State Speech as a Response to Hate Speech: Assessing ‘ Transformative Liberalism ’

‘ Transformative liberals ’ believe that the state should use its non-coercive capacities to counter hateful speech and practices, by seeking to transform the views of those who hold hateful and discriminatory beliefs. This paper critically assesses transformative liberalism, with a particular focus on the theory developed by Corey Brettschneider. For Brettschneider, the state should engage in ‘ democratic persuasion ’ by speaking out against views that are incompatible with the ideal of free and equal citizenship, and refusing to fund or subsidise civil society groups that hold such views. My critique has five parts. I first rebut two central justifications for transformative liberalism, regarding complicity and the undermining of equal citizenship. Second, I show that some of the central policies that Brettschneider advocates are in fact coercive. Third, I raise concerns about the nature of the complex and contestable judgments that transformative liberalism requires the state to make. Fourth, I argue that Brettschneider ’ s view has various troubling implications. Finally, I argue that many of these problems derive from his adoption of a thick conception of free and equal citizenship, resulting in an overly broad definition of hateful viewpoints and of hate speech. A defensible version of transformative liberalism would use a significantly narrower conception.

student groups within state universities can rightly be denied official recognition, and associated benefits such as access to funding. He thus endorses the U.S. Supreme Court's decision in Christian Legal Society v. Martinez, 6 which upheld Hastings College of the Law's decision to refuse to recognise CLS as an official student group-although for different reasons than those offered by the Court. 7 Ultimately, Brettschneider hopes that democratic persuasion will lead all citizens to come to affirm FEC, and to ensure that their comprehensive doctrines are compatible with that ideal. The state can thus tackle hateful speech and practices, without coercive regulation.
I think Brettschneider is right that state speech can be an appropriate response to genuine hate speech. But his transformative liberalism ranges wider than this, leading to several problems, which I explore in this paper.
Before proceeding, it is worth noting that the debate over transformative liberalism is closely related to the debate within political liberalism concerning how the state should respond to 'unreasonable citizens'. Indeed, Quong (2011: 299-312) uses the example of hate speech to illustrate his argument that there is no right to act unreasonably, so the putative rights of unreasonable citizens-rights of free speech in this case-can be permissibly restricted. Others, such as Schwarztman (2012) endorse versions of 'reasoning from conjecture', whereby we seek to persuade individuals to abandon their unreasonableness, rather than curtailing their rights. Such proposals are in a similar spirit to transformative liberalism, since they also seek to ensure that all citizens come to align their comprehensive doctrines with liberal democratic ideals. 8 While I will not explicitly engage with this related debate, many of my arguments are relevant to it.

What Justifies Democratic Persuasion?
A crucial initial question for transformative liberals is what justifies democratic persuasion, so I first show that two putative justifications fail.
One of Brettschneider's central arguments for why the state should engage in democratic persuasion is that otherwise it will be complicit, or seen as complicit, with hateful viewpoints (e.g. 43-5). 9 The liberal state protects the rights of groups that oppose FEC to associate, express their views, and even to widely publicise those views. This state protection of hateful viewpoints could give the impression that the state endorses such views, thus making it complicit with them. This problem of complicity is magnified when the state subsidises or grants tax-exempt status to a group. Engaging in democratic persuasion, by both publicly opposing hateful viewpoints and withdrawing financial support from groups that express them, is necessary if the state is to avoid complicity.
Brettschneider vacillates somewhat between saying that protecting rights without engaging in democratic persuasion actually makes the state complicit with hateful views and saying that it risks the state being seen as complicit. 10 These are distinct claims. Justifying the former 6 561 U.S. 661 (2011). 7 The Court presented its decision as viewpoint neutral, whereas Brettschneider holds that it is not neutral, but there is no need to be neutral with respect to views that violate FEC. For useful analysis of this case, focused on the question of congruence between religious and political values, see McClain (2011). 8 Although they often focus on individuals as the agents who engage in conjecture, rather than the state. 9 Cohen (2012: 395-7) also appeals to state complicity when defending transformative liberal policies. 10 Brettschneider talks both about the state 'being considered' complicit and how it can 'avoid complicity' at 17, 114,119. requires an account of what makes an agent complicit in a wrong, while justifying the latter instead requires evidence that there would be a widespread perception of complicity. Further, we could have cases of actual complicity that isn't widely perceived, and of mistakenly perceived complicity.
The claim of actual complicity seems implausible with respect to the protection of rights such as freedom of speech and association. On Brettschneider's view, the state should equally protect these rights for all citizens, in a viewpoint neutral way. If the state is complicit with some of these views, then presumably it is complicit with all of them. But it is unclear why we should consider the state complicit with all of the myriad incompatible views that citizens express. Indeed, it is a familiar idea that the state can protect rights without being complicit in the ways that citizens use those rights. For example, Rosenblum (1998: 90) notes that Bfreedom of association is no stamp of public approval of the internal life of an association in a liberal democracy.T here are also problems with the claim that the state will be perceived as complicit with hateful viewpoints. Brettschneider does not offer any empirical evidence for this claim, and it is not clear why citizens would have this perception, given their awareness that the state protects rights to express all viewpoints. I, for one, do not think there is a widespread perception that the state is complicit with all the views that it protects the right to express, or all of the practices of civil society associations.
Even state subsidy does not make the state complicit in the way Brettschneider assumes. Tax-exempt status, for example, does not make the state complicit in all of the exempted group's beliefs, values, or speech. The state grants tax-exempt status to a wide range of charitable and religious organisations, based on the state's own purposes-encouraging an active civil society, promoting free speech, supporting charitable activity, and so on. None of this implies endorsement of all of the views of the groups who are granted this status. Similar comments apply to state funding for organisations to provide a particular service. There also does not seem to be any widespread perception of complicity in subsidy cases. If there was such a perception then the state could presumably remove this impression by pointing out the facts I have noted. 11 Both the actual and perceived complicity claims thus appear implausible, with respect to both protections of liberal rights and state subsidy. To the extent that such claims are essential to Brettschneider's argument, his case for democratic persuasion is ill-founded.
My argument has followed Brettschneider in focusing on state complicity created by protecting rights or subsidising groups in a way that appears to endorse hateful viewpoints. However, one might argue that the issues of complicity and endorsement are in fact distinct, with the former having to do with a particular kind of involvement with wrongdoing, such as causal contribution. 12 A state that fails to enact transformative policies might be complicit with denials of FEC in this sense, even if it is not plausibly interpreted as endorsing these hateful viewpoints. While true, this claim cannot be used to justify transformative liberalism, for two reasons. First, not all complicity qua causal contribution can be considered wrongful, since this would be too expansive. The complicity involved in the cases with which we are concerned is plausibly considered non-wrongful-partly in the light of the comments I made above. Second, transformative policies do not remove such complicity anyway, since the transformative state still knowingly permits the wrongdoing and thus causally contributes to it in this sense. 13 11 Koppelman (201411 Koppelman ( : 1025 also dismisses the perception-based complicity argument. 12 See Lepora and Goodin (2013). 13 More would need to be said to properly justify these two points, but I lack space to develop them further here.
The second failed putative justification for transformative liberalism relies on the claim that FEC will actually be undermined in the absence of democratic persuasion. As Archard (2014: 138) notes, Brettschneider's view might Bbe supported by the conflation of a claim that hateful speech affirms or endorses inequality and a claim that such speech directly subverts equality.Ŵ e must distinguish the idea that hateful speech endorses inequality of status from the claim that such speech contributes to, or risks contributing to, such inequality. The former does not entail the latter. This is especially the case since 'hateful speech' here denotes all speech that contradicts FEC, which is a broader category than 'hate speech' as it is traditionally construed. 14 Thus, even if (contra Archard) Waldron (2012: 169) is right that expressions of hate can themselves constitute the dispelling of public assurance of all citizens' equal standing, this would not be true of all 'hateful speech' in Brettschneider's sense, so would not justify transformative policies against all such speech.
Archard's comment extends to some of the discriminatory practices that Brettschneider says make groups liable to democratic persuasion. In many cases, the practice itself does not diminish individuals' status as free and equal citizens. They are prevented from being members of particular groups, or from holding certain offices or fulfilling certain functions within those groups. But this does not in itself constitute political inequality. 15 Even if it is true that groups that practice certain kinds of discrimination thereby express opposition to, or fail to respect, FEC, this is not the same as these practices undermining FEC. Taking these distinctions seriously implies that Brettschneider's justification for democratic persuasion often cannot be that the targeted individuals and groups are actually undermining equality of status.
At times, Brettschneider appears to make the complicity claim and/or assumptions that FEC is being undermined central to his argument. His account is weakened to the extent that it depends on those justifications. To avoid this, his view instead must be that the state ought to promote the ideal of FEC, and oppose conflicting views, irrespective of questions of complicity and whether FEC is undermined. The view thus centres directly on the claim that the state has a duty to try to create congruence between liberal values and the beliefs, values, speech, and practices of individual citizens and civil society groups. This is significant, because my arguments in later sections suggest that this duty is more limited than transformative liberals suppose.

The Means-Based Limit
Let's turn to the first constraint that Brettschneider places on democratic persuasion: it must be non-coercive. It is not clear that all of the methods he endorses abide by this limit. This issue particularly arises with respect to the use of state subsidy, a key tool for Brettschneider's transformative state. Consider the state's threat to withdraw a non-profit organisation's existing tax-exempt status, unless it abandons a viewpoint or amends a policy that the state considers to violate FEC. Is this coercive? Some might hold that this threat is coercive simply because the state is its source. The inequality in power between the state and citizens is sufficiently great that all deliberate attempts by the state to influence citizens' beliefs or expression involve coercion. On this view, all forms of democratic persuasion would be coercive. This seems too strong, however.
At least some state actions can surely be non-coercive. Pure state speech does not necessarily seem coercive, for example (152).
We thus need an account of coercion that distinguishes coercive from non-coercive state actions. Brettschneider draws on Nozick's prominent account. On Brettschneider's construal, coercion is defined as Bthe state threatening to impose a sanction or punishment on an individual or group of individuals with the aim of prohibiting a particular action, expression, or holding of a belief^(88).
Using this definition, much turns on what counts as 'aiming to prohibit'. When the state threatens to withdraw tax-exempt status it clearly aims that the targeted group will respond by changing the relevant practice or refraining from the targeted speech. But Brettschneider denies that this amounts to the state 'aiming to prohibit' that practice, belief, or speech, because the group is permitted to respond by accepting the loss of its tax-exempt status. The group has a right to resist democratic persuasion, so it does not constitute a prohibition or coercion.
We should reject this account of coercion, however, because it allows any threat to be reinterpreted as non-coercive. The state imposing a fine on certain conduct-parking one's car in a restricted area, say-is an archetypal case of coercion. But on Brettschneider's account the state can deny that it is coercive, because it leaves open the possibility of parking in the restricted area-one merely needs to pay the fine. The state aims that drivers will not park there, but this does not amount to a prohibition, because those willing to bear the consequences are free to park. Any threat can be reconstrued in this way-certain consequences are imposed on some act, leaving individuals free to decide whether to bear those consequences.
This problem comes from Brettschneider's use and interpretation of 'aiming to prohibit'. Nozick's original definition appealed to a different aim on the part of the coercer: BP aims to keep Q from choosing to perform action A^. 16 The coercer intends that the recipient will choose not to perform the action. The parking fine clearly fulfils this condition. But so does the threat to withdraw tax-exempt status from a group that presently enjoys it. The state aims that the targetted group will choose to change their practice, belief, or speech, so as to avoid losing their tax-exempt status. The threat is thus coercive on Nozick's account. 17 And we have just seen that there is good reason to favour Nozick's definition over Brettschneider's reinterpretation.
Threatening to withdraw a group's existing tax-exempt status unless the group amends a practice imposes a new cost upon the continuance of the practice, with the aim of causing the group to abandon it. In my view, this is coercive. 18 16 Quoted in Anderson (2015: §2.1). 17 Song (2014Song ( : 1053 also points this out. A complication here is that Nozick's definition includes a success condition: P coerces Q only if P's threat leads Q not to do A. This would mean that the threat to withdraw taxexempt status is only coercive if the group responds by changing its practice or beliefs. Those who resist democratic persuasion are not coerced. The success condition is controversial, and Brettschneider explicitly abandons it. Either way, the key point is that Nozick's account treats this case as analogous to the parking fine case. A different additional condition, suggested by an anonymous reviewer, is that the threat must be plausible, in the sense that the P must be in a position to put the threat into practice (or, perhaps, that Q must believe this). An 'empty thereat' is not coercive. This seems plausible, but does not affect my argument here. 18 An anonymous reviewer highlighted that some might hold that state speech against an illiberal group could also be coercive on this account, since it threatens to impose a 'symbolic sanction', via official public disapprobation. If this is true then it would further limit democratic persuasion, if the state is to abide by the means-based limit. I am agnostic on this issue here, but have argued elsewhere that we should recognise the force of all transformative policies. See Billingham (forthcoming). 162) himself recognises that imposing differential taxes on individuals based on their viewpoint would be coercive. Why are matters different for non-profit groups? Brettschneider's discussion points to two possibilities.
First, he implies that the difference depends on common perceptions of what is coercive. People would generally consider viewpoint discrimination within individual taxation as coercive, so it counts as such (138-9). Similarly, certain means of democratic persuasion should be ruled out because they could reasonably be considered coercive (113). As Song (2014Song ( : 1055 notes, however, a group threatened with the loss of its tax-exempt status will also likely view this as coercive. This does not seem unreasonable-especially in the light of my argument above. A second possibility is that individuals have a right against their tax burden being determined based on their viewpoint, but groups do not. But if this means that coercive democratic persuasion is permissible as long as rights are not violated then it simply abandons the means-based limit. Alternatively, the claim might be that the threat is not coercive because it does not violate rights-whereas varying individuals' tax burden does violate rights, so is coercive. But this invokes a moralised conception of coercion, such that we need a prior account of rights in order to identify coercion, thereby undermining the independent force of the means-based limit. Perhaps for this reason, Brettschneider (185-6, fn. 31) states that his conception is not moralised. The rights-based argument cannot distinguish between individuals and groups, because it either abandons or nullifies the force of the means-based limit.
Brettschneider does seem to make this rights-based move elsewhere, however. He admits that a state employee being fired for holding views inconsistent with FEC seems coercive, but says it is permissible because it is not a case of coercion that violates rights (102). Again, this either amounts to an abandonment of the means-based limit, since some coercion is permitted, or involves adopting a moralised conception of coercion, thereby depriving the means-based limit of independent force.
Brettschneider's claims regarding coercion are thus highly disputable. Of course, one response to this is to abandon the means-based limit and permit coercive uses of state subsidy within democratic persuasion. This might seem an attractive response, given that state subsidy might well be particularly effective at achieving transformative ends. Brettschneider often appeals to such strategies in response to worries that democratic persuasion will be ineffective (e.g. 110).
The problem with permitting some coercion within democratic persuasion, however, is that one must then explain why other coercive policies, such as straightforward prohibitions, are impermissible. Brettschneider, for one, is firmly committed to the means-based limit, due to his endorsement of US First Amendment jurisprudence.
Therefore, while my arguments in this section do not directly oppose transformative liberalism, they do pose a dilemma. Transformative liberals must either jettison the meansbased limit or abandon some potentially effective means of democratic persuasion. Neither option is costless.

The Substance-Based Limit
Alongside the 'means-based limit', Brettschneider endorses a 'substance-based limit', according to which democratic persuasion must only respond to violations of FEC. More specifically, the state is only permitted to criticise, and withhold or withdraw funding from, individuals and groups who hold beliefs, or engage in speech or other practices, that are opposed to, or implausibly consistent with, FEC (65, 90).
In line with the political liberal tradition,  insists that FEC is a purely political ideal. It concerns the equal status of all citizens as members of the political community possessing Rawls's two moral powers: the capacities for a sense of justice and a conception of the good. It is centrally secured by all citizens' rights being equally respected; the wide array of rights protected in liberal democracies is grounded in FEC. One can reject other egalitarian ideas-such as theological or metaphysical notions of equality, and egalitarian theories of distributive justice-while nonetheless endorsing FEC.
Difficulties arise, however, when we consider precisely what beliefs, speech, and practices conflict with this political ideal. Some of Brettschneider's examples involve groups, such as the Klu Klux Klan, that clearly violate FEC by explicitly denying the basic equality of all citizens. Many cases are less clear-cut, however. For example, it is not obvious whether advocacy of the death penalty, opposition to state-mandated universal healthcare coverage, or the practice of male-only priesthood are hostile to, or implausibly compatible with, FEC.
In part, this is because the standard of 'implausibly compatible' is vague. Does this mean that there merely needs to be some argument for compatibility for democratic persuasion to be impermissible? Or does the most plausible argument, or the overall balance of arguments, need to favour compatibility?
Brettschneider's answers seem to shift between cases. Early on, he states that democratic persuasion should only be directed at beliefs that Bclearly conflict with the ideal of free and equal citizenship^(47). Beliefs that can plausibly be seen as compatible should not be targeted, even if their compatibility is contested (90). This suggests that beliefs are exempt from democratic persuasion whenever there is some minimally plausible argument for compatibility with FEC.
At other times, Brettschneider appears to raise the bar. Consider again his discussion of the Christian Legal Society. Brettschneider holds that the CLS's membership policy amounts to a denial of the free and equal citizenship of gay citizens, because it involves status discrimination against them. In reaching this conclusion, he dismisses any distinction between gay citizens' status and their choice of behaviour, on the grounds that it is not a choice to be gay.
Defenders of the CLS could offer several replies. First, they might argue that one can endorse gay citizens' freedom and equality while opposing homosexual conduct. Even if it is not a choice to be gay, it is nonetheless a choice to engage in sexual relations-just as it is for heterosexuals-and discrimination based on behaviour is not equivalent to discrimination based on status. Second, as Vallier (2016: 14-5) notes, some Christians believe on theological grounds that there is no such thing as sexual identity, contra the assumption of Brettschneider's argument. Third, as Garnett (2012: 221) points out, the CLS did not refuse to admit gay members. It refused to admit anyone who did not live in accordance with traditional Christian morality, whether or not they are gay, including those who engage in heterosexual sexual activity outside of marriage. Finally, Garnett (2012: 219) argues that Bit does not usually demean a person, or call into question a person's equal ultimate worth, to exclude her from an association because she does not embrace the association's aims or reasons for being.^McConnell (2000: 472) concurs: BIt is not invidiously discriminatory for a private association committed to certain beliefs and values to limit itself to persons who share those beliefs and values.T hese arguments for the compatibility of the CLS's policy with FEC might be mistaken. But they are not obviously implausible. Brettschneider must be applying a higher standard here, such as that democratic persuasion is justified when the arguments for incompatibility are more plausible than those for compatibility.
In other cases, however, Brettschneider shows greater caution. When addressing the debate over the burqa and headscarf,  takes the fact that it is unclear that these items of closing are at odds with equal citizenship to show that democratic persuasion should not be used. Here, the fact an argument can be made either way seems sufficient to rule out democratic persuasion. The same applies in Brettschneider's (135) discussion of the Catholic Church's male-only priesthood (see also Brettschneider 2014Brettschneider : 1084. Brettschneider says that the ban on female priests is not necessarily the same as refusing women equal status in society at large. The question of the compatibility of the Church's policy with FEC is at least ambiguous, such that it should not be subject to democratic persuasion. 19 These examples illustrate the fact that the standard to be applied in determining whether a group ought to be subject to democratic persuasion is unclear-due both to vagueness within the ideal of FEC itself and apparent shifts in the standard for compatibility. The difficulty here can be seen in the fact that various critics have questioned Brettschneider's judgments on specific cases and the consistency of his judgments across cases (Song 2014(Song : 1051Calabresi 2014Calabresi : 1012Fair 2012: 119).
A major concern about transformative liberalism is that it requires these complex and contestable judgments to be made by the state, especially given the potentially great implications of these judgments for the groups concerned. There are several interrelated reasons to be wary of granting the state such extensive authority to interfere with expressive and associational freedoms.
Most obviously, it is likely that the state will often make incorrect judgments, given the unclarity of the standards of judgment. Brettschneider's own judgments in particular cases often depend on fairly specific features of the case and its context, since these details are often key to determining whether inegalitarian speech and practices amount to denials of free and equal citizenship in particular. For example, his judgment that the Catholic Church's male-only priesthood does not amount to a denial of women's equal status in society at large depends on specific claims about the Church's history, theology, and attitude toward female public officeholders. This makes it hard to develop general rules and policies, and thus leaves the state with a large degree of discretionary power. Liberals should particularly fear that the state is likely to intervene too much, demanding unreasonable levels of congruence, in ways that excessively interfere with the freedom of speech and association of groups whose views or practices do not fully cohere with liberal egalitarian ideals. Importantly, the unclarity here also creates uncertainty for citizens, who cannot be sure which expression or practices will be deemed to make them liable to democratic persuasion, especially since the state's understanding of FEC is likely to shift over time. This is in tension with prevailing conceptions of the rule of law and its value, which emphasise that certainty about what will attract state sanction or censure is crucial. 20 Of course, the liberal state cannot avoid some interference with civil society groups-it must enforce property, taxation, and employment laws, for example. And this sometimes involves complicated judgments based on somewhat unclear standards. While this always creates dangers of excessive intervention, the mere fact that a theory requires complex judgements involving vague standards is not sufficient reason to reject it-although it does provide reason to limit this danger.
The deeper problem for transformative liberalism, however, involves the particular kinds of judgments that are required. Many of the beliefs and practices that are potential targets of democratic persuasion are rooted in theological traditions and doctrines, and rest upon theological conceptions of personal identity, God's creative ordering, natural law, the role of intentions in moral evaluation, the nature of salvation, and so on (Vallier 2016: 13-6). Judging the meaning of a group's expressions and practices in relation to a substantive ideal of FEC thus usually requires an evaluation of these theologically-grounded arguments, entangling the state in metaphysical and theological adjudication.
One objection to this is simply that it will cause resentment among those whose views are subjected to interrogation and judgment, many of whom will believe their views to be compatible with FEC. This will be true even for those whose views are ultimately ruled 'plausibly compatible' with FEC. But it will particularly be the case for those whose own understanding of their doctrines is overridden, and deemed beyond the liberal pale.
Again, the liberal state inevitably makes some judgments concerning civil society groups that run against those groups' self-understandings. It must decide when others' rights have been violated, and can rule that such violations have occurred even when the perpetrators deny this. But Brettschneider's transformative liberalism calls for particular kinds of judgment that the liberal state has good reason to avoid as far as possible: substantive theological adjudication. The state does not merely decide whether rights have been violated, but adjudicates on the acceptability of groups' doctrines-and the expressions and practices they motivate-at the bar of liberalism.
Part of the reason the state should avoid this kind of adjudication is capacity-based. The state does not generally have the knowledge or abilities required to make these judgments. Concern about the state's theological inabilities is a regular refrain in discussions of law and religion, leading to resistance to courts being required to resolve disagreements about the content of religions or to evaluative theologically-grounded materials (Greenawalt 2006: 27-8).
This kind of theological adjudication is also objectionable for more substantive reasons. Groups have a legitimate interest in forming their own doctrines and shaping their practices in line with those doctrines, without their public standing depending upon the state's judgment of the meaning of this conduct and its compatibility with a thick liberal ideal. Freedom of religion, expression, and association should protect groups from the kinds of contestable judgments that are involved in implementing Brettschneider's expansive version of transformative liberalism. As far as possible, we should avoid the state being an arbiter of religious beliefs and activities.
Groups' interests here are clearly not absolute or always overriding. They cannot justify clear violations of basic rights, and do not grant immunity from limited forms of democratic persuasion when there is absolutely no plausible compatibility with even a minimal understanding of equal citizenship. The liberal state is not a passive state. But that state does have good reason to avoid substantive theological adjudication wherever possible, especially when that adjudication is based upon such unclear standards as Brettschneider's expansive notion of FEC. 21 These final comments point to distinctions that I will further discuss below. First, however, I will raise another concern regarding Brettschneider's account by considering more of its implications.

Troubling Implications
Even on Brettschneider's own stated judgments, his view has implications that I consider troubling. One of these implications is that state employees who hold certain conservative religious beliefs are liable to lose their jobs. This follows from  argument that a public school teacher who is a member of the Klu Klux Klan should be fired on the basis of her hateful views. 22 He presents two reasons for this. First, the teacher's views undermine her capacity to promote FEC, which is a duty that she holds as a state representative. Second, the state's duty not to provide positive support or funding for hateful viewpoints includes not employing this individual.
Brettschneider is clear that the Klan member ought to be fired even if she never expresses her views in the classroom; hateful expression outside the school is sufficient grounds for dismissal. But his two arguments also suggest that the mere holding of such views could be sufficient grounds, even if she never expresses them publicly. The fact she holds these views might undermine her ability to convey a message of equal citizenship in the classroom, and means the state is providing financial support to a Klan member.
Whether or not Brettschneider's view does extend to merely holding hateful viewpoints, or is limited to their public expression, 23 his argument presumably extends to those who hold religious views that are deemed incompatible, or implausibly compatible, with FEC. Consider a religious individual who endorses traditional gender roles within the home. He believes that wives ought to be the primary caregivers for children, while husbands have primary responsibility for earning an income to support the family. These beliefs seem to conflict with Brettschneider's understanding of FEC. They seemingly involve denying that men and women have an equal public status with respect to the economic sphere, which is a central sphere in which individuals ought to relate as free and equal, and which greatly influences individuals' life-chances and status as citizens. The individual who holds (or publicly expresses) these traditionalist views would thus be liable to lose his job as a public schoolteacher-or indeed as any kind of state employee. This is true even if he does not express these views in the classroom (or workplace) and treats his male and female students (or colleagues) completely equally.
Similar comments apply to those who oppose state-recognised gay marriage. Brettschneider (134-6) strongly hints that opposing gay marriage is incompatible with respect for FEC. It thus seems that any citizens who (publicly) oppose gay marriage would be excluded from public employment. Again, this will include many citizens with conservative religious views.
An initial observation about this conclusion is that it involves a fairly severe restriction of the job opportunities for these individuals, given the proportion of workers employed in the public sector. According to the OECD (2017: 91), in 2015 15.3% of employees in the USA and 16.4% of employees in the UK were employed in the public sector. The OECD average was 18.1%. 24 One response to this would be to limit the restriction to public sector jobs where individuals' views might directly impinge upon their conduct in their role. Teachers are a prime example, due to their influence upon their students. Many public sector employees' objectionable views cannot have such influence, so they need not be dismissed. Some of Brettschneider's comments about teachers' responsibilities to convey the message of equal citizenship to students might point in this direction.
It is not clear that Brettschneider can make this move, however, given his claims that all public workers have a duty to affirm and promote FEC and that the state is complicit in the viewpoints of all those it funds, including its employees. Further, even if one does make this move, it still means that many adherents to conservative religious doctrines cannot be teachers, politicians, civil servants, doctors, and so on-and will be to that extent marginalised within public life.
Beyond this, it seems troubling to exclude from public employment everyone whose beliefs and values are not fully congruent with liberal egalitarian ideals. Many such individuals are good-willed, respectful, and civil. They do not think others are of less moral worth. They simply hold traditional moral views that many consider incorrect. Further, they often recognise that their views are controversial, and are willing to comply with the requirements of their jobs, including refraining from using their public role to impose or promote their private views. Such individuals should not be denied public employment due to their moral and religious views.
I also do not think that Brettschneider's arguments can justify dismissal. Individuals' duty to promote FEC is insufficient to justify this broad restriction in job opportunities, and for reasons discussed above I do not think that the state is complicit with all of public employees' views.
We should also consider the implications of Brettschneider's theory for groups. Those holding the views I have mentioned will be liable to face democratic persuasion.  hints that the Catholic Church should lose its tax-exempt status if it publicly opposes gay marriage, and many other churches, mosques, and Christian and Muslim charitable organisations would face this fate. They would also be barred from providing government-funded social services-even where their views do not affect their effectiveness as social service providers. 25 This could well result in some religiously-based charitable organisations shutting down.
One result of this would be members of the relevant groups feeling alienated from society. Brettschneider intends democratic persuasion as a means by which citizens are led to reconsider beliefs that conflict with FEC, and to reflectively revise those beliefs in order to align them with that ideal. In reality, however, it seems equally (if not more) likely that many will feel that they are under attack and being marginalised within society. Under pressure, the views of some might become more, rather than less, objectionable.
It is also likely that many will invest in parallel structures-religious schools, explicitly sectarian charities, and so on. Civil society would thus become more balkanised. The cost of the government avoiding 'complicity' with illiberal views might be a less integrated civil society. Liberals have traditionally emphasised the importance of intermediate institutions, including religious institutions, between the state and individuals, both as a check on state power and as a central location for individual flourishing. 26 One of the central purposes of taxexempt status is to support a rich, vibrant, and diverse civil society. 27 Heavy-handed forms of democratic persuasion would have the opposite effect. 28 Brettschneider might offer two responses here. First, his view does not require the state actively to seek out groups to oppose whenever it can, rebut every opponent of FEC, or scrutinise the views of every public employee. Transformative liberalism need not have the extreme implications I have implied. Second, my argument merely shows that there might be consequentialist reasons not to implement all of the possible implications of his view, without undermining the central deontological claim that the state has an obligation to defend the ideal of FEC and avoid providing support to its opponents.
The first response here is well taken. But it does not remove the fact that Brettschneider does think that state employees can be fired for opposing his ideal of FEC and that civil society associations should face the pressures I have discussed. Even if not carried out in an exhaustive way, these objectionable implications are part of the view, and follow from its logic.
Further, on the second point, I think these implications point to a deontological objection to Brettschneider's view. They show that he assigns the state excessive authority to intervene in the workings of civil society, placing direct pressure on large numbers of good-willed individuals and groups to change their beliefs. Freedom of belief and association should protect citizens from this kind of intrusion. The state's legitimate power to promote congruence does not extend as far as Brettschneider claims. This is not simply about the view's consequences, in terms of alienation and balkanisation, but about the limits of the state's legitimate transformative power. I develop this point in the next section.

Free and Equal Citizenship: Thin and Thick
Implicit in much of what I have said thus far is the fact that the ideal of FEC can be defined in thinner and thicker ways. How thick a conception a transformative liberal adopts significantly affects the scope of democratic persuasion, by determining what falls under the label 'hateful and discriminatory viewpoints', and thus what constitutes 'hate speech'. This final section explores this point more explicitly.
At minimum, endorsement of the ideal of FEC involves recognising that all citizens should have their basic rights protected by law, due to sharing an equal public status. On a minimal specification, these rights would include negative rights such as bodily integrity, freedom of expression and association, and so on, along with the right to a fair trial and political rights such as the right to vote and stand for office.
Some groups reject even this minimal ideal of FEC; for example they believe that some citizens should be denied the vote or have an inferior moral and political status. Some of Brettschneider's most plausible examples involve such groups, like the Klu Klux Klan, who certainly hold hateful viewpoints.
Many of those Brettschneider considers appropriate targets of democratic persuasion do not violate the minimal conception of FEC, however. For example, on his view, the CLS's membership policies amount to a denial of the equal citizenship of gay citizens. The CLS's view that individuals who engage in, or indeed endorse, sexual relations outside heterosexual marriage cannot live up to their ideal of the 'Christian lawyer' is incompatible with recognising the equal public status of all citizens. On the minimal conception, this is not the case. The ideal of the Christian lawyer is not a standard for good citizenship or equal civil status, and the CLS do not believe that gay citizens should be denied basic liberal rights or equal protection under the law.
Brettschneider evidently endorses a thicker notion of FEC, although it is not completely clear what is involved in this conception. He seems to vacillate between thinner and thicker conceptions, as we saw above. His official statements make it seem fairly thin, but his discussions of particular cases often seem to thicken it. Nonetheless, it certainly seems to demand more congruence between explicitly political beliefs and groups' internal practices; hence barring discriminatory membership requirements. It also involves a broader understanding of citizenship rights, to include the wide array of rights endorsed by liberal egalitarians, such as gay marriage. Any individual or group that opposes those policies is liable to face democratic persuasion.
The ideal of FEC could be extended further than this, in several ways. For example, liberal egalitarians generally believe that those who respect citizens as free and equal must endorse an ideal of fair cooperation that guarantees some level of primary social goods for all and is incompatible with very large economic inequalities (see Rawls 2005: lvii, 450). Second, Rawlsian public reason liberals believe that acceptance of the ideal of public reason is entailed by endorsement of the idea of society as a fair system of cooperation between free and equal citizens, because the exercise of political power can respect citizens as free and equal only if it is justified by appeal to political values accepted by all reasonable citizens (Rawls 2005: 49-54; Quong 2011: 37-9). Third, one might argue that endorsement of equal citizenship is incompatible with civil society groups being hierarchically organised or selecting their leaders and officials using non-democratic means. 29 All of these elements could be included within an ideal of FEC. And the argument for each of them seems consistent with Brettschneider's approach of arguing based on what is required for citizens to enjoy equal public standing, making it unclear on what basis he can reject the extensions of the ideal to economic matters, views of legitimacy, or democratic associational requirements. His FEC is therefore a potentially very expansive ideal.
Indeed, there is likely to be pressure to extend the principles and values contained within our ideal of FEC, leading to ever-thicker conceptions. A notable feature of the debate about gay marriage is that a policy that was once highly controversial is now considered fundamental to FEC, such that the full force of democratic persuasion would be directed as those who continue to oppose it. Liberal egalitarian thought will continue to develop, with new policies seen as essential to FEC. Conservative groups that fail to adapt their religious and moral beliefs to conform to this new understanding will face the threat of being actively opposed by the state, having tax-exemptions withdrawn, and losing the ability to provide state-funded services.
Keck (2016) makes a related argument against hate speech legislation. Such legislation inevitably has unequal coverage, and unprotected groups will rightly feel aggrieved by this, 29 Levy (2015: 51-3) articulates-but does not endorse-this argument.
creating constant pressure to expand the scope of the legislation. For Keck, the result will be ever-expanding restrictions on free speech, in a way that ultimately undermines the legislation's liberal credentials. I am agnostic about whether Keck is right about hate speech legislation. My claim is that similar dynamics arise with respect to democratic persuasion. New claims concerning denials of FEC will be made, expanding our conception of the ideal over time, such that more groups are found to violate it.
To be clear, I am not arguing that such expanded notions of FEC are incorrect, or that we should not defend policies that extend our understanding of citizenship rights. My claim is certainly not that liberal egalitarian values, and our deepening understanding of their implications, are mistaken.
Instead, the upshot of my argument is that we must distinguish the full, thick, ideal of FEC and the much more minimal ideal that properly captures 'hateful viewpoints'. We should not automatically move from the claim that some principle or policy is part of the full ideal to the claim that the state ought actively to oppose and put pressure on those whose beliefs and values conflict with (or are implausibility compatible with) that principle or policy. Not every belief that conflicts with the thick notion of FEC is a 'hateful viewpoint', the expression of which constitutes hate speech. Brettschneider's transformative liberalism threatens to blur these distinctions. 30 Of course, we might think that our societies would be better if everyone embraced the full gamut of liberal egalitarian values and achieved full congruence between those values and their comprehensive doctrines, such that the internal practices of all associations displayed complete conformity to liberal egalitarianism. But we should not call upon the state to directly pressure individuals to achieve this level of congruence. Democratic persuasion should not be directed at every individual or group whose beliefs are in tension with a thick conception of FEC-including the conception adopted by Brettschneider. Many who reject some liberal egalitarian values or principles do so in good faith, without denying the moral or political equality of all citizens. They might well be mistaken, and we should certainly engage them in robust debate within civil society and defend our values. But these citizens endorse the fundamental core of liberal democracy, as embodied in minimal FEC, and the state therefore lacks legitimate authority to pressure such citizens to alter their views. Such authority is reserved for views that are truly 'beyond the pale', and hate speech should be restricted to this category. We should not use the state's power to create conformity with our full ideal of FEC.
This argument can also be made by appeal to the traditional liberal emphasis on the importance of protecting a vibrant civil society, consisting of 'mediating institutions' that occupy the space between individuals and the state. This involves accepting that some groups we do not like, or consider deeply mistaken or even positively harmful, will exist and enjoy the same public status-including tax exemptions-as groups that we favour. This is a price worth paying for the sake of the broader goal of securing a diverse and active associational sphere, with all of the benefits that this brings.
My arguments do not go so far as claiming that democratic persuasion is never appropriate. But its scope should be limited to groups that clearly and actively oppose the minimal 30 I do not mean to imply that our understanding of the minimal ideal of FEC is completely static or cannot itself include more elements over time. But the fact that this can occur is compatible with maintaining a distinction between the minimal and full ideals of FEC and with insisting that the former should be reserved for flagrant denials of basic liberal rights, and thus is a minimal category that does not expand as quickly or readily as the full ideal. I owe thanks to Matteo Bonotti for pressing me to clarify this. There is clearly much more to say about how we ought to make these distinctions, but I lack space to further develop these points here. conception of FEC. Such groups do hold 'hateful viewpoints' and engage in hate speech, and can be rightly opposed by the state.
Limiting democratic persuasion in this way also helps to evade some of the other problems I have discussed, since it requires fewer contestable judgments about whether beliefs and practices conflict with the liberal ideal. Clearly there will still be difficult cases, and some examination of theological beliefs and practices might be required. But the thinness of minimal FEC will reduce the number of cases under consideration, limit the extent of theological inquiry, and allow for clearer judgments to be reached.