Talk May Be Cheap, but Deeds Seldom Cheat: On Political Liberalism and the Assurance Problem

: In a well-ordered society, democratic officials face an assurance problem. They want to ensure that others will act reasonably when they do the same. According to political liberals, public reason can solve this problem, but the details of how assurance is generated are unclear. This article explains the assurance mechanism in political liberalism. Apart from public reason, mutual assurance is also provided by a long-term record of civic deeds. By performing civic deeds over time, officials signal their reasonableness to each other. This record of civic deeds is costly to unreasonable officials and thus represents a reliable way to differentiate trustworthy fellows from others. The article also shows that a recent critique of political liberalism, which argues that public reason is merely cheap talk and thus political liberalism fails to provide mutual assurance, misses the point. It overlooks that assurance is created through talks and deeds together .

I n recent years, partisan polarization has intensified in contemporary democracies. More and more, politicians use radical methods to achieve their political aims, even if these methods increase partisan hostility. As the retired American politician Paul Ryan bemoaned, "broken politics" has become "the biggest challenge of our time" (The Guardian 2018; cf. Levitsky and Ziblatt 2018). This climate of partisanship can be explained by the erosion of trust between opposing parties. When politicians believe that their opponents are untrustworthy and that their offers to compromise and exercise self-restraint may merely be exploited, they may conclude that there is no point to compromise and may rather try to win by any means necessary. In the face of such partisanship, people begin to lose faith in democracy. Is democracy, even at its best, merely a civil war by other means, a gladiatorial encounter without bloodshed? We do not believe this. Despite serious disagreements, people do not always treat political opponents as their enemies. The problem at stake is that, before trusting others, people want to ensure that others are trustworthy and will not take advantage of them. Trust, in turn, is conditional on the existence of mutual assurance. We call this the assurance problem.
It may be difficult for political theorists to explain how mutual assurance can be achieved in today's democracies, but exploring how mutual assurance works in an ideal democratic society can strengthen our confidence in achieving it in the real world. 1 John Rawls famously describes an ideal democratic society as a "well-ordered society," that is, a society in which members recognize an overlapping consensus on "a family of political conceptions of justice" (Rawls 2005, 450). In a well-ordered society, reasonable citizens are willing to fulfill the duty of civility. For example, they restrict themselves from making political decisions and performing political actions solely out of self-interest, and they act on the ground of shared political values, in accordance with the idea of public reason. However, reasonable citizens are more willing to exercise such restraint when others likewise act reasonably. Mutual assurance is needed. Reasonable citizens may be concerned that others will take advantage of their reasonable behaviors, and refrain from doing their part unless they have confirmed that others are similarly committed (Rawls 2005, 49). Hence, the assurance problem is not only an unfortunate situation that happens in real-world politics. It also occurs in a well-ordered society. It follows that, if assurance cannot be established even in a well-ordered society, people may be justified in distrusting democracy itself. Solving the assurance problem in the idealized case is necessary, though not sufficient, for solving it in the real world.
This article, therefore, draws on the ideas of political liberalism, especially the Rawlsian version, to offer a solution to the assurance problem in a well-ordered society. According to Rawls and some Rawlsians, by offering and responding with public reason, citizens, particularly public officials, manifest that they fulfill the duty of civility and are thus reasonable. Mutual trust is generated, and democracy thereby achieves stability. Public reason acts as a signal whereby public officials can identify each other as trustworthy, reasonable citizens. This account has been challenged recently by many political theorists (Chung 2019, 13-14;2020, 88-90;Gaus 2011, 317;Kogelmann 2019, 200;2022, 180-81;Kogelmann and Stich 2016, 720-21;Thrasher and Vallier 2015, 941-45;Vallier 2014b, 654;2019a, 190). These theorists argue that the assurance provided by public reason is fragile because public reason is merely cheap talk. Our article aims to show that these theorists misunderstand the assurance mechanism in political liberalism. After explaining the assurance problem and the idea of public reason in the first section, we argue that it is mistaken to hold that officials assure each other by only speaking in terms of public reason. In the second and third sections, we maintain that mutual assurance is also provided by civic deeds required by the duty of civility. Finally, we contend that civic deeds are reliable signals because they are costly, and it is therefore difficult for unreasonable public officials to imitate.

The Assurance Problem and Public Reason
Suppose there are two hunters. Both hunters can either hunt stag or hunt hare. Both prefer catching a stag, but no hunter can catch it alone. Catching a stag requires that the two hunters hunt together. Another option is to hunt hare, which can be caught by a single hunter, but this is less preferred. Hence, the most preferred outcome for each hunter is that both hunt stag (Stag, Stag), which is the Pareto-optimal Nash equilibrium. For each hunter, the worst outcome is to hunt stag while the other hunter hunts hare. In this case, the stag hunter gets nothing. The game can be diagrammed as follows: In this game, a hunter will choose to hunt stag only if he knows that the other hunter will do the same. This idea of the stag hunt first appeared in Rousseau's A Discourse on Inequality (1987, 62), though not in a game-theoretical form. Rousseau's account illustrates a problem of coordination that people often face. People are willing to perform their duties only as long as others do the same, so that they are not exploited. Besides Rousseau, Hobbes famously claims that one must refrain from seizing another's goods only on the condition that another will not seize one's own goods (1994,80). Locke states that one should live peacefully with another person unless that other person first attacks, or overtly threatens to attack, oneself (1988,280). Mutual assurance has long been a central problem in the social contract tradition.
Rawls, as a social contract theorist, also takes the assurance problem seriously. In his account of political liberalism, the assurance problem arises when reasonable citizens in a well-ordered society are not certain about whether others are reasonable. "Reasonable citizens" refers to those citizens who accept the "burdens of judgment," that is, the fact that reasonable citizens disagree on issues of morality, religion, and other matters concerning human flourishing. Despite their disagreements, reasonable citizens respect other reasonable citizens by refusing to use state power to impose their "comprehensive doctrines" on others (Rawls 2005, 61). Also, reasonable citizens are willing to comply with the unfavorable laws and policies made by other reasonable citizens, for they know that others are willing to make similar sacrifices when required to do so: This is what Rawls called the "criterion of reciprocity" (2005, xliv). Unreasonable citizens, however, ignore these selfrestraints. They take political actions solely from within the perspectives of their comprehensive doctrines. They may pursue their sectarian goals in radical ways and use state power to impose their comprehensive doctrines on others. When reasonable and unreasonable citizens are in political competition, reasonable citizens who restrain themselves from using radical tactics may be disadvantaged, and they may eventually be forced to adopt these tactics. To avoid this, reasonable citizens prefer acting reasonably when they trust that their fellow citizens will do the same (Rawls 1999, 296). Without this assurance, citizens may worry about being exploited and may be tempted to act unreasonably. When a significantly large number of citizens are uncertain whether their fellows are reasonable, mutual suspicion "may eventually cause the scheme to break down" (Rawls 1999, 211) and thus destabilize the well-ordered society. 2 The problem of assurance is sometimes criticized as ambiguous (Lister 2017, 163). Here, we distinguish between two assurance problems. The first is the compliance assurance problem, which is the worry that some citizens may break the law and take advantage of others. The second is the authorship assurance problem, which is the worry that citizens may violate the duty of civility when they legislate new laws or interpret existing laws. The compliance assurance problem can be solved by a penal system: Given an effective system of penalties, rule breakers will be caught, and citizens can thus be assured that their fellows will obey the law (Rawls 1999, 211). The authorship assurance problem, which usually happens between high-ranking public officials, is more complicated. The duty of civility requires officials to exercise political power in a way that could be justified by a political conception of justice. For a political conception of justice, it means a conception that is elaborated from fundamental intuitive ideas implicit within the common political culture of constitutional democracy, such as "those of political society as a fair system of social cooperation, of citizens as reasonable and rational, and free and equal" (Rawls 2005, 376). Based on these ideas, a 2 Due to space constraints, we assume that the unreasonable population size is above a certain threshold and reasonable citizens have to be worried about deception (Chung 2020, 86-88). Also, although reasonable citizens are not willing to be ruled by unreasonable citizens, they still respect the basic liberties of unreasonable citizens, engage them in public deliberation, and try to persuade them to change their minds (Quong 2011, 291; Wong forthcoming). political conception is freestanding from comprehensive religious and philosophical doctrines (Rawls 2005, 12). Its freestandingness enables a political conception to become a mutually acceptable ground for legislation in a pluralistic society. Reasonable officials should exercise political power according to a political conception, instead of using political power to promote sectarian religious or ethical goals. However, this duty of exercising political power cannot be enforced by law. For instance, legislators are free to debate, negotiate, and bargain in the public domain, so long as they are subject to the control of the ballot box. Also, sometimes the legal and political questions are complex and involve significant areas of uncertainty. It may be unclear whether a legislator acts in accordance with a political conception of justice or, instead, relies solely on a comprehensive doctrine. When controversial issues are at stake, a reasonable official could worry that unreasonable officials may pretend that they are reasonable, disguise their real intentions, and use their power to pass sectarian laws.
Accordingly, the structure of the authorship assurance problem is different from that of the standard stag hunt game. In the stag hunt game, the payoff of the hare hunter is indifferent between the scenario in which both players hunt hare (Hare, Hare) and the scenario in which the other hunter hunts stag (Stag, Hare). A hare hunter thus has no incentive to deceive another hunter that he is going to hunt stag. In the authorship assurance problem, however, an official may gain advantages in political competitions when she acts unreasonably but other officials act reasonably, compared to a situation in which all officials defect and act unreasonably. Hence, a reasonable official may be worried that other officials intend to deceive her and pretend to be reasonable. The authorship assurance game can be represented as follows: The authorship assurance problem can be explained by the following example. Suppose that Audrey and Bruno are leaders of two parties in Congress and have different views of distributive justice. Audrey is a liberal socialist and Bruno is a property-owning democrat. Although both endorse reasonable comprehensive doctrines, they sometimes disagree with each other's political proposals. As a socialist, Audrey believes that major resources, such as land and oil, should be publicly owned (Edmundson 2017, 33). Nowadays, social media platforms, such as Facebook, control vast quantities of personal data, which is very valuable to politicians and business executives. This kind of data is even called the "new oil" of the digital world (World Economic Forum 2011). Platform owners greatly benefit from selling the data and become politically influential. Audrey, therefore, proposes a bill to nationalize Facebook. 3 If oil should not be left in private hands, then neither should the "new oil." Suppose that Audrey's party has the majority in Congress, and her bill is very likely to pass. Bruno now faces a difficult question: Should he acknowledge the forthcoming nationalization as legitimate? To Bruno, nationalizing Facebook is controversial and radical. Bruno may suspect that Audrey and her party are turning the state into a sectarian socialist polity. If this happens, Bruno thinks, he has to resist this nationalization by civil disobedience, enabling more officials to realize its injustice. To counter Audrey's radical reform, Bruno and his party also consider proposing retaliatory proposals that could only be justified by their comprehensive doctrine and mobilizing their supporters through extremist political discourses in the next election. These actions will obviously foster partisan hostility, but Bruno thinks they are necessary if Audrey is dragging the country in a terribly wrong direction. Hence, it is important for Bruno to discern whether Audrey is reasonable or not, for this determines whether he should accept the unfavorable result of the legislation in question or resist it by radical political actions.
In this case, the most preferred outcome is that Bruno confirms that Audrey is reasonable. He thereby chooses to compromise and act reasonably. Political opponents continue to be civil and moderate. This is the outcome in which both officials act reasonably. However, if Bruno believes that Audrey is unreasonable, then he is unwilling to compromise. He will turn to promoting his sectarian goals in radical ways. By doing so, Bruno can mobilize his supporters and resist the hegemony of Audrey. Eventually, both sides will suffer from the mutual hostility caused by rising partisanship (the outcome in which both officials act unreasonably). Nonetheless, this unpleasant outcome is still better for Bruno than acting reasonably, losing totally in the political competition, and being subjugated to the socialist hegemony of Audrey (the outcome in which one official acts reasonably and the other acts unreasonably). In brief, Bruno would be more willing to act reasonably when he has confidence that Audrey is reasonable, and he would be more tempted to act unreasonably when he determines that Audrey is unreasonable. Without a method to identify Audrey's reasonableness, Bruno, as well as many other reasonable officials, may misunderstand Audrey and turn to acting unreasonably. This may intensify partisanship and destabilize the well-ordered society.
Some political liberals argue that speaking in terms of public reason is the solution. In a democratic society, although officials endorse different comprehensive doctrines with incompatible epistemological, metaphysical, and ethical beliefs, they share a family of political conceptions of justice. Since these conceptions are based on fundamental intuitive ideas shared among reasonable officials, they form public reasons that are mutually acceptable in public discussions. Public reasons also include those accepted general beliefs, the forms of reasoning found in common sense, and the methods and conclusions of science that are not controversial. These reasons serve as a shared ground of justification because they are accessible to each official, regardless of their comprehensive doctrines. It means that these reasons are regarded as sound reasons according to some common evaluative standards shared among officials. Such standards are principles and values used by "citizens…to decide whether substantive principles properly apply and to identify laws and policies that best satisfy them" (Rawls 2005, 224;cf. Quong 2011, 264). Officials scrutinize, dispute, or affirm reasons offered by others by using these standards. 4 Political liberals typically justify the use of public reason by appealing to the moral values that underpin it, such as respect and autonomy (Audi 2000, 30;Boettcher 2007;Larmore 2015, 77-79;Neufeld 2022;Nussbaum 2011), civic friendship (Ebels-Duggan 2010, 55-58; Leland 2019; Leland and van Wietmarschen 2017, 157-64;Lister 2013, 105-10), noncoercion (Nagel 1991, 36;Wong 2020), justice (Quong 2014, 273-74), coauthorship (Bird 2014, 201-4), and civic equality (Kugelberg forthcoming). 5 Some political liberals are notable exceptions and recognize the social benefit of public reason, but their accounts are brief. Hadfield and Macedo argue that "by offering public reasons…we reassure one another of our cooperation on fair terms that we can all share" (Hadfield and Macedo 2012, 10), while Paul Weithman explains in one influential account that public reason is needed because "members of the [well-ordered society] all need assurance that everyone else acknowledges the authority of the unified perspective on fundamental questions that the political conception provides" (Weithman 2010, 10;cf. Weithman 2015). 6 Similarly, Watson and Hartley argue that public reason "strengthens the stability of the overlapping consensus" because it "deepens the strength of mutual assurance between citizens that they will honor the terms of social cooperation" (Watson and Hartley 2018, 51). According to these political liberals, by speaking in terms of a shared language of public reason, officials assure one another that they are reasonable and are thus willing to act reasonably.
We can use the case of Audrey and Bruno to explain this point again. Facing the doubt of Bruno, Audrey can show that she is reasonable by explaining that nationalizing Facebook is a way of preventing important resources, such as personal data, from being concentrated in private hands. By eliminating the political influences of the Facebook owners, citizens are more likely to have a roughly equal chance to affect political outcomes. Hence, this promotes the fair value of the political liberties (Edmundson 2017, 128-38). Although Bruno disagrees with Audrey's policies, he will agree that the fair value of the political liberties is of great value, and he will understand that Audrey still adheres to a political conception of justice based on these shared political values. They merely differ in the way that they rank or interpret these values. Bruno will, therefore, choose to compromise and acknowledge that the nationalization is legitimate.
However, this seems to exaggerate the power of words. Why would Bruno trust Audrey to be reasonable simply because she speaks in terms of a kind of political language? Recently, a number of political theorists have argued that public reason fails to be a costly signal that enables reasonable officials to identify each other (Chung 2019, 13-14;2020, 88-90;Gaus 2011, 317;Kogelmann 2019, 200;2022, 180-81;Kogelmann and Stich 2016, 720-21;Thrasher and Vallier 2015, 941-45;Vallier 2014b, 654;2019a, 190). Their arguments rest on the signaling theory in economics. According to the signaling theory, a costly signal is a way of showing commitment that one is going to act in certain ways. The signal of commitment is credible from the perspectives of others because it is costly; the cost of sending this signal is much higher for those people who fake such a commitment. Hence, only those people who will truly act in the committed ways are likely to send the signal, and other peo-ple can identify them. In other words, the costly signal enables one to differentiate credible people from others (Skyrms 2003, 80). The cost of presenting political views in terms of public reason is, however, too low. By definition, in a Rawlsian well-ordered society, everyone is supposed to be familiar with at least a political conception of justice. In other words, everyone is able to offer public reasons to justify their decisions when they are asked. Also, the meanings of political values are usually vague, and thus public reason can be used to defend many vastly different policies. 7 Thus, "giving reasons from [a] political conception of justice in public discourse will not be costly" (Kogelmann and Stich 2016, 721). Using public reason is virtually free. Public reason, therefore, is merely a kind of "cheap talk" (Aumann 1990).
Hence, public reason fails to create mutual assurance in the authorship assurance game mentioned above. Suppose we return to the case of Audrey and Bruno. If Audrey is a reasonable official, then she would certainly offer public reasons to explain her proposal of nationalization. However, if Audrey is an unreasonable official, then she could also use public reasons to hide her sectarian ambition. Audrey's secret plan could actually be to nationalize Facebook so that her party could use it as a platform to disseminate propaganda for a socialist conception of a collectivist good life. 8 Yet, she could fool Bruno into not resisting the first step of this plan (the nationalization of Facebook) by clothing it with justifications appealing to public reasons. This lip service costs Audrey hardly anything, but she can benefit when she successfully disguises herself as trustworthy. Therefore, no matter whether Audrey is reasonable or not, Audrey would offer public reasons. From Bruno's perspective, public reason fails to serve as a costly signal that differentiates reasonable officials from unreasonable officials. Political liberalism, therefore, is vulnerable to the authorship assurance problem. We call this objection the cheap talk critique.
We believe that political liberalism can resolve the authorship assurance problem. However, the key is not the discourse of public reason, but rather the civic deeds linked to the duty of civility. The following three sections will explain the idea of civic deeds and show how it responds to the authorship assurance problem and the cheap talk critique.

FIGURE 1 Frequencies of Stag Hunt in the Experiment
Notes: SI: Signal with Information, SNI: Signal without Information, NSI: No Signal with Information, NSNI: No Signal, without Information. The experiment (Charness and Grosskopf 2004, 388) shows that players in the stag hunt game have the highest percentage of achieving Pareto-optimal outcomes when they receive both signals that pertain to each other's moves in the present round and the information about each other's true actions (SI). This percentage is higher than the percentage of a situation that players merely receive the signals of each other's moves (SNI).

Generating Mutual Assurance from Deeds
Although the authorship assurance problem is in some sense similar to the stag hunt game, it is helpful to examine how the stag hunt game can achieve mutual assurance. In this game, players achieve the Pareto-optimal outcome when they place trust in others and are willing to take a risk. Two economists, Gary Charness and Brit Grosskopf, found that knowing the actions of the other player plays a critical role in generating trust (Charness and Grosskopf 2004). In their experiment, 144 participants were divided into 12 groups. Participants were asked to play 10 rounds of the stag hunt game, with the pairings reshuffled randomly for each round so that no players were aware of the identity of the other player in the pair. The 12 groups played the game in four different settings. They might or might not have information about each other's actions, and they might or might not receive signals pertaining to their moves in the present round (whether or not they will choose to hunt stag) from the other players in their pairs. Supposing that "B" denotes the Pareto-optimal outcome that both players hunt stag, the frequencies of B in these four settings can be presented as in Figure 1 (Charness and Grosskopf 2004, 388). The worst two settings with respect to generating trust were NSI and NSNI, in which the players received no signals pertaining to others' moves in each round. The best outcome was achieved in the setting SI, in which the players both received signals and had information about their counterparts' actions, so that it was transparent to the receivers whether the senders of the signals cheated. There was a substantially higher percentage of Paretooptimal outcomes in the groups that played in the SI setting than in the groups that played in the SNI setting: 79% (142 out of 180) as compared to 49% (88 out of 180) (Charness and Grosskopf 2004, 385). Hence, information about their fellow players' actions enables players to judge the credibility of a signal. Furthermore, as Figure 1 shows, the more rounds players play, the more knowledge that players have about their fellow players' past actions, and the easier it is for the player to trust the signals leading both players to choose to hunt stag. Suppose that a receiver repeatedly found that senders pay for their words (i.e., the senders sent a signal to hunt stag and then they really went to hunt stag); the receiver is more willing to trust the signal of the senders and choose to hunt stag. Senders, in this setting, also tend to be credible because they know that any cheating would be made known to the receivers (Charness and Grosskopf 2004, 386). In sum, mutual assurance is robust when people can observe others' deeds and thus ascertain trustworthiness with respect to others' signals.
The experiment shows that the role of action should not be overlooked in mutual assurance. In fact, we can describe these actions as a kind of costly signal. When players are willing to take risks and pay for their words, other players will know that there are some trustworthy players and are thus willing to take risks as well. However, there are three differences between this stag hunt experiment and the authorship assurance problem, and the final one is crucial. First, in the experiment, the organizers will take care of relaying the senders' actions to the receivers. In the authorship assurance problem, the officials have to actively show their actions to other officials. Second, in the experiment, players are anonymous to each other. In the authorship assurance problem, officials know the identities of each other. This is helpful because if officials continue to pay for their words, then other officials will increasingly believe that those officials are trustworthy. Third, and most importantly, in the experiment, it is obvious to the players what they should observe to assess others' trustworthiness, namely, hunting stag or hare. If players claim that they will hunt stag and they finally choose to hunt stag, then their action matches their words. However, in the authorship assurance problem, the action itself may be the subject of disagreement. As Rawls acknowledges, even in a well-ordered society, it is unrealistic to expect that all reasonable officials endorse a single conception of justice. Rather, there is a family of political conceptions of justice in the public political culture. 9 The "just" legislative action of some officials may be problematic to other officials. For example, Audrey's action of proposing to nationalize Facebook is the action that leads Bruno to suspect that Audrey is unreasonable. Hence, the question is, given that officials endorse different political conceptions of justice and have different political proposals, do they have a shared standard for judging whether others act reasonably?

Civic Deeds in Political Liberalism
We argue that they do. Apart from endorsing a political conception of justice, a reasonable public official should perform the duty of civility (Rawls 2005, 217). The duty of civility refers to a set of moral requirements that officials should fulfill when they address political disagreements. It contains three parts: (1) officials should present public reasons that are accessible to others to justify their political proposals (Rawls 2005, 226), (2) officials could rely on comprehensive doctrines in political decision making only if they could also provide public reasons to support their positions in due course (Rawls 2005, 219), and (3) officials should always maintain an appropriately deliberative attitude in political discussions (Rawls 2005, 217). Political theorists mainly discuss (1) and (2); they are mainly concerned with what the duty of civility requires public officials to speak and think (Neal 2008;Rickless 2001;Thunder 2006, 144-47). However, (3) is no less important because it outlines how the duty of civility requires public officials to act. 10 Facing political disagreements, officials should maintain "a willingness to listen to others," as well as "a fairmindedness in deciding when accommodations to their views should reasonably be made" (Rawls 2005, 217). They 9 See Rawls (2005, 450). Recently, several philosophers have argued that Rawls's political liberalism cannot be reconciled with justice pluralism, though they have not examined in detail the impact of justice pluralism on the assurance problem. See Gaus (2016, 150-54), Kogelmann (2017), Thrasher and Vallier (2018, 399-400), and Vallier (2019b). Although this article aims to reply to the cheap talk critique, our arguments about civic deeds can also show how a well-ordered society can be stable, even if officials endorse different political conceptions of justice. For another reply to this criticism, see Weithman (2017), which defends the viability of political liberalism by showing that political autonomy remains available even when justice pluralism exists. 10 The deliberative attitudes mentioned in the duty of civility are also discussed in Boettcher (2020), which argues that public reason theorists, no matter whether they endorse the consensus conception or the convergence conception of public justification, should encourage these attitudes.
should also act in a way such that their actions "best advance the common good" (Rawls 2005, 219-20). Their actions should show that, instead of advancing their self-interests or pursuing sectarian goals, they act in a way that is publicly justifiable to other officials.
Rawls never clearly defined what actions are required by the duty of civility. Nevertheless, by summarizing the description above, we can conclude that the duty of civility requires public officials to act in ways that are publicspirited and answerable to others. We shall call these actions civic deeds. The following examples can illustrate civic deeds: (A) When Audrey supports or rejects a law or policy, she patiently explains her arguments to the public in terms of public reasons, especially in terms of why she upholds this position after having carefully considered counterarguments. (B) When Audrey encounters other officials who oppose her view, she is frequently willing to listen to the arguments of the other officials respectfully. (C) During the discussion, Audrey is willing to contribute to public discussions by improving the opponents' arguments, even if this may strengthen their position. (D) Whenever opponents point out a weakness in Audrey's argument, she can look at her own position impartially. If the weakness truly exists, Audrey acknowledges the insights of her opponents and revises her view accordingly.
This list of civic deeds is not supposed to be exhaustive. Other actions that reflect public spirit and answerability may also be included. These actions provide a basic standard that enables officials to evaluate the trustworthiness of each other. Although officials endorse different conceptions of justice, they would identify each other as reasonable officials if they observed each other performing civic deeds over time.
We shall use the example of Audrey and Bruno again. Suppose that Audrey, in addition to using public reasons to justify the nationalization of Facebook, also performed the civic deeds listed above in political discussions over a long period of time, observable to Bruno. The civic deeds A and B show that Audrey has spent enormous time and effort to understand the opponents' views and respond to them seriously. It may be easy for Audrey to pay lip service to the importance of mutual respect and public deliberation, but her consistent answerable behavior shows that Audrey is truly reasonable. In addition, the civic deeds C and D show that Audrey is public-spirited and is willing to sacrifice her possible gains for the common good. Her main goal is to make the best decision for the public, even if the political outcome is not the most favorable result for her. The willingness to give up these possible gains shows her disposition to fairly respect the interests of others and deliberate impartially. This long-term record of civic deeds shows that Audrey has always been willing to abide by the duty of civility, and thus she is a reasonable official.
A real-world example of how civic deeds generate trust is the famous friendship between two United States Supreme Court justices, Ruth Bader Ginsburg and Antonin Scalia. Scalia was a staunch conservative who believed in originalism, whereas Ginsburg was a radical liberal who believed in a living constitution. 11 During their time together on the Supreme Court, they were seldom on the same side on controversial issues and neither shied away from voicing their opinions. Nevertheless, they are also famous for their close friendship. Ginsburg warmly described that "[w]e were best buddies" after Scalia died (National Public Radio 2016). Several reasons can explain their friendship. For example, they were both New Yorkers and close in age, and they shared a love of opera. Despite these similarities, a key reason why Ginsburg had a sense of trust in Scalia was his civic deeds. Ginsburg recalled that, in 1996, she had to write the majority opinion in U.S. v. Virginia, a landmark opinion that ended women's exclusion from the Virginia Military Institute. Scalia, obviously, opposed this opinion because "[t]his is not the interpretation of a Constitution…but the creation of one" (Scalia 2020, 178). Although they disagreed with each other, Ginsburg was grateful for how Scalia disagreed. Before Ginsburg completed the majority opinion, Scalia kindly gave her a copy of his dissent so Ginsburg could read it in advance and further improve her opinion. As Ginsburg admitted, "[t]hinking about fitting responses consumed my weekend, but I was glad to have the extra days to adjust the Court's opinion. My final draft was much improved thanks to Justice Scalia's searing criticism" (Ginsburg 2016, 249).
In our terms, Scalia here performed a civic deed. He showed his public spirit because his action enabled the decision of the Supreme Court to be more wellgrounded, even though he personally disagreed with the decision. 12 Furthermore, this kind of civic deed 11 Originalism insists that the Court should interpret the text of the Constitution according to its discoverable public meaning at the time of its initial adoption. In contrast, the living constitution approach suggests that the Framers of the Constitution intended the meaning of the text to adapt to the times. Thus, according to the latter approach, the Court could interpret certain clauses of the Constitution in a manner that might have surprised the Framers, whereas for originalism this is an abuse of judicial power. 12 One might argue that Scalia did a favor for Ginsburg because his side had already lost. At the moment that Scalia gave Ginsburg his had been continuously performed throughout Scalia's life. Ginsburg admitted that "whenever [she] wrote for the Court…Scalia gave [her] just the stimulation [she] needed to write a more persuasive account of the Court's decision" (Ginsburg 2016, 249). As Justice Elena Kagan described, no matter whether Scalia agreed or disagreed with her position, he always spent enormous time and effort in discussion with her, suggesting ways to improve her opinions (Roberts et al. 2016, 6). This shows the answerability of Scalia. He not only claimed to respect others' views but was also willing to spend enormous time and effort to engage with others in public deliberation. In brief, Scalia's civic deeds explain how he won the trust of other officials, despite ideological opposition. 13 One can also find relations of trust based on civic deeds in legislative settings. Widely considered the first Leader of the Opposition in the modern sense in Britain, Charles James Fox was an open supporter of both the American and French Revolutions and was disliked by King George III. Yet, his speeches in Parliament were praised by his fellow members of Parliament as vigorous, clear, lucid, and with great argumentative subtlety, to the point that even his political opponents would remark that Fox "gave some dignity to faction" (Namier and Brooke 1964;Thorne 1986). The respect was mutual: Fox also highly praised his chief opponent, Prime Minister William Pitt the Younger, remarking that Pitt's absence would "render every debate flat and uninterest-ing….
[O]ne feels as if there was something missing in the world" (Thorne 1986). Pitt had a reputation among his fellow members of Parliament as a "servant of the Commons, bound to interpret and carry out its wishes, rather than as its leader," even if the initiatives he supported were defeated (Namier and Brook 1964). Pitt and Fox became a famous pair of rivals in the early British Parliament who respected each other despite their opposition. They relentlessly appealed to reasoned arguments in their discussions, "believed that audiences could be swayed by argument," and were willing to join forces with opponents should they be convinced by their arguments (Leonard and Garnett 2019, xii-xiii, 256-57).
Accordingly, in light of experimental game theory, the stag hunt problem can be resolved when players can dissent, it was confirmed that Ginsburg's side had the majority in the Supreme Court. However, as many of Scalia's colleagues mentioned, Scalia did something similar throughout his career, before and after the Supreme Court made final decisions (Roberts et al. 2016). 13 Ginsburg also helped improve Scalia's arguments when he wrote opinions (Scalia 2017, 378). These reciprocal civic deeds explain how mutual assurance occurred between Scalia and Ginsburg. However, due to space constraints, we focus on Scalia's civic deeds.  others' actions and check those actions against others' words. The more information about others' past actions that is available, the more likely it is that players will cooperate (see Figure 1). Performing civic deeds is an action that can solve the authorship assurance problem. Offering public reasons is insufficient in convincing others that one is a reasonable citizen, since this signal is not costly enough. However, by performing civic deeds, officials can show their counterparts that they have continued to be answerable to other fellow officials and deliberate in a public-spirited way. Hence, they have performed the duty of civility throughout their public life. From these deeds, their counterparts can be assured that, even if they endorse different conceptions of justice, this official is still a trustworthy, reasonable citizen. Civic deeds, therefore, serve as a costly signal to resolve the authorship assurance problem. 14

The Problem of the Cheap Talk Critique
Having established the relationship between civic deeds and mutual assurance, we can now examine how the cheap talk critique misses the point. It wrongly assumes that the discourse of public reason alone is the source of mutual assurance. In fact, mutual assurance is created by the discourse of public reason together with civic deeds. It may be simple to present one's view in terms of public reason, but it is never easy to continuously perform civic deeds over time. To perform civic deeds continuously, an official has to pay three costs, and these costs enable other officials to confirm the reasonableness of this official. First, it takes enormous time and effort to be a public-spirited and answerable person in open discussions. Public deliberation can be very time-consuming.
14 Some might doubt the usefulness of civic deeds. They might argue that cheap talk is common in politics, and the political judgments of people in the real world are mainly affected by their partisan interests. It is hard to deny this, especially in polarized societies. However, some political scientists have argued that people are not necessarily driven by partisanship. If there are more contacts and civil exchanges among people, then they are more likely to reduce their prejudices and reflect on their political judgments (Mason 2018, 130). Collective deliberation can dramatically narrow differences between political opponents on issues where they are initially deeply polarized. Accordingly, several measures may have to be adopted to encourage officials to have more civic engagements with each other in the political domain and listen to each other's arguments more patiently, such as softening the party discipline rules and promoting critical, nondemonstrative parliamentary debates (Leydet 2015). We also suggest how civic deeds can be encouraged in civic education and public culture at the end of the next section.
Officials have to patiently explain their views to others and be willing to listen to the opinions of the opposing sides. Facing challenges to their decisions, officials have to attend carefully to the public reasons raised by others. Sometimes, to discover the best way forward, officials should be willing to put themselves in the shoes of the opposition and even put effort into improving the opposition's arguments. If an official continues to perform these civic deeds for years, and even decades (as Scalia did with Ginsburg), then it makes sense for other officials to believe that the official is a genuinely reasonable citizen, rather than a secretly unreasonable citizen. In fact, showing that one is willing to spend time and effort on certain actions has long been a widely used costly signal in many areas (Spence 1973, 359). Second, civic deeds involve the exercise of selfrestraint against certain psychological inclinations. People are naturally inclined to talk to those with whom they already agree. We also tend to perceive people who disagree with us in extreme, black-and-white, simplistic ways. Instead of sympathetically understanding these people, we are more likely to see them as unintelligent, immoral, or dishonest (Kahneman 2013). Civic deeds require officials to overcome these kinds of psychological tendencies. Officials should leave their comfort zones and engage with their political opponents respectfully. They should also be aware of being biased in an echo chamber and try their best to cure cognitive biases. These are mentally demanding endeavors and require a level of self-control that is well beyond average. A persistent willingness to restrain one's inclinations is therefore reliable evidence of one's commitment to reasonableness.
Third, civic deeds require people to give up possible gains for the sake of the common good. Sometimes, even if officials have enough power or support to make certain laws, they should refrain from doing so if they realize that these laws could be defeated by public reasons stronger than their own. Sometimes, officials should contribute to improving the arguments of their political opponents, even if this may eventually lead to a political outcome unfavorable to these officials. These altruistic actions enable other officials to understand that an official's commitment to public spirit and answerability is not merely lip service. Giving up one's possible gain is, in fact, a common costly signal in society. For example, hunters usually share their individually harvested resources unconditionally with the other hunters in their gathering. Some hunters consistently provide more than others while sharing, and these hunters are usually more trusted by others in the long run (Gintis et al. 2001, 116). Hence, like hunters, officials can also develop mutual trust by showing that they are willing to gain less than they could.
We shall conclude by addressing three possible objections.
(i) Are civic deeds truly a costly signal? Some might agree that the performance of civic deeds is costlier than using public reason in political speech. However, it does not mean that a civic deed is a costly signal. For a signal to be costly in political liberalism, the cost of delivering it must be much higher for unreasonable officials than it is for reasonable officials. Thus, unreasonable officials would find it difficult to pretend to be reasonable and deceive others. In brief, what matters is not the magnitude of the cost of the signal, but whether the signal can enable a player to differentiate credible players from others.
However, a civic deed is reliable because it imposes differential costs on reasonable and unreasonable officials. The cost of a signal can be understood in two ways. A signal has an absolute cost, which is the expenditure of energy and resources required to send the signal. It also has a relative cost, which is the proportion of an individual's overall budget that this individual is prepared to pay for delivering the signal (Fraser 2012, 272). For reasonable and unreasonable officials, the absolute cost of performing civic deeds over time is the same. Both have to spend large amounts of time and effort and endure psychological discomfort. Sometimes, they may have to forego possible gains. Nevertheless, their relative costs are different. Reasonable officials have the political virtue of honoring the duty of civility (Rawls 2005, 224). They have mentally prepared to act in the public interest and are used to spending time and effort in reciprocal activities. The cost is already taken into account in their life planning. Accordingly, they do not think it is a burden to perform civic deeds. Unreasonable officials, however, lack political virtue and thus reject the duty of civility. They do not think that they are supposed to spend so much time and effort on civic deeds.
For example, suppose a reasonable public official spends 5 hours in a political discussion. Nevertheless, the official knows that public deliberation is important; thus, he has already prepared to spend 10 hours in it. Hence, to this reasonable official, the relative cost of his civic deed is merely 0.5. In contrast, an unreasonable official who does not see the value of public deliberation may only be prepared to spend 1 hour in it. To this unreasonable official, the relative cost of a 5-hour political discussion is then 5. This explains why, although reasonable and unreasonable officials may spend the same time performing civic deeds, it is much costlier for unreasonable of-ficials to do so. Even if some unreasonable officials can bear these costs and perform civic deeds, it seems difficult for them to continue doing so over a long period of time. 15 (ii) What is the difference between civic deeds and conjectural discourse?
Recently, Kogelmann and Stich (2016) suggested that the assurance problem in political liberalism can be resolved by conjectural discourse. Officials can convince other officials to endorse their political proposal by offering nonpublic reasons that those other officials would accept. To offer such nonpublic reasons, officials have to be familiar with the comprehensive doctrine of other officials. This action, which is called conjecture by Rawls (2005, 465-66), can be seen as a kind of costly signal, for it takes time and effort to study the relevant comprehensive doctrines in order to offer such nonpublic reasons.
Here, we shall briefly explain the difference between our solution and conjectural discourse. 16 At first glance, both conjectural discourse and civic deeds are forms of deliberative engagement, which involve discussing and explaining ideas to others patiently. However, they are different because a civic deed involves offering and discussing public reasons, but conjectural discourse is about offering nonpublic reasons that others would accept. Although we do not deny that sometimes conjectural discourse can have a positive influence on public discussions (Wong 2019), we aim to show that the actions of exchanging public reasons themselves are sufficient in creating mutual assurance. Even if reasonable officials do not know each other's comprehensive doctrines, they can still develop mutual trust through civic deeds.
(iii) How can civic deeds be encouraged in democracies in the real world?
Due to space constraints, we shall only offer two brief suggestions. First, civic education should promote these deeds. Some political liberals suggest that a component of citizenship education should be to let students participate in formal debates concerning political issues (Neufeld 2022, 132-33;Wong 2021, 780). In these debates, students are required to use public reasons to justify their positions. We shall further suggest that civic deeds can also be encouraged in these debates. Instead of which side wins, what matters more is whether debaters can behave in a public-spirited and answerable manner.
For example, debaters sometimes distort opponents' arguments or present them in a biased way in order to win the match. This should be avoided. Rather, debaters should address the opponents' arguments as respectfully and sympathetically as they can. They should acknowledge the strength of the opponents' arguments and show how their arguments can be improved in light of the opponents' contribution. To encourage these civic deeds, the judges may include an item of "civility" when they award debaters' points. In sum, a debate match is more than a zero-sum competition. Debaters should take it as a collective activity that enables every participant to improve their viewpoints.
Second, citizens can publicly praise those politicians who continuously perform civic deeds. For example, inspired by the valuable friendship between Ginsburg and Scalia, Derrick Wang, a composer-librettist, created the opera Scalia/Ginsburg in 2015 to compliment these two justices on their mutual respect and civility (Los Angeles Times 2015). This opera has been widely acclaimed and continues to be produced across the United States and abroad. It shows a way that citizens can influence the public culture by raising people's awareness of the civic deeds of politicians.

Conclusion
This article suggests that civic deeds are effective means to resolve the authorship assurance problem. By showing this, it contributes to the "consensus versus convergence" debate in vogue among political liberals. Recently, several political theorists have proposed a convergence conception of political liberalism, or "convergence liberalism" for short, as an alternative to the Rawlsian "consensus" conception of political liberalism (Billingham 2016;Gaus 2011;Kogelmann and Stich 2016;Muldoon 2016;Vallier 2014a). Both convergence and consensus views agree that laws must be publicly justified to all reasonable citizens in order to be legitimate. The two views differ regarding the type of reasons that should be used for such public justifications. Convergence liberals hold that citizens are permitted to endorse a law for purely nonpublic reasons, so long as the law can be justified to all citizens in terms they can accept. Rawlsian shared public reasons are not necessary for public justification. 17 One of the major critiques frequently used by the convergence liberals is the cheap talk critique. According to convergence liberals, the consensus conception of political liberalism exag-gerates the function of public reason in resolving the assurance problem. Public reason, they argue, is only cheap talk. This constitutes a forceful criticism of the consensus conception.
Our article, however, shows the problem of the cheap talk critique. The convergence liberals wrongly believe that in the consensus conception, the cost of signaling reasonableness is that of speaking in terms of public reasons. Instead, in the consensus conception, the true cost is that of performing civic deeds over time. Civic deeds play a crucial role in generating mutual assurance. Hence, our article provides a robust defense of the consensus conception. 18 In addition, our article suggests new directions for discussing political liberalism and mutual assurance. Both consensus and convergence views have been concerned with the role of talk or discourse with respect to public reason. The focus of disagreement is merely whether speaking in terms of public reason can generate assurance. By contrast, our article argues that deeds should be given more attention. No matter what is said, what is done is centrally important when people assess each other's trustworthiness. Talk may be cheap, but deeds seldom cheat.