government and political process.”6 Constitutional essentials must be subjected to the strictest standards of public reasongiving, not primarily for epistemic reasons but so as to avoid falling into a modus vivendi in which parties bide their time while struggling for the power to enforce incommensurable policy preferences.7 For Rawls, constitutional essentials are the background against which ordinary political affairs are staged. They are not to be treated as ordinary political decisions. We might say that in his attempt to draw regime-justifying guidance from broad principles of justice, Rawls did not anticipate the sheer multiplicity of potential constitutional essentials, though of course he did note that constitutional matters often come under attack from locations far from the obvious foundations of a polity. Like an optical illusion, foreground institutions and background institutions are sometimes transposed, and the distribution of public goods becomes a threat to the separation of powers: a properly functioning market undermines our trust in public institutions or makes us wish that offices and honors were not quite so equally accessible. In light of the complicating interplay of plans, policies, events, and accidents, political theorists must add many other matters previously taken to be nonessential to the traditional list of constitutional essentials. These include: the tax structure (for instance: intergenerational shifts in inequality); market regulation (for instance: large private firms or novel financial instruments); the education and incentive structure of the administrative class; tariffs and trade (for instance: interest rates and current account deficits); immigration and labor policy. Decisions about any one of these “emergent essentials” may end up appearing to be, in hindsight, the crucial epistemic test of a government’s decision procedures. This emergent theory of constitutional essentials has at least three forms: skeptical, pluralist, and manageably-risky. Skeptics about the democratic quality of constitutional essentials fall into two camps. First, those who are concerned with what Peter Suber called “the paradox of self-amendment,” by which any truly self-governing polity must be understood to be free to change the laws that govern them as well as the rules for making such laws.8 Second, those who fear that democratic decision-making will underestimate systemic risks in economic, environmental, and cultural arenas. The “self-amendation” problem is non-trivial yet overly abstract. As it ignores the non-voluntary “spirit of the laws” that preserve political culture, especially the conservative and communalist elements, “self-amendations” are hampered by procedures that hold the promise of self-amendation but have 56 The Good Society largely shown themselves to be unusable in the 20th Century. In the US, the procedures for constitutional amendment (ratification and constitutional convention) have given way to a focus on alternative institutional means.9 In contrast, there is a sense of constitution tied not to the document but to the panoply of background institutions that constitute a regime. Unforseen systemic risks threaten unscrutinized regions of the economy and political culture. Innocuous questions of law, finance, or even education disguise potential cataclysmic risks. A radically skeptical theory of emergent constitutional essentials might hold that there is no hope to detect or predict these eventualities, and so we must build our institutions on risk-independent principles respondent to cataclysms only after they have become generally recognized. Against this view, the overall efficacy of research in the political science, economics, and sociology suggests that the skeptical position is untenable. Epistemic proceduralism subscribes to the pluralist account of emergent constitutional essentials. Estlund notes that what counts as a risk is subject to “reasonable” dispute, since a Christian might “reasonably” claim that the fate of a population’s immortal souls is more important than its capacity to avoid economic collapse or famine.10 The pluralist account argues that even among experts, there is reasonable disagreement on how to best evaluate environmental risks like climate change against economic costs like food shortages risked by corn ethanol, or human rights abuses in Iraq or Darfur against the outcomes of warfare. The skeptic believes a citizenry that cannot agree on constitutional essentials cannot constitute a stable or legitimate regime. If reasonable citizens prefer to dissolve their associations rather than be bound together, the regime becomes unstable. Thus, all supposed essentials are threatened with alteration as demographics change. More practical is the “manageably-risky” account of emergent constitutional essentials. Given the right preparation and the right data, we suspect that well-educated experts who belong to a community of fellow inquirers and embrace fallibilist, multivariable analyses can do a better job staving off cataclysms than the general populace. If this is true, we hope to prove that it would be unacceptable for democratic citizens to institute procedures that override these experts purely because they have the audacity to make invidious comparisons between classes of cataclysm-avoiders. Imagine you could choose between two societies that had a varying chance of making the right choice on some subset of policy-matters. For simplicity, they face only two types of GS 18-2_10.indd 56 11/2/09 4:11:15 PM B E YO N D ‘ U T O P O P H O B I A’ Boring A B 95% 10% Cataclysmic 10% 95% policy issues: frequent, boring decisions of low potential impact and infrequent, cataclysmic decisions that may well destroy the regime. Society B is clearly preferable to the Society A, because no matter how the distribution of decisions affects the overall reliability of the social decision mechanism, avoidance of cataclysmic failure is the condition of possibility for superior decision-making in boring matters. As a reasonable public, we ought to focus our attention on decisions with the highest cost of failure and the greatest benefits for success. Unfortunately, the terrain of policy-making is considerably rockier than this. No amount of risk-tolerance can outweigh the discounted affects of future cataclysms. Regardless of whether cataclysmic decisions are an annual matter or only come about every few decades, there is no number of balanced budgets that can justify a genocide. Nor can a state afford to focus on filling potholes while choosing to ignore a looming financial collapse. The consequences for deciding sub-optimally vary from decision to decision. Unbalanced budgets might lead to ethnic tensions that bring about the genocide, and crumbling infrastructure might lead to financial collapse. Fallibility extends to issue-identification, and we have no guarantee that we perceive the gravity of any particular decision correctly. One reason that epistemic proceduralism may be less reliable is the political economy of information sharing. The more serious the consequences, the more like that distortions will appear in the information economy which citizens draw upon to deliberate. This is likely so because more serious problems tend to be more complicated. Ideological positions tend to incorporate constitutional essentials, and thus provide intense, emotional issue identification on the matters that are the most contingent and subject to rapid change. The distribution of consequences are likely to be at their most uneven the more benefits or harms there are to distribute. Further complicating such decisions is the fact that the consequences may never be distributed evenly among deciders: many primary bads weigh unequally on some subset of citizens. Because epistemic privilege often travels with material and political privilege, those who are least privileged epistemically (either because they do not know or because they cannot make themselves heard) are often the most vulnerable to the consequences of bad decisions. Thus, we are not asking public decision-makers to decide correctly. We are asking them to devote their attention to finding correct decisions to supply optimal reliability on cataclysmic questions. This would likely include that decision-makers consult those who are most effected (and thus best able to report failures) but least likely to effectively participate (because excluded, ignorant, or irrational.) As a result, issue selection and epistemic justice must be built into cataclysmic decision procedure in a consequentialist way that prioritizes results over procedural legitimacy. The sky is not constantly falling, but it is constantly at risk of falling down. What crystallization of the many minutiae of politics might combine to bring about its downfall is unknown. Ideal procedural theory, even when epistemically reliable, is not reactive enough to meet this challenge. C. The Scope of Epistemic Privilege In order for Estlund to make his case, he must not only show that some fair procedures are better than others at achieving the best policy, he must also show that no unfair procedure is better at avoiding cataclysm. Where he spends most of his time dealing with formal and procedural accounts of politics that forgo epistemic reliability, Estlund lumps all unfair but epistemically superior political theories into the category of correctness theories and offers a comparatively few criticisms. Here, we respond to four of them. 1. The Prevention of Primary Bads Must Be Generally Acceptable Estlund argues that experts are not bosses, and that the best reason to prefer democratic institutions is that their policies are authoritative.11 It is not enough to be right: a coercive order must also be recognized as a legitimate command if we are to avoid thinking of the state as a particularly clever hostage-taker. For a law to be authoritative, it must obligate the consent of those whose obedience it demands, such that their refusal to consent violates a prior norm. This norm is the requirement of general acceptability to all qualified points of view. Estlund’s defense of the general acceptability requirement requires that qualified citizens be insular, that they recognize each other’s qualifications to accept or reject a given institutional arrangement.12 This mutual recognition functions as a foundational norm for qualifying to be considered. As such, Estlund depends on a version of the “reasonable rejection” criteria. While he quibbles on the term “reasonable,” Estlund does adopt a pluralist version of contractual rejection.13 Specifically, he adopts the view that authoritative institutional arrangements must pass a veto by many plural interests. This tends to lead to less-than-optimal arrangements for accomplishing particular objectives, but the hope is that meeting with general acceptance will achieve better policies generally in order to prevent primary bads. This generalism is Volume 18, Number 2, 2009 57 GS 18-2_10.indd 57 11/2/09 4:11:15 PM SYMPOSIUM untenable when it responds sub optimally to particular questions on cataclysmic matters. Epistemic proceduralism succeeds insofar as it is supported by a political economy that can generate a decisive subset of the deliberating population that is epistemically reliable. Estlund rejects Condorcet’s Jury Theorem, in which epistemic reliability emerges statistically from a large population of people who are individually better than a coin flip.14 The combination of deliberating and deciding by vote justifies Estlund’s claims on the epistemic reliability of democracy. If voting alone could accomplish this, then Estlund would be able to adapt Condorcet-style formal voting analysis to his ends. If it were just about deliberating, then Estlund would have to accept a deliberative dictatorship as equivalent to deliberative democracy. If there is an epistemically reliable set of deciders in any population, why are they decisive sometimes but not all the time? For one thing, when we make decisions rooted in deeprooted stereotypes, habitual interpretations, or ideologies that are systematically incorrect, we find ourselves making judgments correctly at rates that can be worse than a coin flip.15 We may also be mistaken based on imperfect information or an imperfect perception of the information we have. Deliberation cannot always overcome such biases, yet for most decisions it may be the best method available. When open and inclusive deliberative voting performs less reliably than other methods, however, epistemic proceduralism may grant the authority of less democratic methods. By focusing on voting and deliberation, Estlund downplays the importance of the coercive rule-making of largely unaccountable administrative spheres. Formal and informal realms of democratic politics are only orthogonally related to the most important institutions of governance. Epistemic considerations quickly lead to the conclusion that the realm of true constitutional essentials frequently is, and always ought to be, exempted from the general acceptability requirement. Constitutional essentials that remain subordinate to general acceptability are so subordinated at great potential risk. Professor Estlund may respond that we have merely repeated the general acceptability requirement, showing for one particular set of decisions the way in which general acceptability can be applied to itself. He could reiterate his claim that democratic procedures are not an end in themselves, but rather a means to the truth. In one sense, this is correct. Since the survival of the mostly fair and epistemically reliable regime is needed for improvements in fairness or epistemic reliability to be realized, 58 The Good Society no rational, qualified citizen would choose a regime that was less likely to survive because it avoidably stumbled into one of the catastrophic regime busting decisions. However, in precisely that sense, Estlund must acknowledge that it is not enough to be better than a coin flip. Tolerating an unnecessary likelihood of cataclysm is all it takes to be unreasonable, to have one’s objection to a biased but reliable regime disqualified. Thus, any governance structure that is inferior when it comes to cataclysms is reasonably rejectable because anyone whose contributions are likely to count against such governance ought to be excluded. 2. Epistemic Reliability Vis-à-Vis Cataclysm Trumps Bias Estlund rejects procedural privileges for the epistemically reliable cataclysm-avoiders, because the educated are not prejudicefree. Estlund writes that we cannot accept such biases even if they tend to produce wiser policies in general, because general wisdom may conceal particular prejudices.16 Because educated voters may tend to discount the political harms done to some subsection of uneducated voters, we cannot allow any invidious comparisons between better and worse policy-makers to be ensconced in our policy-making procedures. Even though we must all grant that a better education (somehow conceived) improves the ability to rule wisely, it is not unreasonable or disqualified to suspect that there will be other biasing features of the educated group, features that we have not yet identified and may not be able to test empirically, but which do more epistemic harm than education does good.17 Estlund gives two further concerns inadequate attention. First, a particular expertise (as an engineer, for instance, or in classical political theory) may not translate into general wisdom. This is the problem of the public intellectual, who tends to mistake a specific insight for expanded competency in all public matters. Second, inclusive majorities may contain worse biases than an educated subset. This is a special case of the problem of public ignorance. Many such biases can be countermanded by rightsbased trumps to maximally-inclusive decision-making, but not all of them can be. The arguments against supplying procedural privileges to the epistemically privileged conflate ordinary concerns of justice with concerns over constitutional essentials. Just as education can improve a population’s epistemic reliability with GS 18-2_10.indd 58 11/2/09 4:11:15 PM B E YO N D ‘ U T O P O P H O B I A’ regard to some of the highly technical questions of governance while leaving reliability on other axes untouche majoritarian decision-making can improve the legitimacy of some decisions while leaving other axes untouched. Where deliberation may grant decisions general acceptability, it may also be the case that majorities have “biasing” features that make them less qualified for preventing “regime busting” cataclysmic failures than experts or the educated. Yet while political and civil rights serve to protect minority interests from some prejudices, there is not a similar juridical mechanism for protecting the population from systemic risk. Unlike a rights violation, a systemic risk does not announce itself as an outrage to everyone with a conscience. Despite our preference for inclusion, we’d still prefer that an expert inspect our parachutes. maintains that voter ignorance prevents them from synthesizing partial views of their own interests with the policies that will lead to the best possible outcome.19 He thus concludes that elite administrative institutions are the best policy-makers and democratic institutions must be designed so that free speech and administrative control are maximized. Obviously, this is a brand of epistocracy, which Estlund rejects. Estlund’s counter-argument to Schumpter has two steps. First, he denigrates the importance of voting and deliberating because “voters other than officials do not often vote on binding policy decisions directly” though he does acknowledge that “a smart decision about whom to vote for would seem to require some information and understanding of the merits of policy proposal, since these will be some of the major reasons to prefer one candidate to another.”20 Estlund asserts “all the official decisions [of regulators, commissioners, and appointees] must be counted as among the democracy’s decisions.”21 Yet this is absurd. Some of the most powerfully coercive administrative agencies are almost completely unaccountable to voters and their representatives. Like the Cyclops blinded by Odysseus, those coerced by the Federal Reserve might as well curse the gods as the agents of their coercion. The bureaucratic structure of the administrative state has indelibly ensconced invidious comparisons into the rule of law. Second, Estlund analogizes politics to parenting, listing an impressive number of questions upon which the average parent is ignorant, and thus potentially worse than a coin-flip, and then concluding that though experts would likely do better than parents, we cannot select these experts without violating the demands of general acceptability by making invidious comparisons between (for instance) developmental psychologists or medical professionals and ordinary parents. Estlund has thus covered his bases: (1) voters don’t matter much and (2) insofar as they do they can’t mess things up too badly. When added to his argument about the uncounted biases of the educated, he seems to be suggesting (3) that anyway the next best group is often even worse. Given the role that experts play at the state level in providing product and food safety, and in preventing abuse and neglect, Estlund’s example of parenting is a curious choice. Experts frequently intervene when parents’ ignorance threatens the health, safety, or welfare of a child. State paternalism in this case is not reasonably rejectable. Perhaps for legitimacy reasons, Estlund’s conclusion is supportable when considering decision-making generally, just as most non-cataclysmic parenting outcomes are justified generally. Yet Estlund’s theory assumes that epistemic proceduralism’s decisions will be generally distributed amongst decisions that are equally important. In instances where decisions are potentially cataclysmic, specialization and exclusivity are still warranted. Volume 18, Number 2, 2009 59 3. No Reasonable Pluralism Can Justify Ignoring Primary Bads Estlund suggests that pluralism prevents rejectable claims to expertise when he points out that some groups of citizens advance accounts of unmitigated risks that do not meet with the general acceptance of the rest of the population. The example he supplies is the Christian account of damnation. Since the potential costs of eternal damnation are so high, a Christian citizen would be justified in insisting that the regime devote its attention to this question. Yet many other citizens would complain that this concern is a distraction. In the same way, a group of concerned scientists and environmental activists might insist that the most pressing cataclysm is the instability that will result from continued carbon emissions, while other citizens claimed that the underlying phenomenon of global warming is a propagandistic effort to seize control of the mechanisms of issue selection. In both cases, coercive measures are called for to prevent a cataclysm. The weighing of cataclysmic risks by their potential to become real is a decision the political regime cannot escape. No matter how illegitimate this seems to non-experts, this subset of decisions cannot be shirked or left to sub-optimal decision procedures. Estlund comforts us with a theory of epistemic proceduralism that ignores equally the privileged claims to expertise of both climate scientists and theologians, but if either group is right we ought not to be so comforted. In response to these concerns about incommensurable cataclysmic concerns, Estlund’s rebuttal to Joseph Schumpeter’s famous argument against the existence of a common public interest is telling.18 Schumpeter argued that voters’ interests are ultimately at odds because they are enacting policies designed to distribute finite public goods. Even if some outcomes are preferable for all (because, perhaps, some policies increase the total supply of public goods available for distribution) Schumpeter GS 18-2_10.indd 59 11/2/09 4:11:15 PM SYMPOSIUM 4. Inclusion Produces Epistemic Attitudes Ultimately, Estlund’s concerns about experts aren’t just procedural: he also worries that they aren’t actually that smart. He cites Philip Tetlock’s survey of pundits, academic experts, and policy makers showing that they are often less reliable than a coin flip.22 This is why he emphasizes the epistemic value of the attitudes cultivated under epistemic proceduralism, and downplays the reliability of the procedures as practiced in the ‘real speech situation’ by the ‘actually existing’ public. Yet Estlund also acknowledges that voters tend to be ignorant, that policies are often made by officials independent of elected politicians and traditional forms of democratic accountability, and that opinion leaders and pundits tend to do worse than a random prediction of the likely effects of policies enacted by the state. In the face of this incompetence, irrelevance, and the clownish efforts of our opinion leaders, Estlund manages to navigate safely between the shoals of despair and the rocks of complacency. From whence does he draw his hope? Here, Estlund’s defense of epistemic proceduralism can be better understood as an account of fair epistemic culture, one that supplies the administrative class with well-educated bureaucrats, a vibrant and useful informal public sphere, and deflects public disapproval of the policies decided upon under the qualified acceptability requirement through widespread lip service to inclusive procedures that do not actually steer the mechanisms of the state. We offer a compromise and a caveat: while inclusion produces epistemic attitudes (especially a preference for fallibilist experts), administrative governance harnesses these attitudes and is not beholden to them. As such, we believe Estlund must embrace a ‘correctness view’ when it comes to primary bads, and that he must expand his list of primary bads to include ‘emergent constitutional essentials.’ In part, Tetlock’s study bolsters this view, since it also demonstrates that some experts, those who reject “grand schemes” or theoretical parsimony and the ones that embrace their own fallibility, are much better than average. Individually, experts and public intellectuals are foolish: collectively, they might be the least fallible group available. in showing that there is less tension between fairness and truth than either pure proceduralists or strong epistocrats would like to believe. We agree with Estlund that it must count in favor of a procedure that it tends to produce the best decision and that some decisions are more important than others. All that remains is to resolve the empirical question of whether democracy is the best means of satisfying these requirements. Thus we must turn our attention from the passionate world of democratic deliberation and focus on the technocratic institutions of the administrative state. Estlund does not wish to legitimate the authority of such institutions—we suspect that he must. Our conclusions are elitist, exclusive, and troublesome. We understand why Professor Estlund resists them. We wish we could resist them as well. Joshua A. Miller is a Ph.D. Candidate in the Department of Philosophy at the Pennsylvania State University. His recently completed dissertation is entitled, “Hannah Arendt and Deliberative Judgment.” He is also an Adjunct Professor at George Washington University. Steven Douglas Maloney is a lecturer at the University of Saint Thomas. He recently co-authored “Justice is Hard, Let’s Go Shopping!: Trading Justice for Efficiency in the New Aggregate Settlement Regime” in the Saint John’s University Law Review. Endnotes 1. David Estlund, Democratic Authority (Princeton: Princeton University Press, 2008), 33. 2. Ibid., 98. 3. Ibid., 160. 4. Ibid., 163. 5. Ibid., 235. 6. John Rawls, Justice as Fairness: A Restatement (Cambridge, Massachusetts: The Belknap Press of the Harvard University Press, 2001), 28. 7. See Ibid., 192. 8. Peter Suber, The Paradox of Self-Amendment: A Study of Law, Logic, Omnipotence, and Change (New York: Peter Lang Publishing, 1990). 9. For various accounts of how and why this has happened, see Bruce Ackerman, We the People: Transformations (Cambridge, MA: The Belknap Press of the Harvard University Press, 1998), Theodore J Lowi, The End of Liberalism: The Second Republic of the United States, Second ed. (New York: W. W. Norton, 1979), Michael J. Sandel, Democracy’s Discontent (Cambride, MA: The Belknap Press of the Harvard University Press, 1996), Mark A. Graber, “Why Interpret? Political Justification and American Constitutionalism,” The Review of Politics 56, no. 3 (1994). 10. Estlund, Democratic Authority, 50. 11. Ibid., 22. 12. Ibid., 55–56. 13. Ibid., 44. 14. See Chapter 12 of Democratic Authority. D. Conclusion If some political decisions are really so important that mistakes entail the destruction of a relatively attractive political regime, it is more important to decide correctly at the expense of trivial matters. Of course, accuracy in trivial matters is part of what makes a regime attractive in the first place, but this is only when we compare relatively stable regimes. Epistemically reliable deciders in cataclysmic contexts deserve privilege and a constitution that provides fertile soil for them to grow. In this essay, we have found ourselves making arguments that give us a deep discomfort. Estlund has done a remarkable job 60 The Good Society GS 18-2_10.indd 60 11/2/09 4:11:15 PM B E YO N D ‘ U T O P O P H O B I A’ 15. “If you are to ask, ‘How could a person be worse than a coin flip?’ the answer would be ‘easily.’ People have more or less systematic views about many issues. If their system is bad, so to speak, then they could easily be wrong all the time.” Ibid., 16. 16. Estlund advances this argument in response to John Stuart Mill’s proposal that votes be weighted in value by educational attainment. 17. Estlund, Democratic Authority, 222. 18. Ibid., 26–27. 19. See Joseph A. Schumpeter, Capitalism, Socialism, and Democracy (New York: Harper and Row, 1976). 20. Estlund, Democratic Authority, 260. 21. Ibid., 161. 22. Philip Tetlock, Expert Political Judgment: How Good is it? How Can We Know? (Princeton: Princeton University Press, 2005). Volume 18, Number 2, 2009 61 GS 18-2_10.indd 61 11/2/09 4:11:15 PM