Article body

Consensus holds a preeminent position in philosophy (Rescher, 1993), especially within liberal theory of political legitimacy (for a critical discussion of the importance of consensus in political theory and a defence of compromise, please refer to Bellamy, 1999; Thrasher & Vallier, 2013; Vallier, 2011). Rescher characterizes consensus as a “widespread and pervasive agreement” (Rescher, 1993, p. 45). Merriam-Webster defines it as “a general agreement about something,” “an idea or an opinion that is shared by all the people in a group” (“unanimity” is offered as a synonym).[1] Consensus-oriented theories claim that political decisions and institutions are legitimate insofar as all citizens (would at least hypothetically) consent to them in one way or another.[2]

The underlying idea is that the legitimacy of political decisions and institutions depends on reasons and justifications that should be understandable and hypothetically agreed on by all citizens (Vallier 2011, p. 262). The influence of consensus is visible in concrete decision-making procedures too, such as United Nations General Assembly resolutions, Occupy Wall Street decision-making procedure, and Quakers’ decision-making procedure. The Nice Treaty (2001) set consensus as the decision rule for the EU Council before the Lisbon Treaty (2007) replaced it by qualified majority.

Such influence is visible not only in consensus-based mechanisms, but also in the positively valued proximity in political theory between consensus and other principles or procedures, most notably majority rule and its variants. In liberal-democratic theory, part of the justification for majority rule is that, when consensus is impossible, the best alternative is to approximate it by getting the consent of as many individuals as possible or of a majority. Qualified majority is a closer approximation than simple majority and, thus, strengthens the legitimacy of ensuing decisions.

In economic parlance, consensus appears as the first best for public decision making for part of the liberal tradition (equivalent to the Pareto optimum). According to mainstream economics, even if the first best is unattainable because one of its conditions is constrained, achieving other conditions would still be desirable. Thus, if we accept the relevance of economic concepts for political theory, and the possibility that consensus could be presented as a first best, then a manner to look at situations where consensus is not reachable is to call to the second-best theorem (TSB hereafter).

The general theorem for the second best optimum states that if there is introduced into a general equilibrium system a constraint which prevents the attainment of one of the Paretian conditions, the other Paretian conditions, although still attainable are, in general, no longer desirable.

Lipsey and Lancaster, 1956-1957, p. 11

The TSB captures the idea that there are situations where it is preferable not to try getting as close as possible to the first best or to try fulfilling as many first-best conditions as possible (Gaus, 2016, pp.14-15). As later noted by Richard Lipsey, “a ‘second best situation’ referred to any situation in which the first best was unachievable” (Lipsey, 2007, p.352). Another point needs to be stressed: the departure should be general (i.e., all first-best conditions dropped).

Authors have underlined the fecundity of the TSB for political, moral, or legal theory (Coram, 1996; Estlund, 2014; Gaus, 2016; Lipsey, 2006; Margalit, 1983; Räikkä, 2000; Rebello, 2002; Rodrik, 2008; Sher, 1997), but often in brief mentions without elaborated discussion of how to interpret Lipsey and Lancaster’s theorem. Attempts at deciphering TSB’s meaning and implications for political theory have been rare [Juha Räikkä (2000) is one of the very few exceptions].[3]

This article is an exercise in applied political theory that partially addresses such a lack by starting from Lipsey and Lancaster’s formulation and tackling an original issue: consensus as a liberal first best. It initiates an evaluation of TSB’s usefulness for liberal legitimacy, without providing a full-fledged analysis. The goal is to contribute to the few initiatives that aim at evaluating the promises and challenges of importing the TSB in political theory. Thus, I hope to contribute to discussions about the process of importing the TSB itself, the relevance of the theorem for political theory (which question remains under-studied) while applying the theorem to a specific field: political legitimacy.

In the first section, I argue that consensus stands as the first best for part of liberal theory. In welfare economics, the first best is the Pareto optimum[4] resulting from the fulfilment of “first-best conditions.” Microeconomics textbooks introduce perfect competition as the result of conditions (homogeneity of goods, atomicity of actors, perfect information, free entrance/exit from the market, and perfect mobility of productive factors). First-best theories stipulate that if one condition is unattainable, the others should be fulfilled anyway. For example, if homogeneity of goods cannot be achieved, fulfilling the other conditions will still enhance social welfare. Structural reforms imposed by the International Monetary Fund on developing countries follow this premise: if all first-best conditions for perfect competition cannot be met, it is still preferable to implement some rather than none (Rodrik 2008).

The TSB is useful because it provides the vocabulary for identifying consensus as the first best for part of the liberal tradition of political legitimacy. According to this tradition, if the consent of all individuals on a complete set of propositions (which is the only condition retained here for the discussion) cannot be obtained, it is still desirable to approximate it. Two versions have been popular among liberal theorists: tacit and hypothetical consent. John Rawls’s overlapping consensus is another illustration of this strategy of approximating a full-blooded consensus in the original position through an agreement on a deflated set of conditions.

The second section builds on Richard Bellamy’s criticism of the central role of consensus in political philosophy, including liberal theory. His criticism tackles the assumption of and the search for consensus on moral matters in pluralistic societies. My aim is to show how consensus raises similar issues regarding factual and epistemological disagreements.[5] A liberal society depends on consensus not only for political matters, but also for facts and epistemological rules.

The third section is about reasons for taking seriously factual and epistemological disagreements. The territorial claims of Canadian First Nations illustrate how such disagreements may challenge consensus-based theories of legitimacy by creating second-best issues. Such disagreements are endemic in liberal democracies and, more importantly, they offer reasons for modifying epistemological standards, especially for stability reasons, which raises second-best issues.

The fourth section presents conditions under which factual and epistemological disagreements undermine a liberal society based on consensus as the first best. The debate on GMOs illustrates such a challenge. I claim that considerations of trust and stability offer a solid ground for considering that factual and epistemological disagreements may create second-best issues for consensus-based approaches to legitimacy. Then, I present two issues that political theorists interested in “importing” the TSB into their field need to address. The first has to do with identifying a general departure from first-best conditions that is liberal. The second is to take seriously stability as a constraint on potential second bests. I end the article by noting that the TSB intersects with debates on the value of contextualism and pragmatism in political theory.

1. CONSENSUS AS LIBERAL FIRST BEST

As shown by Bellamy (1999), some liberals (and other political theorists) value consensus as the first best for the legitimacy of political decisions, institutions, or regimes. This value implies that when the consent of all is out of reach, the inclination is to approximate it. This also implies that political decisions or institutions become more legitimate the closer they get to consensus (which explains, for example, that a qualified majority appears to offer a stronger legitimacy to a political decision than a simple majority).

Few words about this process of approximation are necessary here. On matters of political legitimacy, the first best could be understood as encompassing consensus (i.e., everyone truly agrees/consents to all features of a decision, policy, institution, or regime). Hence, the first best actually has two dimensions: extension (everyone truly agrees/consents to a decision) and depth (everyone agrees upon all the features of the decision). Then, if consensus is out of reach, approximation could be to obtain the consent of either as many people as possible on a fixed set of propositions (e.g., majority rule) or all people on an altered set of propositions (e.g., compromise or Rawls’s overlapping consensus).[6] My discussion will mostly focus on (without being restricted to) the extension dimension. Consensus could also be used hypothetically—that is, as a justification for political principles (for instance, under idealized and counterfactual conditions).

This does not mean that consensus is the core of all liberal conceptions of legitimacy. It just means that consensus, like freedom or equality, is a central concept for many liberals. The liberal attachment to consensus flows from a commitment to individual autonomy—more precisely, to a regime of equal liberty, where every citizen has access to the same maximum level of fundamental liberties, including political ones (e.g., expression, autonomy).

Consensus is about individuals’ consent—that is, a situation where all individuals agree to a decision or a proposition. In a political regime, it applies to the justifications of public decisions and institutions. Indubitably, some conceptions of political legitimacy are independent of individual consent or agreement, such as conservative conceptions that are based on tradition and religion, or technocratic conceptions that are based on expertise or bureaucracy. But the specificity of liberal legitimacy is to be rooted in the fundamental idea that, ultimately, political decisions and institutions are legitimate insofar as individuals who are affected by them accept them in a less or more formal manner (i.e., consent or agree to them).

Then, consensus characterizes situations where all the parties to the decision-making process agree on or consent to decisions, institutions, or political regimes. Additionally, consensus could be interpreted as the agreement or consent of all bearing on all the features of the decisions, institutions, or political regimes. (This paper mostly focuses on the first dimension.)

Jeremy Waldron clearly expresses this importance of the consent of all for liberal theory.

Liberals are committed to a conception of freedom and of respect for the capacities and the agency of individual men and women, and that these commitments generate a requirement that all aspects of the society should either be made acceptable or be capable of being made acceptable to every last individual.

Waldron, 1987, p. 128, emphasis added

One of the reasons why, within a liberal society, social rules may not be seen to be as restricting of individual freedom as in other political regimes lies in consensus qua consent of all. The fundamental liberal thesis is that “a social and political order is illegitimate unless it is rooted in the consent of all those who have to live under it” (Waldron, 1987, p. 140, emphasis added). In other words, consensus grounds the legitimacy of political decisions and institutions in a liberal order (i.e., regime or society).

The idea could be found in Locke’s Second Treatise (II, section 95), where individual consent is placed at the heart of liberal societies: “Men being, as has been said, by nature all free, equal, and independent, no one can be put out of this estate, and subjected to the political power of another, without his own consent” (Locke, 2003, p. 141). Even if Locke immediately introduces the majority rule for enacting the consent of all after the initial establishment of the political community, which confirms the status of the majority rule as a substitute for consensus, the foundation of the political community and the ultimate justification for the political regime yet lie in the consent of all.[7]

The idea of political legitimacy as founded in the consent of all is the backbone of contractarianism (Vallier, 2011). It expresses itself through the principle of popular sovereignty (e.g., as originated in Hobbes, Locke, and Rousseau).[8] Indeed, consensus can be interpreted as being the first best—the optimum that serves as a benchmark for suboptimal alternatives—of legitimacy for part of the liberal tradition.

As a practical matter, consensus is rarely reachable and few theorists seriously consider that legitimacy requires the actual, explicit consent of all to any democratic decision or institution. They have instead taken one of these two paths: tacit or hypothetical consent. Both are altered versions of consensus, retaining some of its features. Their value in terms of legitimacy largely stems from the extent to which they approximate the first best.

Still in his Second Treatise (II, 119), Locke (2003, pp. 152-153) contrasts express and tacit consent. If the former is the unmistakable expression of one’s liberty, the second is simply assumed when individuals benefit from the security offered by the government. The rationale behind tacit consent is quite clear: when actual consent of all is impossible to obtain, the strategy is to retreat to a solution that approximates the first best.

The problem with tacit consent, in the absence of further qualifications, is that it may express less consent than imposed choice or restricted option, especially in non-ideal settings. Moreover, tacit consent alone (i.e., unqualified) is too loose a criterion for political legitimacy. It is too encompassing. If individuals can be shown to benefit from the security offered by an, even totalitarian, government, they could be assumed to be tacitly consenting.

The second path is hypothetical consent. In that case, there is no need to identify some eventual benefit. Consent is assumed under some hypothetical conditions and settings: political decisions and institutions are legitimate if it can be demonstrated that individuals would consent to them under specific conditions [e.g., of rationality (Kant, 1996)].

John Rawls’s original position (Rawls, 1971, p. 13; Rawls, 1980) illustrates hypothetical consent. The original position is a device that hypothetically generates the consent of all (or shows that such consent is a reasonable assumption) on the principles of justice that regulate the basic structure of the society. From A Theory of Justice to Political Liberalism, the role of consensus is not restricted to the original position and the “values of political justice”; it includes the “values of public reason” too (Rawls, 1993, p. 224). The “liberal principle of legitimacy” stipulates that

our exercise of political power is fully proper when it is exercised in accordance with a constitution the essentials of which all citizens as free and equal may reasonably be expected to endorse in the light of principles and ideals acceptable to their common human reason… Only a political conception of justice that all citizens might be reasonably expected to endorse can serve as a basis of public reason and justification.

Rawls, 1993, p. 137, emphasis added

However, securing the consent of all hypothetical agents and securing the consent of all actual individuals are two different things. The latter individuals know their position in the society, hold religious or moral views, belong to socioeconomic categories, and so forth. They might be tempted to adopt self-serving principles. Moral pluralism is also a serious obstacle for consensus-based legitimacy. It strips consensus of part of its appeal.

The issue is not limited to pursuing an ideal in a constrained situation. The existence (or reasonable assumption) of pervasive pluralism undermines the value and traction exerted by consensus. Rawls’s answer is to propose to ground legitimacy on “overlapping consensus” (Rawls, 1993, pp. 133-172), which is a light version of full-blooded consensus. Despite the “fact of pluralism,” Rawls considers that consensus is still possible among individuals holding “not unreasonable comprehensive doctrines” (Rawls, 1993, p. 140). The consensus is limited to the values of public reason, while “it is left to citizens individually…to settle how they think the values of the political domain are related to other values in their comprehensive doctrine” (Rawls, 1993, p. 140).[9]

In short, consensus plays a central role in liberal legitimacy for some authors (e.g., for Rawlsians), at least when it concerns the basic structure of the society and public reason. In other words, it is the first best: the legitimacy of political decisions or institutions in a liberal regime depends on the actual or hypothetical consent of all individuals. When consensus is not attainable under reasonable assumptions, the solution is to approximate it, by retaining some first-best conditions: the consent of all is kept, but, for instance with Rawls, restricted to a limited set of principles or values.[10] This restriction stems from a deflationist strategy that distinguishes metaphysical and political principles, insulates the latter from moral controversies, and searches for a consensus on these principles (and not on a full-blooded conception of the society).

Consensus is more than a theoretical concern for liberals. It also pervades concrete decision-making procedures. The United Nations General Assembly uses consensus and unanimity rules. The Council of the European Union used it until the Lisbon Treaty of 2007. Representative democracy is another example. Given the impossibility of consensus-based political decision making, liberal states rely on the representation principle, the representatives being chosen by voting. Representative liberal democracy approximates consensus in two manners: the consent of all is replaced by the consent of elected representatives and the consent of all elected representatives is usually replaced by the consent of the majority. The value of the majority rule may be interpreted as deriving from its approximation of consensus. Finally, qualified majority is a closer approximation of consensus than simple majority.

In sum, consensus is not only the theoretical first best for part of the liberal tradition of political legitimacy. Its value is not restricted to the choice of principles that regulate the basic structure of the society. It also inspires concrete practices and decisions ranging from representation to voting.

2. CONSENSUS, FROM MORALITY TO EPISTEMOLOGY

According to Bellamy (1999), consensus is at the core of liberalism and large segments of political theory (e.g., Rawls, Walzer, and Hayek). This pre-eminence explains why compromises are sometimes regarded by theorists as being politically, if not morally, inferior. Bellamy stresses that the liberal quest for a minimal consensus works best “when principles are largely agreed or else kept off the agenda” (Bellamy, 1999, p. 98). Problems emerge, however, when political values clash. In a liberal society, consensus is reachable only because, by definition, disputable matters have been put off the agenda (e.g., Rawls posits that individuals are capable of distinguishing political from moral principles, and of finding arguments in the latter for supporting the former). Ultimately, Bellamy challenges the assumption that the “underlying consensus is a liberal one” (Bellamy, 1999, p. 99).

While Bellamy underlines the shortcomings of consensus (as the first best) in cases of moral disagreements in a liberal society, my aim is to extend the discussion to factual disagreements (Kappel, 2017). First there is one objection to consider. One may argue that Rawls, and other liberals, place factual and epistemological issues outside the scope of their consensus-based approach. If right, that would entail that discussing consensus for factual issues would be irrelevant.

The objection calls for two remarks. Firstly, Rawls himself is explicit on consensus playing a role outside moral/political disagreements, at least for the foundations of the political community. He indicates that justifications “on matters of constitutional essentials and basic structure” should “appeal only to presently accepted general beliefs and forms of reasoning found in common sense, and the methods and conclusions of science when these are not controversial” (Rawls, 1993, p. 224). Thus, consensus applies to factual and epistemological components as a central part of the justification of liberal institutions and policies (their legitimacy) or, at least, consensus is the condition for allowing factual components to enter the justifications for political decisions and institutions (Jønch-Clausen and Kappel, 2015).

Furthermore, many liberal theorists consider factual and epistemological material as more hospitable to consensus than axiological material since the validity of epistemological rules and facts may be assessed by scientific enquiry. Thus, factual and epistemological disagreements do not benefit from the same attention from liberal theory. If moral controversies on issues such as the purpose of life or religious education might appear so unsolvable and prone to endless disagreements, even among reasonable agents, factual disagreements have been assumed to belong to the “solvable” category, viz. given proper enquiry, they could (at least potentially) be answered by scientific observation. In other words, if political liberalism conceives moral disagreements as being permanent (and, furthermore, being the unescapable consequence of the freedom of moral agents), in contrast, epistemological and factual disagreements are seen as transient.

Thus, facts and epistemological material are part of the decision-making process and consensus a priori applies to this material—that is, to the part that has not been falsified yet. They are filtered out in ways similar to moral material: by skimming off the controversial (i.e., scientifically challenged) part. This process matches Rawls’s “method of avoidance” (Rawls, 1971, p. 231), but applies to facts, reinforcing Bellamy’s point that liberal legitimacy would hold only because the most contentious (moral or epistemological) issues are pruned off during public decision making.

Secondly, if we move away from the legitimacy of the basic structure and other “essentials” to the actual legitimacy of public policies and institutions, legitimacy often depends on a mix of moral, political, and factual assumptions. Moreover, the various types of assumptions may be difficult to disentangle from one another or to divide into controversial and noncontroversial elements. Alternatively, reasons for policies may rely on justificatory blocs where moral and factual elements are tied together.

For instance, public-health policies rely on moral judgments (e.g., it is a good thing to be healthy) and factual claims about the effects of lifestyles (e.g., smoking causes cancer), some being controversial [the net effect of physical activities on health in urban environment (e.g., biking, running in car fumes) or on obesity (Malhotra et al., 2015)]. The legitimacy of public-health policies is conditional on factual assumptions that are probable in a statistical sense. Decision makers can resent or object to this lack of certainty, which could fuel discontent. Furthermore, assumptions’ being true and factual does not mean that they are uncontroversial.[11]

Consensus plays a prominent role for political legitimacy on moral and epistemological grounds. The liberal society relies more or less explicitly on facts and epistemological elements that are assumed to be uncontroversial—for example, assumptions about human nature (instrumental rationality, autonomy), epistemology (the existence of truth, the hypothetic-deductive method as a firm basis for science), historic events (presence on a delimited territory, unity of a nation), and so forth. Liberal policies assume causation in the domains of health (e.g., between lifestyles and pathologies), the economy (among macroeconomic aggregates), politics, and so forth. Once that point is acknowledged, the next step is to determine how factual and epistemological disagreements may challenge consensus-based conceptions of legitimacy. Moreover, how do they affect decision rules that approximate consensus (e.g., majority rule, representation)?

3. EPISTEMOLOGICAL AND FACTUAL CHALLENGES TO CONSENSUS

In some circumstances, factual and epistemological disagreements may jeopardize consensus-based conceptions of political legitimacy. Numerous decisions or institutions in a liberal democracy need to be grounded on facts or epistemological rules, which can be controversial due to lack of conclusive evidence (e.g., health effects of nanoparticles, macroeconomic relations between inflation and unemployment). In such cases, there is no possibility of setting aside controversial facts or epistemological procedures to reach a minima consensus: the decision process cannot abstain from controversial facts and procedures. In other words, the first best cannot be reached by political institutions.

In addition, the possibility of producing scientific evidence may not settle controversies among citizens. Epistemological or factual disagreements may not be suppressed, even in the presence of scientific evidence (e.g., disagreements about genetically modified organisms, GMOs hereafter), because citizens who are part of the public debates or decision-making process are too committed to their epistemic beliefs. As a result, decision makers and citizens could remain deeply divided on policies without any hope of consensus, which again presents a challenge for consensus as a liberal first best.

Situations where factual or epistemological disagreements impair the legitimacy of political decisions and institutions by fragmenting the population include debates on the recognition of past wrongs (e.g., the Armenian genocide, the Soviet occupation of the Baltic States, the positive role of French colonization or its qualification as a crime against humanity). A contemporary example is Indigenous land claims.

In Canada, proving an ante-European, and at least since colonization, continuous presence on a territory is a prerequisite for the First Nations in order to obtain the recognition of an “aboriginal title” or “ancestral rights” as protected by the 1982 Constitution Act (section 35).[12] Some claims can be settled by historians and archaeologists producing evidence before governments and courts whereas, for others, evidence is controversial, subject to conflicting interpretations. Then nothing is to be expected from scientific enquiry, at least not in the current state of the art of science. A third category covers cases where available evidence will not lessen factual disagreements. For instance, historical presence on a given territory could be so central to Indigenous or non-Indigenous identity that evidence cannot win the agreement of all or contribute to attenuating controversies. A last category includes cases of true epistemological disagreements (i.e., disagreements on epistemological rules and not only on the facts). The society at large (Indigenous and non-Indigenous citizens) may be deeply divided on public policies because of controversial epistemological aspects.

The Canadian jurisprudence illustrates the depth that factual and epistemological disagreements can reach in liberal democracies. For asserting their pre-European and continuous presence on the claimed territory, Indigenous peoples often use oral histories—for example, stories, legends, or oral accounts transmitted from one generation to another (Etinson, 2008). Courts have traditionally refused to consider such accounts as proofs equivalent to written documents or archaeological artefacts. Under the general regime of proof, oral accounts, when not stemming from direct witnesses, amount to no more than hearsays. However, the jurisprudence has been evolving. Some courts have placed oral accounts seemingly on par with other kinds of evidence (e.g., written historical documents) to prove Indigenous presence on a given territory.[13]

Why consider individuals’ epistemological commitments in examining claims such as Indigenous peoples’ ancestral rights? In situations characterized by epistemological uncertainty, the absence of indisputable evidence may advocate for giving more weight to such commitments. However, the argumentative force is not in the commitment, but in the absence of indisputable evidence. What about situations where evidence exists or could be produced? Why consider controversial factual or epistemological beliefs?

Three reasons may support considering factually ungrounded or controversial epistemological beliefs and, thus, diverging from the first best. The first is that redressing past injustices and inequalities justifies modifying epistemological standards. However, such exemption is constrained: it cannot be used in matters unrelated to the past wrongs that justified modifying standards in the first place.

The second reason is that respect is due to the decision makers as participants in public reason or debates. The justification is that “relaxing” standards might be necessary for including decision makers in the democratic process. However, such a justification is too broad: standards cannot be “relaxed” for all decision makers in any situation, except at the price of a collapse of rational deliberation. Such a justification needs to be circumscribed by identifying a conception of respect that could justify “relaxing” standards only for some decision makers (e.g., for Indigenous peoples and not for other citizens who inhabit the disputed territory).

In liberal theory, respect is usually grounded on the (posited) possession of given faculties (viz. reason, rationality, or autonomy) by all individuals. However, “relaxing” epistemological standards only for some decision makers does not do justice to decision makers in terms of reason, rationality, or autonomy. A possibility for qualifying the second reason is to elaborate on the respect due to specific decision makers. Liberal theory already justifies differential treatment. Take the example of impaired capacity. Because of their limited abilities could justify a differential treatment. Because of their impaired capacity to distinguish true from false statements or to follow sound epistemological rules, respect imposes to relax epistemological standards for some decision makers. Hence all decision makers will be included in the public exchange of reasons. It is then possible to present second-best issues in relation to consensus as expressing a tension between an epistemic defence of democracy, according to which democracy and truth enquiry are mutually supportive, and democracy as equal standing and inclusion (MacGilvray, 2014, p.117). Simply put, democracy as a process to hold true beliefs might threaten democracy as equality in the public exchange of reasons.

While justified for individuals with limited abilities (e.g., children, people with dementia), such reason is abhorrent when used to justify “lowering” epistemological standards for Indigenous peoples or any mobilizing group acting upon different standards. The differential treatment is rooted in a depreciative view of the individuals who hold different epistemological views. Such a ground for lowering standards adds insult to the injury of injustice. Furthermore, if respect implies treating people equally and treating them as equals (Dworkin, 1985, p. 190), this is also disrespectful of decision makers to whom original standards continue applying.

A third reason is to allow controversial facts or epistemological considerations to enter the decision-making process for pragmatic reasons, such as political stability. Because the exclusion of such considerations may nurture political instability, it may be better, everything considered, to give weight to them during decision making. For instance, proofs may exist that a genocide took place one century ago. Part of the population may ask the government to release an official recognition of the tragedy or to compensate for the victims. Another part of the population may disagree so fiercely on the reality of the tragedy or on its consequences that the government may finally decide to give way to their opposition by not enacting any kind of recognition or compensation.

While the first argument makes sense when dealing with justice issues such as Canadian First Nations and the second argument is controversial for the reasons I already mentioned, the third one is interesting due to the general framework it offers for thinking about second-best departures from the first best for reasons of stability, which occupies a central place in the liberal thought. Many developments of Rawls’s Political Liberalism are devoted to it and to the means for securing individuals’ support to democratic institutions (Barry, 1995; Klosko, 1994). Stability is so central for Rawls that it is one pillar of political liberalism, alongside justice (Klosko, 1994, p. 1183). (I will come back to stability in the next section.)

In any case, this does not imply that the three reasons are convincing. They illustrate that reasonable grounds exist for considering epistemologically controversial arguments during public decision making, even where evidence could be available. Then, the following question emerges: in situations where only controversial facts or epistemological elements are available or where decision makers are too committed to their epistemological beliefs, why not call for traditional decisional procedures that approximate consensus (e.g., voting)? In other words, why do such factual and epistemological disagreements undermine consensus as the first best?

4. SECOND BEST AND POLITICAL LEGITIMACY

The previous sections offer two arguments: (a) consensus is the first best for part of liberal legitimacy, and (b) factual and epistemological disagreements may challenge the consensus-based approach, especially for pragmatic reasons like political stability. There are two manners in which to understand this challenge: one radical and another moderate. The former is that disagreements necessarily incur a second-best issue and, consequently, a general departure from first-best conditions (as suggested by Lipsey and Lancaster). The latter is that disagreements under specific conditions impose a general departure from first-best conditions (i.e., from consensus approximations). In other words, only some disagreements would constitute or raise second-best issues. Only the latter is plausible, though, since it is difficult to see how all instances of factual and epistemological disagreements could raise second-best issues.

Before proceeding, I need to establish a preliminary condition. As seen, a second-best problem emerges when the first best is not attainable due to factual or epistemological disagreements. There are cases where a full consensus on facts or epistemological rules is not attainable because of uncertainty[14] [e.g., disagreements over risk evaluation, such as for the “Black Swans” (Taleb, 2008)] or decision makers’ epistemological commitments. In addition, the disputed facts or epistemological rules should be necessary for decision making—that is, for justifying a political decision, policy, or regime.

For instance, when regulating chemical pollution, public institutions cannot avoid using contamination thresholds. They need measurement and impact studies to determine which chemical to regulate and how. These elements are necessary for policy making and often cannot be set aside, even in cases of deep disagreement. If the disputed material is not necessary (e.g., if the disagreement is about the kind of cancer a chemical favours, while there is no disagreement about the chemical creating a hazard), it could be ignored during the decision-making process (e.g., when setting a release threshold for the industry) and a consensus could be hammered out from the undisputed material (the hazardous aspect of the chemical and the necessity to regulate).

The preliminary condition for a second-best problem is that the first best cannot be reached due to factual or epistemological disagreements bearing on element(s) that are necessary for decision making. Although necessary, this condition is not sufficient. If the first best—consensus—is not attainable, a second-best problem does not necessarily arise. Approximating the first best could still be possible or desirable. In general, a second-best issue arises only if the remaining first-best conditions are impossible or undesirable (Räikkä, 2000).

Concerning the first best for liberal legitimacy qua consensus, the preliminary condition requires that the consensus not be able to be reached and that approximating the consent of all is either impossible or undesirable. Only then does a second-best issue arise. Theoretically, it means that second-best issues characterize situations where any mild or altered form of consensus, such as the overlapping consensus, is either impossible or undesirable. Practically, it implies that aiming at getting a majority through voting and representation may be neither possible nor desirable.

The preliminary condition raises a question: under which conditions is pursuing the remaining first-best conditions either impossible or undesirable, and, in particular, what kind of disagreement could bar approximating the consent of all regarding the legitimacy of institutions, policies, or public actions? As previously evoked, the main risk posed to legitimacy by factual and epistemological disagreements occurs when they deeply fragment the political community or accentuates prior profound division. For a second-best problem to arise, citizens would have to be so divided that trying to get the consent of as many citizens as possible on the contested facts or epistemological rules would undermine the legitimacy of political decisions and institutions. This condition has two parts: citizens would have to be divided on the factual and epistemological issues at stake and the divide would have to cut so deep it threatens legitimacy when institutions try approximating consensus.[15]

The kind of social division threatening a liberal regime is one that endangers stability. In addition, stability represents a key liberal concern. Rawls proposes an overlapping consensus because it increases the probability of individuals adhering to the principles of a liberal society and, therefore, enhances political stability.[16] Then, various situations may present a threat for liberal legitimacy.

An important category embraces situations where the combination of the pursuit of the first best and factual/epistemological disagreements undermines mutual or institutional trust (i.e., trust among citizens or citizens’ trust towards their institutions) and where it is very likely that consensus approximation will undermine trust. The second-best issue could worsen if, due to damaged trust, citizen participation in liberal institutions declines to the point that the perpetuation of such institutions might be endangered.

An illustration is GMOs. European citizens do not trust national and European institutions on genetically modified food (van Kleef et al., 2006, p. 58). In these cases, vote or representation may generate or accentuate distrust, whereas metastudies have underlined the absence of proved harms to human health by GMOs (Barrows et al., 2014; Nicolai et al., 2014), while no comparable metastudy has proved the contrary.[17] Such situations are risky for liberal democracies in several respects.

First, the risk is that consensus approximation (mostly voting) may worsen tensions between the majority, opposed to GMOs (Gaskell et al., 2010), and the minority, including experts who do not evaluate GMOs as more harmful than their non-GM equivalents. Another risk is the possibility that consensus approximations (viz. voting) support suboptimal public policies in a Paretian sense—that is, policies that would make no one better off without degrading the situation of someone else when at least one alternative exists where an agent could be better off with no one else being worse off.

In relation to suboptimal outcomes, a stronger claim is that opposing GMOs actually harms people (Potrykus, 2010). The future collapse of traditional crop yields, overpopulation, deforestation, and climatically driven changes (e.g., water shortages, irregular rainfalls, higher drought frequency, increase of extreme climatic events, salt saturation in coastal lands, mineral depletion in soils) carry significant risks that might be (partially) addressed by broad cultivation of GM crops (under certain conditions).[18] Therefore, the rejection of GMOs based on consensus approximations might turn out to be detrimental to most of humanity within a few decades.[19]

If we accept scientific evidence that GMOs are not more harmful to human health than their non-GM equivalents, and if we accept GMOs’ gains in yield and in pest resistance (e.g., Klümper and Qaim, 2014), public opposition to GMOs creates social gridlocks. These gridlocks happen when the conditions of agents (e.g., consumers, farmers) could be improved without worsening the condition of anyone else (or, at least, without worsening the condition of another agent in proportions larger than would have been the case with non-GMOs).

A third risk is the rejection of scientific evidence as a legitimate ground for public policy (when scientific evidence and popular sentiments conflict and consensus is approximated), which may lead to grounding public policies on unproven or false premises, a risk comparable to the one posed by populism. A further challenge is the danger of increasing the intensity of factual/epistemological disagreements in the future, because of science’s loss of authority (due to its declining perceived legitimacy within a population).

Hence, the question is the following: Which resources could the TSB offer to public decision making embroiled in factual and epistemological disagreements? This question captures the challenge of mobilizing the second best: to offer specific enough theoretical and practical resources. The issue is that the literature on the TSB in political theory is underdeveloped. Moreover, this literature is crippled by misunderstandings and approximations (Räikkä, 2000). Most of political theorists who appealed to the TSB did so in a very brief manner (in no more than a couple of paragraphs or sometimes a few sentences).[20] Therefore, there is still a long way for the TSB to help deal with the political issues of legitimacy, education, institutions, justice, and so forth.

To make the TSB relevant for liberal legitimacy (and political theory), two issues ought to be addressed head on. The liberal theorists who want to mobilize the TSB need to specify how to understand the general departure form first-best conditions in a liberal sense and the constraints stability imposes on second bests. However, this task should be carried out while keeping in mind Lipsey’s reservations and warnings. First, finding a second best seems more difficult than achieving the first best (Lipsey, 2007, p. 356). Second, finding a second best is about “piecemeal improvements in welfare” (i.e., about adopting a pragmatic approach).

4.1. Liberal departure

According to Lipsey and Lancaster’s theorem, a second best implies a “general departure” from first-best conditions (i.e., a change of all conditions). Therefore, for the TSB to be meaningful for liberal theory, it is important to constrain the principles or rules used for identifying a second best. In the context of the present discussion, these principles should be compatible with liberalism, but alien to a consensus-based approach. The principles supporting a second best should constitute a general departure from first-best conditions (viz. no voting or other approximation procedure) and be liberal. They need to be liberal because liberalism is, by definition, the encompassing justificatory architecture for the conception of political legitimacy discussed in this article.

Therefore, the issue of what could constitute liberal second bests for political legitimacy (among other issues) emerges. Without discussing the matter in detail, let me consider just a few possibilities to give the reader a rough idea of the kind of conceptual work to be undertaken by anyone who wants to apply the TSB to political theory.

One possibility is to adopt a rights-based approach. When factual and epistemological disagreements jeopardize the realization of the first best, political decisions and policies could be compared and adopted based on their impact on individual rights. This presupposes mobilizing a background theory of the nature of rights and rules of adjudication between competing claims, which excludes many group-rights approaches (which are anti-individualistic, limit individual autonomy, etc.). Another option is to adopt a welfarist approach. The legitimacy of decisions and policies may be assessed by assessing their impact on welfare. There again, further discussions are necessary on the nature of welfare, the possibility of making welfare comparisons across individuals, groups, situations, and so forth. As for the rights-based approach, the justificatory scheme ought to be compatible with liberalism.

These options do not come without conceptual and practical difficulties. Moreover, they do not cover all possibilities. Nonetheless the liberal-departure issue shows that one might still endorse consensus as a first best, but, due to specific constraints, adopt a backup justificatory scheme. Thus, justifications might come into play at different levels (which is a different point from the distinction between ideal and non-ideal theories). In other words, the TSB may advocate for a pragmatic decompartmentalization between competing conceptions of political legitimacy. It argues for tuning down some differences among theories of political legitimacy and adopts some sort of justificatory pluralism based on pragmatic considerations.

4.2. Stability

This liberal concern, present in Rawls, requires that the application of second-best principles not undermine political stability (e.g., by fuelling distrust toward democratic institutions). This condition could be made stronger by requiring that the application of these principles strengthen stability. In other words, on top of being compatible with liberalism, the principles that provide guidance for identifying a second best ought to also guarantee social stability. This condition is practical and, as such, open to debate. However, the important point to retain is that, again, the TSB seems to point to highly contextual reflections. In that sense, further investigations of the TSB will show a strong potential overlap with methodological discussions on contextual approaches in political theory (e.g., Carens, 2004; Kukathas, 2004).

5. CONCLUSION

Looking back at the TSB fifty years on, Lipsey (2007, p. 356) indicates that second-best issues call for pragmatism, especially because no global second-best solutions can be tailored for economics (while the case is still open for politics). Therefore, only piecemeal changes are left to policy makers. Coupled with the underqualification of the TSB, this notice of caution could fuel the view that the TSB’s fecundity for political theory remains limited. Moreover, one could nurture the view that whether factual and epistemological disagreements undermine political legitimacy is a purely contingent matter. Then, the conclusion could be that the TSB offers no compelling or workable challenge to consensus-based conceptions of legitimacy.

Certainly, the existence of factual and epistemological disagreements does not necessarily undermine political legitimacy, and, as indicated, the existence of such disagreements does not necessarily create a second-best problem—that is, a general departure from first-best conditions (consensus). In this article, I recognize these points and discuss a more modest thesis: factual and epistemological disagreements could create second-best problems for the liberal tradition of political legitimacy based on consensus. I tried to clarify two points: (1) what consensus as the first best means for part of the liberal tradition as concerns moral and factual/epistemological disagreements and (2) how such disagreements could create second-best problems for consensus-based approaches. I ended up by underlining, on the one hand, the risks associated with first-best approximations in situations shaped by a deeply divided society and, on the other hand, the necessity to evaluate more closely the resources the TSB could offer to political theory.

That being said, it is true to some extent that the claim that factual/epistemological disagreements create second-best issues for consensus-based legitimacy depends on contingent factors (e.g., the degree of trust, the intensity of disagreements, the nature of disagreeing parties, etc.). Nevertheless, the challenge for liberal legitimacy founded on consensus is not contingent (in the sense of anecdotal). For any political theory that relies on the proximity to consensus for assuring the legitimacy of its foundational principles and institutions, any reasonable expectation of dissensus is a direct challenge to that theory’s plausibility and expected outcomes, like stability, (and therefore its traction) (Thrasher and Vallier, 2013). The existence of enduring factual and epistemological dissensus forces consensus-based approaches to legitimate the rules and outcomes of decision making with other principles, which should be compatible with liberalism. This dimension is largely (if not totally) absent from political theory.

However, for the political theorists interested in the task, a suspicion looms over the whole project. To the question “Are there general policy rules for piecemeal improvements?,” Lipsey (2007, p. 358) answers with a clear “no,” reducing the entirety of the TSB to a piecemeal approach. Thus, in the absence of serious discussions of the TSB by political theorists, it seems that appeals to the second best are doomed to either remain rhetorical (as false equivalents for non-ideal approaches, for instance) or advocate for political casuistry.