A Republican Conception of Counterspeech

Abstract

‘Counterspeech’ is often presented as a way in which individual citizens can respond to harmful speech while avoiding the potentially coercive and freedom-damaging effects of formal speech restrictions. But counterspeech itself can also undermine freedom by contributing to forms of social punishment that manipulate a speaker’s choice set in uncontrolled ways. Specifically, and by adopting a republican perspective, this paper argues that certain kinds of counterspeech can dominate when they contribute to unchecked social norms that enable others to interfere arbitrarily with speakers. The presence of such domination can pose just as much a threat to freedom of speech as unchecked formal restrictions by threatening an individual’s discursive status, revealing a problem for those who defend counterspeech as a freedom-protecting alternative. Rather than rejecting both counterspeech and legislation outright, however, this paper argues that the republican principle of parsimony ought to be exercised when deciding on appropriate harmful speech response. While the principle of parsimony allows for suitably-checked formal punishment for some of the most egregious forms of harmful speech, citizen-led counterspeech must be guided by a reliable set of norms against the use of social punishment where those who do engage in social punishment face certain costs. The presence of robust, widely-known, and reliable norms thus supports both formal and informal responses to harmful speech while maintaining a secure discursive status for all.

1 Introduction

Consider the story of Palestine-born US-citizen Majdi Wadi (Mouk 2020). Since arriving in Minneapolis in the early 1990s, Wadi had made a great success in founding and running ‘Holy Land’, a large Mediterranean-themed restaurant, supermarket, and catering company that eventually employed around 200 workers. In June 2020, Wadi’s 24-year-old daughter was exposed on social media as having posted a series of racist and anti-Semitic tweets as a teenager.¹ The posts were shared widely across a range of platforms, drawing widespread negative attention and criticism, mostly aimed at Wadi himself. Protestors, both online and in-person, publicly rejected the racist messaging of the tweets and urged the public and other companies to boycott the business. The furore surrounding the exposure resulted in significant cost to Wadi’s livelihood (Fadel 2020). Around $5 million worth of business contracts were lost, the landlord in charge of the largest of Holy Land’s premises chose to evict the business, and Wadi eventually had to move his family out of their home after threats were made to their safety.

In such instances of counterspeech, understood here as expressive challenges to harmful utterances, the aim of exposure,² which is often accompanied by criticism, is to amass public rejection of a harmful message and to challenge the authority of the speaker to enact a racist utterance (Radzik et al. 2020: 50). Counterspeakers in this case may have felt that they were exercising a duty to speak out and prevent the racist utterances from enacting norms that subordinate Black and Jewish members of society (Maitra 2012), to assure members of those groups that their equality was a matter of public importance (Lepoutre 2017: 864), and to deter would-be racist speakers from making similar utterances in future (He et al. 2021).

But the cumulative effect of the actions of individual counterspeakers in this case is illustrative of the way in which counterspeech can also authorise destructive forms of social punishment, very often far beyond the grasp of individuals who, in exercising duties to ‘call out’ oppressive expression, have contributed to the online cacophony of disapproval (Tosi and Warmke 2016; Teekah 2015; Aly and Simpson 2019). The outsized influence of social media in today’s world means that the kind of experience suffered by Wadi is not uncommon³ (Thomason 2021; Klonick 2016).

What are we to make of such cases? For those who defend ‘more speech’ as a noncoercive, free speech-protecting alternative to extensive formal bans, everything appeared to have unfolded as expected: individual counterspeakers did not call upon the state (or the social media company) to impose costs upon Wadi or his family, and instead his daughter’s racist tweets were publicised and her authority to enact a racist norm challenged (Langton 2018a). The internet counterspeakers who called for the exposure, criticism, and boycott of Wadi’s business will thus have sent a strong message that racist speech is not acceptable and will likely have deterred (at least some) would-be hate-speakers who came across the story (Miškolci et al. 2018).

While the counterspeech in this case could be viewed as a success from an efficacy standpoint, there are good reasons to be concerned about such forms of counterspeech that- either intentionally or inadvertently- also authorise social punishment in this way. Counterspeech theorists themselves have taken seriously the potential risks posed by counterspeech practices, constructing their approaches in ways that minimise burdens on victims (Howard 2021) and that shield against the oppressive harms of amplification (McGowan 2018) and backlash (Saul 2021). In parallel, a burgeoning literature on the ethics of public shaming has urged caution when engaging in online discourse, due to the potential effects on targets’ social reputation, material circumstances, freedom of speech, and mental health (Frye 2022a; Billingham and Parr 2019; Elford 2021; Fritz 2021).

This paper, which argues from a republican standpoint, maintains that we should also take seriously the ways in which counterspeech can contribute to domination. Just as arbitrarily enacted and applied formal regulations expose individuals to unchecked interference, counterspeech practices can, at the same time as they challenge harmful expression, also risk the enactment of social norms that authorise significant levels of interference. In environments where such interference is especially costly and uncontrolled, the presence of such a risk manipulates a speaker’s choice set in a way which distorts the conditions necessary in order to enjoy freedom of speech effectively. So-called ‘free speech infrastructures’ of this kind will thus threaten the equal, non-dominated status of citizens by preventing them from taking part in the discursive contestation and construction of the shared rules and norms of society. Just as there are checks on state regulation of speech, then, there ought to also be checks on the social regulation of speech.

My argument will proceed as follows. In Sect. 2 I begin by critically assessing the claim that counterspeech is a reliably noncoercive, free speech-preserving substitute for formal speech regulations. Section 3 articulates and defends an alternative republican approach to harmful speech which provides a more substantive account of the wider formal and informal conditions required in order to enjoy a free, discursive status. One formal way in which one’s discursive status can be protected- hate speech law- is then critically examined (Sect. 4) and shown to be unreliable in many cases. Section 5 advances a republican account of counterspeech as an informal method of securing a non-dominated, discursive status using speech. Certain forms of counterspeech are shown to interfere just as arbitrarily with freedom of speech as formal regulation, thereby posing a threat of domination. In response, those concerned with threats to non-dominated discursive status from both formal and informal sources must exercise parsimony, which in the case of counterspeech requires securing robust norms against the use of social punishment.

Before starting, a brief preliminary note. In this paper, I take a broad view of the kind of speech that counterspeech may be used in response to. In that sense, I will not confine my area of concern to ‘hate’ speech only, understood as expression which denigrates or undermines the equal standing of members of society because of their identity (McGowan 2012; Langton 2012), but will also consider speech that is oppressive in nature (McGowan 2019: 106). One of the positives of counterspeech is that it can form part of a coordinated response to all sorts of harmful speech, including speech which contributes to damaging stereotypes or associations (Lemeire 2021), propaganda (Stanley 2015), ‘dog-whistles’ (Saul 2018a), and ‘microaggressions’ (Rini 2020), as well as other generally disparaging ways in which certain groups are discussed and addressed in the public sphere. With this goal in mind, then, I use the term ‘harmful’ speech to describe all those forms of speech that contribute in some way to hierarchies of status that rank individuals according to their perceived membership of a particular identity group.

2 Counterspeech as a Threat to Freedom?

The political theoretical literature on harmful speech is, for the most part, concerned with solving the following puzzle. That is, how should we respond to speech which undermines the equal dignity and perpetuates the oppression of some of society’s most vulnerable members, without in turn undermining one of the fundamental components of democracy: freedom of speech?

Defenders of counterspeech, including those who argue in favour of legal bans and state-led ‘expressive’ measures in limited cases, suggest that we need not sacrifice freedom of speech in order to protect those targeted by such harms. Instead⁴, we can look to the actions of individual citizens themselves, who, in exercising mechanisms carefully conceptualised by ongoing work in political theory and feminist philosophy of language, can intervene in (McGowan 2019: 106), challenge (Fumagalli 2021), silence (Saul 2018b), ‘block’ (Langton 2018b), or otherwise undermine harmful speech using particular context-sensitive discursive ‘moves’.⁵

For sceptics of speech regulation, counterspeech practices provide a solution to harmful speech without posing a threat to freedom of speech. On this view, state bans on speech which expresses an idea- whether hateful⁶ or not- cannot be enacted without at the same time limiting the conditions of democratic legitimacy and undermining the autonomy of individual citizens (Post 2011; Strossen 2012: 379). Democratic legitimacy requires that the state afford citizens a sizeable sphere of freedom from intervention, within which they can exercise their expressive freedoms among themselves, deliberate on matters of democratic importance, and negotiate the terms of political life without undue influence or threat from above (Meiklejohn 1961; Baker 2012: 63). This ‘democratic background’ essential to such an arrangement can only be achieved where all are able to exercise voice and make their thoughts and opinions known (Baker 1989: 59; Post 2005: 144-7). While harmful speech may very well contribute to ongoing discrimination and inequality, so this argument goes, it nonetheless must not be intervened with by the state.⁷ Once we exclude content-based speech interventions from our arsenal of possible solutions to harmful speech, then our only hope is that individual citizens take the initiative and confront such speech themselves (Howard 2021: 926). Here, and following Justice Brandeis’s 1927 concurring opinion in Whitney: “The fitting remedy for evil counsels is good ones… If there be time to expose through discussion the falsehood and fallacies, to avert the evil by the processes of education, the remedy to be applied is more speech, not enforced silence.”⁸

Not all are so optimistic about the efficacy of ‘more speech’ for securing freedom, however (Delgado and Stefancic 1996; Nielsen 2012). Harmful speech, as has been extensively argued, also negatively impacts the equal standing of targets in ways that prevent them from taking part in democracy effectively (Shiffrin 2011; MacKinnon 1993; Waldron 2012: 155). Quite apart from the emotional and psychological toll that such speech takes on victims (Delgado 1982; Matsuda 1989), the status-undermining effect of pervasive harmful speech will mean that it is likely that their attempts at confronting those who use such speech will be both burdensome and ineffective (Cohen 1993: 256; Delgado and Yun 1994). This will be especially the case, of course, for members of target groups whose social position automatically places them at a discursive disadvantage (Langton 2012).

The well-established negative impact of harmful speech for victims thus provides an obligation for those of us who are at minimal risk of harm to support the counterspeech of others by challenging such speech in our day-to-day lives (Goldberg 2018; McGowan 2018). Apart from those cases where counterspeech proves to be particularly ineffective (Cepollaro et al. 2022), most argue that we have no good reason to stand by and permit harmful utterances from continuing to shape and diminish the social standing of our fellow citizens (Ayala and Vasilyeva 2016; Langton 2018b: 161). As a set of practices that are believed to pose a minimal threat to freedom of speech, then, counterspeech continues to be an essential tool in our battle against pervasively harmful speech.

As the Wadi case and similar demonstrate, however, it appears that some instances of citizen-led counterspeech can impose costs that present just as much of a threat to freedom of speech as certain kinds of formal regulation. To understand how counterspeech might behave in this way, we need to explore how social punishment works as a method of norm enforcement. Social norms, understood as “clusters of normative attitudes plus knowledge of those attitudes” (Brennan et al. 2013: 15) are key determinants of human behaviour. The maintenance of norms is in large part dependent on the (actual or predicted) reactive responses of others, where

the spontaneously emerging attitudes of others towards us when we act in norm-governed contexts operate as a distinctive kind of intangible sanctioning mechanism, culminating in us externalizing and ultimately conforming with norms (ibid.: 225).

Public shaming, including the online variety (Billingham and Parr 2019; Frye 2022a), is an especially powerful method of social norm enforcement that ensures that those norms perpetuate over time (Anderson 2000; Jacquet 2015). By holding wrongdoers (i.e., norm-breakers) to account in public settings, shamers make wrongdoers and audience members (Radzik 2016; Detel 2013) aware of the presence of a morally authoritative norm while simultaneously expressing the moral commitments of the community⁹ (Billingham and Parr 2020).

In reality, however, norm enforcement of this kind will also give rise to harms that are very frequently coercive in nature (Aitchison & Meckled-Garcia 2021). According to Linda Radzik, social shaming inflicts harms on targets by playing on their “need for human contact and cooperation” and “desire for goodwill” from others (Radzik et al. 2020: 36). Shaming also impairs a target’s self-esteem by encouraging forms of social disapprobation that aim at the root of who they are as persons (Smith 2010). The infliction, or threat of infliction, of these forms of social punishment are coercive when they manipulate the choice set available to the shamed individual in future by rendering some of their actions extremely costly. These costs are amplified where the shaming also involves authorisation of ‘tangible’ costs on the shamed. Revealing a target’s personal information online (also known as ‘doxing’), or issuing demands to their employer that they be disciplined or sacked are common examples which involve the imposition of high costs.

Where such threats are commonplace and where the costs are high, it seems that counterspeech can, just as formal responses to harmful speech, also pose a substantial threat to our freedom of speech. One such concern has been raised by Elford (2021), who, in drawing a parallel between state and social imposition, describes how the Internet has exposed individuals to a greater level of reputational vulnerability than ever before. This threat of social punishment online, with its downstream ‘real-life’ consequences, limits freedom of speech by attaching substantial costs to one’s speech. Following Radzik’s account of social coercion, then, we might say that, where such discursive environments are commonplace, whether individuals are directly engaged with them or not, then the speech choice sets of some will be unduly limited in a way that diminishes freedom of speech.

So far, then, we have found that both formal and informal harmful speech response potentially threaten freedom of speech under certain conditions. But, given the debilitating effects of harmful speech on target groups, a ‘hands off’ approach also does not appear to be sufficient. What should be done? One way of avoiding the worst excesses of necessary norm enforcement practices is by abiding by a set of duties. For some, duties to avoid imposing foreseeable risks of coercion on others (Elford 2021: 163) and to engage with the moral agency of speakers (ibid.: 168) preclude us from using methods that enact and authorise social punishment. Others, however, argue that duties of this sort can be compatible with social punishment so long as certain conditions are observed (Billingham and Parr 2020). Across the range of duty-focused measures is a stress on engaging appropriately with the discursive agency of wrongful speakers, where such engagement is considered both respectful and necessary for successfully persuading them out of the beliefs that motivated their utterance in the first place.

The idea that we might have duties of this kind seems plausible. However, it is not so clear that a call that counterspeakers exercise certain duties is sufficient to reliably secure freedom of speech. Absent a set of checking mechanisms, it is not clear that such duties will be consistently exercised, especially by those who are already disposed to supporting harmful counterspeech practices. In such environments, then, individual speakers will live under threat of potential coercion and so will not enjoy free speech in the proper sense. Risks of interference will be amplified for those from marginalised groups. In figuring out what a reliable freedom of speech might require, then, what is needed is an account of how discursive environments more broadly secure or undermine the freedom of speech of individuals. This aim, I suggest, can be achieved by adopting an alternative republican framing of harmful speech response which takes seriously the domination posed by particular discursive arrangements. Here, a ‘free speech infrastructure’- which takes free speech not as pre-institutional but as dependent upon well-designed and controlled formal laws, rules, and social norms of society- must be of a sufficiently freedom-promoting quality in order for individuals to enjoy a robust discursive status.

3 Republicanism and the Free Speech Infrastructure

The republican tradition¹⁰ is at heart concerned with the freedom afforded to individuals when they are self-governing, where the constitutive political and social arrangements of society are decided ‘on the people’s terms’ (Pettit 2012) and subject to ongoing critical scrutiny from an active citizenry committed to exercising ‘eternal vigilance’ (Pettit 1997: 250). To ensure that the laws and norms of the land are decided upon in such a way, individual citizens must enjoy a status in which they are free in a non-dominated sense. Freedom as non-domination, as articulated by Pettit (2012: 58), arises not simply when an individual experiences a lack of interference from others, but when they enjoy a lack of arbitrary (or “uncontrolled”) interference. Along such lines, a republican will be concerned with the kinds of structural arrangements that need to be in place in order for an individual to enjoy a robust freedom as non-domination, such that they feel secure that they can go about their lives without having to submit to the unpredictable and arbitrary whims of others. Importantly, the republican ontological position on the necessary interdependency of human beings takes free status as something that depends upon, rather than is hindered by, a combination of the rule of law and strong social norms. As Skinner (2010: 97) describes, one who enjoys such a status as a ‘freeman’

is consequently someone to be reckoned with, someone who can look you in the eye, who can reason and negotiate on terms of equality without ever feeling the need to doff the cap or bend the knee. As a result, freemen are said to enjoy the respect- and the self-respect- that comes of being known to speak frankly and behave without fear or favour, acting solely as reason and conscience dictate.

The robustness of my status as a free person, importantly, must be such that everyone is aware of my status, including myself (Montesquieu 1977: 202). In that sense, individuals “should have an undominated status both in the objective and the subjective or inter-subjective sense of status” (Pettit 2012: 83).

The subjective component of non-domination, satisfied where the laws and social norms of society are such that they serve as a reminder that one’s standing is secure, is important for the following reasons. For one, a robust awareness that the state will protect against arbitrary infringements on one’s freedom provides individuals with the confidence to engage in the democratic process, which in turn ensures democratic legitimacy (Rostbøll 2015). Put differently, when the knowledge that I am respected as an equal co-deliberator in the decision-making process is enshrined in the law, I come to view myself as someone whose voice will be taken seriously (Laborde 2008: 16). Further, when the norms that make up our social relationships and our civic interactions also reflect my standing as an equal, I can thereby form bonds of trust with my fellow citizens (Pettit 1997: 261), which not only satisfies key features of social interdependence but also allows me to exercise independence from others (Braithwaite and Pettit 1990: 67; Pettit 1997: 241). Individuals who enjoy non-domination, then, are free in both a political and social sense, in a society that respects their contributions as a co-creator of shared rules and norms and which allows them to enjoy social relations of a sufficiently high quality (Pettit 1997: 265; Laborde 2008: 17 − 8).

The presence of robust norms and formal rules for the enjoyment of non-domination also applies to the enjoyment of freedom of speech. In carving out a characteristically republican understanding of free speech, Philip Pettit marks a distinction between two different ways in which our speech among a range of options can be considered ‘free’ (Pettit 2018: 61). On the first (typically liberal) account, which Pettit calls ‘unhindered’ speech, I am free in my speech options to the degree that I am free to choose from among a range of options (ibid.: 62). The second (typically republican) account of ‘protected’ speech requires not only that I am free to choose from among a range of speech options, but that I am securely protected in my choices (ibid.: 64).

Importantly for this form of protected speech is that this security of discursive status is something that I can rely upon in my communicative interactions with others. A protected discursive status thus requires that there are costly burdens in place that deter would-be interferers from preventing me from speaking. To enjoy a secure discursive status, such that I feel confident to live alongside others without fear of interference in my speech, it is not enough that these costs reduce the likelihood that I will be interfered with. Rather, free speech on this account requires a robustly reliable set of formal and informal norms to be in place. The costs attached to norm-breakers will need to be

severe enough to enable you to say your bit on any topic- and, figuratively, to look others in the eye as you do so- without the power of interference of others giving you good reason, by local criteria, for fear or deference (ibid.: 65).

The benefits of protected speech over unhindered speech are clear. Under unhindered speech, I am free to speak so long as others, no matter how much power they have over me, decide not to interfere with me. For example, I may work for an employer who has the power to discipline workers for complaining about working conditions with colleagues. So long as I am in my employer’s good favour I would be free in an unhindered sense. In contrast, to be free in a protected sense would require that there are regulations and norms in place that prevent employers from disciplining workers for their grumblings. We can see that in the latter case, I enjoy a discursive status not afforded to me in the former. Where my free speech is protected in this way I do not have to continuously monitor my speech among colleagues for fear of future punishment, nor do I need to ingratiate myself to my employer by flattering them or bending to their whims. Importantly, now that I am free to discuss working conditions with colleagues, we will have the capacity to check and monitor the power of the employer. Protected speech offers us a richer form of free speech than unhindered speech alone. As Pettit notes, the distinction reveals how freedom of speech is not to be understood pre-institutionally, but rather

[i]t is only by dint of law and regulation- and supportive social norms- that speech gets to be protected, and gets to count as free (ibid.: 67).

The requirement for a robust set of suitably-controlled laws and norms as a prerequisite for freedom of speech has notable implications for how we approach our responses to harmful speech. Taken together, the formal and informal aspects of a free speech infrastructure need to be such that one’s discursive status is suitably protected. This framing thus demands a rather more substantive set of conditions for the enjoyment of freedom of speech in the face of harmful speech than those described in Sect. 2. Rather than working out how to minimise interference in the realm of speech, then, we must consider what both the formal and informal of speech conditions ought to look like in order for individuals to enjoy freedom of speech effectively, whether one is a potential victim of harmful speech or an ordinary speaker. In the following, and by employing the above republican framework, I first critically examine a common formal response in the form of hate speech law for its suitability for protecting discursive status. Laws of this type, I argue, struggle to satisfy the demands of non-arbitrariness, thus making them unreliable for securing non-domination except in a very narrow set of cases.

4 Formal Responses to Harmful Speech and the Problem of Arbitrariness

The republican understanding of freedom as conditional means that, when approaching the question of speech regulation, republicans will be concerned not with the demonstration of ‘harm’ but with determining whether the at-issue speech (whether libel, incitement, or hate speech) must be limited in order to secure freedom as non-domination (Pettit 1997: 36). Importantly, the proposed regulation in question must also satisfy the conditions of being both ‘co-exercisable’ and ‘co-enjoyable’, so as to avoid state domination towards a minority of citizens.¹¹ A free speech infrastructure that provides protection to some groups and not others in ways which increase domination overall will thus not be one in which all enjoy the independent status required to take part in public life effectively.

There are three key ways in which hate speech laws potentially increase, rather than reduce, the domination of citizens. Firstly, it is not clear that many hate speech laws- in whatever form- can be written and applied in a way which satisfies a republican commitment to non-arbitrariness.¹² One level of arbitrariness arises in the way such laws describe the kind of speech they deem to be punishable. Take, for example, one of the most common categories of hate speech ban and that which is defended by Waldron (2012) as offering minimal threat to free speech: group defamation law. Such laws can be criticised for an inherent vagueness, which makes them difficult to implement in fair and balanced ways. Article 20 of the International Covenant on Civil and Political Rights (ICCPR), for instance, outlaws speech involving “[a]ny advocacy of national, racial or religious hatred that constitutes incitement to discrimination, hostility or violence.” A similar version can be found in Sect. 18(1) of the United Kingdom Public Order Act 1986, which targets those insulting words where the speaker “intends… to stir up racial hatred.” In Canada, the Supreme Court has defined hate speech as that which “is likely to expose” individuals to hatred, defined as “unusually strong and deep-felt emotions of detestation, calumny and vilification” and “enmity and extreme ill-will” which “goes beyond mere disdain or dislike.”

As commentators have pointed out, however, the broad and overly-vague wording of such laws makes it extremely difficult for the courts to apply them consistently (Heinze 2006). From a republican perspective, the greater the amount of discretion at the hands of those involved in the criminal justice system- in this case, the jury- the more prone a particular law or legal practice is to produce domination. As Strossen (2018: 76) argues in her reflections on the Canadian example:

If you were a juror, would you be able to distinguish between speech that conveys “disdain,” which is not punishable, and speech that conveys “detestation” or “vilification,” which is? And if you were voicing your own strongly negative views about some person or group, how secure would you feel that officials would consider your words to communicated protected “disdain” rather than punishable “enmity” or “extreme ill-will”?”

Such uncertainty is intensified by the inherent heterogeneity of contemporary liberal democracies. Police, prosecution, and jury identification of ‘vilification’ and ‘disdain’ will depend strongly on how the individuals involved are socially-situated, meaning that implementing such laws evenly, even while taking great care to exercise caution, will be a considerable challenge in plural societies. Differences in culture, religious belief, class, gender, and race all have been shown to strongly influence public perceptions of the ‘harm’ in speech, thereby limiting the possibility of satisfying a standard ‘reasonable person’ test (Cox 2016; Moran 2003). Existing power inequalities in society between the majority and those with minority values thus “ultimately implements a majoritarian consensus that is prejudicial towards minorities or, more accurately, those holding minority views” (Kohl 2018: 114).

A second kind of arbitrariness can be found in how hate speech law necessarily distinguishes between those groups which fall under their protection and those which do not. The problem, according to Eric Heinze, is that such formal distinction between protected and unprotected groups within the law will “inevitably create two tiers of citizens- those who are protected from offensive speech, and those left unprotected from equally offensive speech” (2006: 555). To illustrate, Heinze points to common forms of prejudice which, while not falling under formal definitions of hate speech, are nonetheless just as damaging to their victims, pointing out that

many physical, mental or psychological conditions that provoke hurtful speech, such as low intelligence, ugliness, frailty, mental illness, physical debility or malformation, poor coordination, blindness or deafness may be neither self-induced nor easy to change, and have histories of stigmatism stretching further back than any current form of racism (ibid.: 566-7).

It could reasonably be argued that, while the above groups of people will suffer the harms of ‘offensive’ speech, such harms do not contribute to the systemic forms of discrimination experienced by those groups typically protected by hate speech laws. This argument can be supported with reference to the so-called ‘symbolic’ or ‘expressive’ argument for formal bans (Brown 2015: 240). According to this view, such laws can be justified not by the frequency of incidents but by their expressive or symbolic role in reminding systematically subordinated groups of their equal standing. In republican terms, both the existence of the law and its enforcement are believed to provide that intersubjective reminder of robust status necessary for freedom as non-domination¹³. By enacting a law designed to protect a specific group, then, the state is not expressing the idea that the group are indeed morally superior to other members of society (Galeotti 2002: 156). Rather, they are responding to the fact that there exists a background social hierarchy in which that minority group are already placed in a diminutive social position. By enacting a law of this type for symbolic purposes, then, the state could be said to be attempting to shift social understandings of that group such that they and their fellow citizens can ‘walk tall and look others in the eye’ in the way republicans describe (Pettit 2012: 3).

In practice, however, the demonstration of systemic discrimination has not taken such a central role in the decision over which groups to protect and nor, when it is invoked, does it tend to be very effective at securing equal standing in the way defenders of the symbolic argument claim (Lepoutre 2020)¹⁴. The UK Public Order Act (1986), for example, has been guided in their selection not by the relative social standing of groups but on predicted prosecution rates. It is because of this criterion for qualification that the Act has been criticised for excluding disability and gender-identity as protected characteristics (Brown 2017b: 300).

Further, the differing threshold evidential burdens required for prosecution across groups means that full protection of certain groups is not possible without posing a threat to freedom of speech. As it stands, a higher evidential burden is in place in cases involving religious hatred compared to those involving racist hatred. Together with an additional set of clauses, the higher standard required for cases of incitement to religious hatred aims to protect non-hateful criticism of religion and to reduce the risk of a ‘chilling effect’ (Barendt 2011) at the same time as it protects religious individuals from hate (Brown 2017b: 35). Arguably, however, some of the most common and influential anti-LGBT and anti-Muslim speech, for example, operate under the guise of ‘criticism’ (Mondon and Winter 2017; Kolhatkar 2014). That current hate speech law in the UK aims to protect non-hateful criticism is, of course, a positive from the perspective of free speech, but it still leaves us with a large proportion of harmful speech uncovered. If incitement laws do not cover some of the most common and consequential cases, then, their purpose comes under question (Clooney and Webb 2017).

These problems will pose a worry to republicans, not least because of the high risks already attached to state intervention in speech in the first place. Were hate speech laws proven to be ineffective at protecting the vulnerability of target groups, then their implicit imposition on citizen freedom can no longer be justified. Republicans recognise the chief importance of the law in securing citizen freedom (Pettit 1997: 36), but stress that the decision to attach state-imposed costs on behaviour can only be justified where the strengthening effect on dominion (i.e., non-domination) when regulation is in place far outweighs the significant weakening effects on dominion attached to regulation. The decision to invoke the criminal justice system as a response to harmful speech, then, must be guided by the principle of parsimony, on which Braithwaite and Pettit (1990: 79) state:

Every act of punishment has the certain cost of diminishing someone’s dominion and if we are concerned about dominion then every act of punishment will need positive justification. The rest-response will be mercy, the response that needs vindication punishment. More generally, the rest-response will be non-intervention, the response that requires justification will be intrusion.

What we can take from this is that the more difficult it is to distinguish between non-hateful public discourse and the incitement or stirring up of hatred (intended or foreseen), the less comfortable we should be with lowering the threshold for prosecution. Those groups that are not currently included under most hate speech legislation thus will tend to be those where the distinction between offensive speech and hate speech is more difficult to discern. To extend current incitement laws to include them would either require maintaining a high threshold on prosecution (and thereby to punish very few cases), or to unjustifiably infringe upon freedom of speech in a way which limits public conversation about important issues. In either case, the discursive status required for the enjoyment of non-domination remains unsecured. We must, then, look to other key aspects of our discursive environment for a response to harmful speech that is both reliable and freedom-protecting.

5 Counterspeech: A Republican Defence

Given republicanism’s commitment to a joint system of laws and norms as necessary features of the equal standing of all, it appears that we have good reasons to count counterspeech as one of the necessary practices of securing such a system.¹⁵ Counterspeech might thus be described as an exercise in enacting and expressing core civic values through the exercise of vigilance (Laborde 2008: 152). In the context of harmful speech, then, this not only means that a republican framing recognises that harmful speech is a widespread problem for freedom (Bonotti 2017), but that protecting the psychological, subjective (and intersubjective) aspect of equal standing requires that individual citizens take part in the contestation of that speech. In short, then, the apparent necessity of norms of citizen-led counterspeech in the maintenance of a free and equal society means that counterspeech is not just a form of desirable behaviour but is one of the central components of our free speech infrastructure.

As discussed in Sect. 2, however, counterspeech that uses forms of social punishment can also manipulate a speaker’s choice set in uncontrolled or arbitrary ways. In most cases, no individual contribution to collective counterspeech on its own will threaten freedom of speech. Instead, online ‘pile-on’ scenarios will involve a range of uncoordinated individuals, described by Frank Lovett as a ‘team’ (2022: 32), who are drawn together temporarily by a “contingent convergence of purposes” (ibid.) to engage in the joint intentional action of norm enforcement. Importantly, the emergence of the team, and their success at manipulating a speaker’s choice set, depends upon certain background conditions supported by social norms. These conditions provide the team with the capacity to exercise power over a speaker and distinguishes ‘latent’ teams that pose a domination threat from all of the other potential teams that might have the power to interfere (ibid.: 45).

Certain structural features of social media platforms make it possible for a set of previously-uncoordinated counterspeakers to contribute to a joint intentional action which aims towards holding a speaker to account for using harmful speech (Saul 2021; Coe et al. 2014). These features- which includes ‘sharing’, ‘liking’, and ‘re-posting’ (Marsili 2021)- allow individual contributions to form a collective mass that can have major influence over the choice sets of others. For this form of collective counterspeech to dominate, however, we need to determine that the collective: (i) Has the power to interfere; and (ii) Has the capacity to interfere. The kinds of interference involved include those I described in Sect. 2, such as the social cost of public shaming, shunning, and stigmatisation as well as the authorisation of more ‘tangible’ material costs of loss of employment and physical threat to safety. The collective of counterspeakers in this case will indeed have the power to impose or authorise those social forms of punishment, the cost of which likely will be determined by the size and social standing of the punishing audience.

But does a newly-coordinated team of counterspeakers have the capacity to interfere with speakers? Here, we need to look to the background conditions in place at the time of the coordination. Social media, in the last several years, has provided unprecedented opportunities for large numbers of people to congregate and communicate from all over the world (Papacharissi 2009). Twitter, for example, continues to be highly influential when it comes to influencing public debate in certain sections of society (boyd 2011). The opportunity to connect with a wide audience, in addition to the ongoing public influence of the platform, suggests that public shaming events that take place on social media have the potential to have far-reaching consequences. We also have on social media a communicative culture in which ‘calling out’ norm-breakers is common (Bouvier and Machin 2021; Bouvier 2020). The option for anonymity of users, plus the relative low-stakes nature of engaging in online environments in comparison to those in-person, makes contributing to online forms of punishment typically less costly than in-person forms of norm enforcement (Anderson et al. 2014; Santana 2014). This, we could say, has led in some part to the emergence of a norm that encourages users not to stay silent when they witness harmful speech, and to take part in counterspeech against it (Saul 2021:140-3).¹⁶

For the power and capacity conditions for domination to both be satisfied, then, counterspeech needs to take place in an environment in which the norms in place authorise, impose, and encourage harmful forms of social punishment for speaking (Frye 2022b). Unfortunately, the Internet is one such place where norms of this kind are commonplace (Basak et al. 2016; Stroud 2016; Goldman 2015). Where such norms are in place, counterspeakers will not censure one another for engaging in harmfully coercive counterspeech practices and may even join in themselves (Gervais 2015; Neubaum 2018). When audience members perceive that large numbers of people support these counterspeech practices then such methods come to be authoritative and therefore seen as appropriate and proportionate. Where such norms become widespread, employers who perceive a consensus regarding the actions of a particular speaker may feel that they can discipline or fire them without consequence (Broderick and Grinberg 2013).

How, then, might the republican framework of free speech approach informal methods of counterspeech of this kind? The central role that social norms play in the maintenance of a free and equal republic means that republicans will be comfortable with at least some sort of social consequence for norm-breakers (Pettit 1997: 254). In their work on the ‘economy of esteem’, for instance, Brennan and Pettit (2004) build on a centuries-long tradition in political theory on the central role that the ‘intangible hand’ of social approval and opprobrium can play in securing ‘good’ social norms, especially where the use of the law is either ineffective or inappropriate. Here, a suitably-harnessed desire for esteem offers an underutilised force with which we can affect human behaviour for the good of society.

There are two key limits to what the economy of esteem can do for counterspeech without worsening overall freedom. The first, which can be thought of as an individualised danger, is seen in those cases where public opprobrium results in a permanent or severe loss of reputation, such as when an individual becomes a source of stigma (ibid.: 318).¹⁷ Secondly, and on a societal level, a culture in which systems of esteem and disesteem lose their sensitivity to behavioural changes will also no longer be capable of influencing behaviour in socially-desirable ways (Braithwaite 1989). Republicans will thus be wary of those methods that hinder the cause of securing norms that support non-domination, either by engaging in practices that authorise the imposition of severe costs on speaking or by distorting the economy of esteem such that it ceases to influence behaviour in a positive direction, for

if the infamous are stuck with their notoriety, and the famous with their celebrity, then they are not going to be subject in the ordinary way to the discipline of an intangible hand. There will be little prospect for the infamous of redeeming their reputation, and little prospect for the famous- at least if they play it safe- of losing their reputation. And so neither will be susceptible in the manner of ordinary folk to the desire for esteem or the aversion to disesteem (emphasis added) (Brennan and Pettit 2004: 318).

Putting the practical effectiveness of forms of social punishment to one side, severe or unpredictable social punishment in the realm of speech will also negatively impact the robustness of a citizen’s discursive status. As with vague and overly-broad speech laws, a discursive environment in which counterspeech is guided by norms endorsing coercive forms of social punishment will also be one in which one’s discursive status is insecure (at the very least) or potentially non-existent (at the very worst). Such an environment would have serious downstream effects for both individuals and for society as a whole by undermining citizens’ subjective reminder of discursive status and by deterring them from speaking on matters of social or democratic importance. This will, of course, have an even more worrying impact on members of historically silenced or marginalised groups, who already experience higher levels of scrutiny and enjoy a relatively low degree of authority when engaging in public discourse.

If social punishment, especially in today’s online world and as it pertains to forms of counterspeech, is as much of a threat to freedom as I have suggested, does this mean that counterspeakers must simply avoid social punishment completely? Building on the discussion of the principle of parsimony (Sect. 2), it seems that we must also invoke parsimony when deciding how to best to use counterspeech when responding to instances of harmful speech. As with the constraints imposed on the criminal law by parsimony, counterspeech measures must also be guided by norms which reduce the likelihood of posing a greater threat to dominion than the proposed cure. Where the conditions required for efficacious norm change via social sanction is shown to be absent, then there should be a set of robust and reliable norms in place against such sanction.

What should these norms look like, and how can we ensure consistent compliance? While I do not have space to fully explicate a positive account here, I will briefly outline a few thoughts on how online environments in particular might foster robust norms which aim to protect discursive status while also tackling the effects of harmful speech. In online environments, it is necessary to secure norms against methods of counterspeech which authorise the imposition of substantial costs on speakers. There should thus be norms in place against the use of harmful social punishment and against the imposition of tangible harms on speakers, such as contacting a wrongful speaker’s place of employment or revealing their personal details online (Robards and Graf 2022; Douglas 2016; Han and Brazeal 2015). Importantly, such norms should be suitably costly, such that potential punishers are deterred from following through with their punishment (Brennan et al. 2013: 38).¹⁸

To ensure that norms of freedom-protecting counterspeech are reliable and that they continue over time, however, we must overcome two fundamental hurdles specific to the online realm. Both concern the difficulty of holding wrongful punishers accountable given the structure of online discursive environments. The first, is that ‘teams’ of online counterspeakers are large, unstable, and made up of agents who each hold varying degrees of authority (Ahrne and Brunsson 2011). A second accountability-limiting feature is that counterspeakers online have the option of engaging anonymously, making them less sensitive to disapproval. On the first issue, it might be possible to stabilise counterspeech behaviours by encouraging the growth of widely-recognised practices across online groups. Once those practices have become established, and guided by their own set of norms, then those who engage in counterspeech will be moved by social pressure to follow those practices over others. One example of such norm stabilisation can be seen in the online counterspeech movement #iamhere. The movement, which originated on Swedish Facebook in 2016, practices a form of organised counterspeech in which members scan online articles and comments sections for harmful speech before, by use of the hashtag, signalling to other group members to join in to counter or correct the speech. The movement, which now comprises over 150,000 members worldwide, has been shown to be extremely effective at combating hate by mobilising a collective while maintaining an ethos of respectful and open dialogue (Friess et al. 2020; Buerger 2021). Importantly, the norms of the movement require that counterspeakers do not harm others while engaging in counterspeech. The continued and well-established practices used by #iamhere reliably prevent those involved from using harmful methods by maintaining a group identity¹⁹ guided by values of respect (Sunstein 1996/1997). Long-term and large-scale use of standardised counterspeech practices online can secure robust norms which can be relied upon in ways that protect discursive status while undermining the effects of harmful speech.

The second issue of anonymity can also be dealt with by establishing the form of collective group identity found in the above #iamhere example. Here, it is not necessary that individuals engaged in counterspeech use their real identity when taking part in counterspeech. Instead, the creation of an activist group identity can go a long way towards securing compliance with a set of norms of behaviour while counterspeakers carry out their work (McCarthy & Zald 2001). The creation of a large-scale common (yet flexible) counterspeaker identity, with its own set of norms, allows others to observe and imitate ‘good’ practice which, over time, solidifies those norms as the commonly-accepted standard (Gerbaudo and Treré 2015). Public figures who hold substantial sway over the attitudes of the community can also have a great deal of influence over whether or not the group, and its associated norms, become established over time (Chen and Liebler 2022).

6 Conclusion

Can we rely on counterspeech to tackle the harmful effects of hateful and oppressive speech without posing the same threats to free speech typically associated with formal regulation? As this article has demonstrated, we must challenge the assumption that counterspeech is a reliably freedom-protecting alternative to formal bans. The growing influence of social media allows previously-uncoordinated individual counterspeakers to impose a risk of significant costs on others. In this way, and when those costs are uncontrolled, both formal and informal forms of harmful speech response can contribute to the unfreedom of citizens by increasing domination.

In Sect. 2, I explored, and ultimately rejected, a common assumption that freedom of speech and formal regulation are necessarily in tension with one another. Such a framing suggests that formal responses to harmful speech are, despite their potential benefits, ultimately freedom-limiting, whereas counterspeech is freedom-protecting. Where social responses do cause harm, it is suggested that we rely on citizen duties to refrain from using counterspeech in harmful ways. Sceptical that such duties can reliably protect freedom, and by adopting an alternative republican framing (Sect. 3), I argue instead that the discursive status necessary for the enjoyment of freedom of speech is not pre-institutional but is instead conditional on certain formal and informal conditions. Formal bans are critically assessed for their suitability in fulfilling this task (Sect. 4) and shown to potentially undermine freedom when arbitrarily enacted and applied. Counterspeech as a freedom-promoting method of responding to harmful speech is discussed in Sect. 5 and is also shown to impose risks to freedom, especially in certain discursive environments. When approaching potential harmful speech response it is thus necessary to be guided by the principle of parsimony, which requires taking account of the conditions (i.e., formal and informal norms) provided by the entire free speech infrastructure and which provide the foundation of a secure discursive status for all.