On Hate Speech & Free Speech
Contents
1 Introductory Remarks
2 Mill and Freedom of Speech
3 Hate Speech
4 Free Speech
5 Arendt and Epistemic Arguments against Free Speech
6 Social Media
1 Introductory Remarks
With the recent assassination of Charlie Kirk, a phenomenon, strange even for American standards, was evident. The MAGA camp began invoking the rhetoric of ‘hate speech;’ Pam Bondi said,
There’s free speech, then there’s hate speech. And there is no place [...] for that in our society.
This raises to salience a topic of political, and legal philosophy that dominated the online discourse in the 2010s; yet in our current climate, the roles of the respective political sides have swapped. Before, the caricature ‘liberal-feminist’ who had such authority to wield the mythical sword of ‘cancel culture’ were met with the ‘free speech absolutists’ who defended speech regardless of their content or their context. Whereas now, it is the executive office echoing the words of their avowed ‘woke’ enemies. Ironically utilizing the terminologies and rhetoric frequented by the supposed crazed ANTIFA organizations. Hence, we believe that it is apt to investigate the debates surrounding hate-speech and its relations to free speech.
This prompt, therefore, outlines prominent arguments for and against hate speech and free speech respectively. Thereby coming to question whether these two principles, if held, can co-exist, or whether they are either (i) theoretically exclusive, or (ii) politically and/or practically exclusive. Section 1 will cover the classical arguments provided by Mill in defense of free speech outlined in On Liberty. Section 2 explores what is meant by ‘hate speech’ outlining Katharine Gelber’s definition that hate speech occurs within systematic structures. Section 3 involves the dialectic between Jeremy Waldron and C. Edwin Baker, with the former outlining the harmful effects of hate speech on the fabric of society; the latter providing a novel argument for freedom of speech that basis itself on democratic, and liberal foundations. Section 4 explores Arendt’s Truth and Politics, where I shall question Mill’s epistemic justification for freedom of speech and give arguments on epistemic grounds for censorship.
2 Mill and Freedom of Speech
Mill’s On Liberty sets out a systematic political philosophy with one core principle, which to use Isiah Berlin’s terminology, consists of negative liberty. Mill writes,
The object of this Essay is to assert one very simple principle [...] That principle is, that the sole end for which mankind are warranted, individually or collectively, in interfering with the liberty of action of any of their number, is self-protection.
Included within this principle is the negative liberty to express any form of speech without government interference. Mill, however, gives an epistemic justifications independent of the said principles. Which are, summarized by Brink, as such:
A censored opinion might be true.
A censored opinion might contain partial truths.
Even if wholly false, wrong opinions provide opportunities to reinforce the strength of true claims.
If unchallenged, even true beliefs become dogmatic, and will thus lose meaning.
As Brink puts it, 1) and 2) are truth-tracking justifications. Freedom speech are instrumentally valuable in guarding against the loss of true claims by censorship. However, taken by themselves, they do provide strong reasons against censorship. This is because i) we can have censorship that is successful in truth-tracking, ii) Mill does not necessarily justify why truth in itself is valuable (intrinsically, or extrinsically), iii) it needn’t be that free expression may maximize the ratio the number of true claims as opposed to false ones. The final worry seem the most pertinent to our current situation. With the maximilisation of mass media through technology, the so-called ‘market place of ideas’ are thoroughly polluted with manipulative language.
Hence, 3) and 4) are needed for an exhaustive justification. Brink dubs the latter two as the deliberative reasons. Mill’s claim is that true claims, if unchallenged become dogmatic and thus become stale and loses their meaning. This results fro Mill’s distinction between knowledge and true beliefs (with the implicit belief that knowledge is true belief with justification). As justification requires a form a deliberation, a choice among alternatives, censoring false beliefs will limit this capacity of deliberation. With the threat outlined here being that we risk becoming harboureres of true beliefs, rather than knowledge.
My general concerns will only be briefly touched here. But the most salient issue that is evident in Mill’s justification for free speech is the fact that it is ultimately an epistemic justification. From this, two questions appear the most important here for the purposes of our debate:
i) is free speech the best method of maximising the number of knowledges in our society?
ii) is Mill’s epistemic justification for free speech at all relevant to discourses around free speech and hate speech?
The answers to i) and ii) seem far from obvious from my point of view. If they are not, it seems necessary to search for new arguments against any forms of censorship.
3 Hate Speech
In Differentiating Hate Speech: a Systemic Discrimination Approach, Gelber rejects the harm definition of hate speech. The argument that hate speech is that which constitutes harm to the audience. This definition spawns ambiguity and a lack of clarity as to what harm constitutes, and bypasses one core property of hate speech – that it functions within structural oppression. Gelber argues that the sole emphasis on the harm that hate speech produces is not a sufficient understanding of hate speech. Hate speech operates within oppressive structures and one of its core functions is the reinforcement of the marginalization of othered groups. If focused just on harm, then hate speech may come to equivocate discriminatory speech against a white person and such speech against marginalized and oppressed groups. This, to me, seems right. Placing the locus of the effect produced by hate speech beyond strictly just the hearer and to have an overall impact on society is a more accurate and coherent approach than an emphasis on harm.
Thus, Gelber proposes understanding hate speech as speech that reinforces systemic discrimination. Gelber appeals to Mary Kate McGowan’s rule accommodation model to help define systemic discrimination. By engaging in hate speech, the rules of accommodation is updated in such a way that ‘enact facts about what is subsequently permissible in the conversation’. However, hate speech cannot be directed towards people who are not historically marginalised. For example, discriminatory speech against white people does not qualify as hate speech as it does not add to the systemic discrimination they face. They cannot be marginalised systemically through speech acts in the current structure. Thus, for Gelber, hate speech operates similarly to McGowan’s accommodation thesis except that it is expanded beyond language and placed within systemic discrimination. It goes beyond language as what is made permissible is the reinforcement of oppression itself, not just oppressive language.
Therefore, hate speech is one that is ‘publicly directed at a member of a group that is identifiable as being subjected to systemic discrimination in the context within which the speech occurs’. Systemic discrimination is ‘best understood as pervasive, institutionalized exclusion presenting in patterns and practices that are perpetuated in, and through, ostensibly neutral institution principles’. Gelber’s systemic discrimination approach to hate speech moves beyond the previously orthodox harm definitions and tries to incorporate structural issues and oppression into the definition.
4 Free Speech
The societal harm that hate speech incites is an orthodoxy in the literature around it. However, what to do about it is a major topic of contention. C. Edwin Baker was one of the most prominent defenders of free speech absolutism, arguing that no state intervention is speech is justified. His arguments are found in many writings, but I shall focus my review on his Hate Speech which appeared in The Context and Content of Hate speech (2012).
Whereas many writings on hate speech begins with the values of democracy, Baker bases the value of democracy on the value of individual liberty. He argues that what justifies our allegiance to democracy is its unique political structure that embodies a normative principle of equality among all of its members. It respects individual liberty and self-determination. Therefore, it is individual liberty itself that justifies democracy, and places democracy as an offshoot of the normative value of personal autonomy. Therefore, under this framework and understanding, any arguments against hate speech which basis itself on the values of democracy fails, as it is these basic liberties that affirms democracy in the first place.
Nevertheless, hate speech, Baker affirms, can lead to greater societal harm. Thus, he lays out five conditions that ought be met for restriction on hate speech to be justified. These are:
Hate speech occurs in cases of genocide.
Hate speech contributes to these matters causally.
Legal prohibitions of hate speech are actually an effective place to intervene in the casual chain.
Enactment of hate speech prohibitions creates greater benefits than costs.
These legal bars on hate speech would not reduce the efficacious news or the likelihood
of other intervention that would be more effective in preventing virulent racist acts.
Baker admits that (1) is uncontroversially true. Genocides in history are littered with dehumanizing language hurled at targeted minority groups. Further, Baker notes that (2) is ‘plausible’, though hate speech may often not be the spawn of hate itself. What Baker critiques the most are (3), (4), and (5). He argues that legal interventions in restricting speech are likely to be ineffective in putting a halt to hate and have dangers of intensifying those with these bigoted views. He outlines six ways in which hate speech could backfire.
(a) By legally restricting hate speech, we will force it to go ‘underground.’ This would protect hate speech from valid criticisms and into echo chambers. More speech against these views will be more effective.
(b) By causing racism to go underground, prohibitions are likely to obscure the extent of the problem and the location or the human or social carriers of the problem. Further, it may make us ignorant of valid ways of refuting forms of hate speech.
(c) Legislation may increase the racist individuals and group’s sense of oppression thereby increasing their rage and their desperation for action.
(d) Prohibiting expression of any values are likely to reduce democratic cultural self- understanding that conflicts are better dealt with politically than through restrictive legislation.
(e) These legislations runs the danger of diverting political energy from more meaningful political responses.
(f) There is the danger of a ‘slippery slope’ that the state may extend ‘hate speech’ legislation to crack down on dissidents.
John Stuart Mill’s line of arguments are clear to see in (a) and (b). Ignorance of the arguments of hate speech is argumentatively forceful as by not interacting with hate speech, we may be ignorant of defeating them via arguments. Further, (a) in conjunction with (c) may come to further reinforce the racist’s certainty of their view. By restricting their speech, we are restricting our ability to respond to it, and with the racists’ increase in resentment and rage, their attitude towards their belief may strengthen – essentially, we would be brushing the issue of racism and hatred under the carpet, rather than attempting to eliminate them, which is the argument of (e). Finally, though quite a common argument (f) is a genuine concern. Already, the famous case of Mahmoud Khalid’s arrest and threat of deportation due to his pro-Palestinian activities in Columbia serves as an example of a government willing to curb an individual’s right to speech. Giving the state the legislative authority of hate speech may embolden and further strengthen the ability to get rid of dissidents.
I consider (f) to be the greatest worry posed by Baker, with the others being outweighed by the benefits of hate speech legislation and with (e), I argue not being the case. A refutation of Baker’s line of arguments is given succinctly by Jeremy Waldron in his The Harm in Hate Speech, where in chapter 6 he directly challenges arguments forwarded by Baker. In response to the suggestion in (a) that more speech may be a better solution, the so-called ‘market place of ideas’, Waldron argues that the analogy between markets and speech regulation fails. The argument goes that an economic model that prioritises efficiency cannot be placed on a public discourse that concerns itself with truth as the fundamental value. Waldron writes ‘they may produce efficiency, they may not produce, or they may undermine, distribute justice’ (156). Waldron challenges the assumption that ‘more speech’ would lead to the prevailing of truth. As evidenced by current discourse especially in America, Waldron seems to be right here.
Waldron provides further exposition of Baker’s normative valuation of individual autonomy, especially regarding speech. He writes, for Baker, ‘individual speech reveals the way in which ideas are connected to persons: it presents persons as the locus of ideas’, he further elaborates ‘if we enact laws against racist violence or arson, we deprive him of at least one mode of self- disclosure [...]. If, however, we enact laws against hate speech, we deny the racist his elementary autonomy of self-disclosure’ (164). Thus, for Baker, speech warrants special right to absolute freedom as it is the core mode of self-disclosure and thus it is the centre of connecting ideas, and basis of democratic expressions.
Waldron argues that though these arguments do hold, they do not undermine the genuine threat and harm that hate speech poses to an ordered society. Especially, it threatens the mutual assurance that we provide to each other, and specifically minority groups that they would not be the target of injustice and inequality. This destructive function of hate speech, the erosion of mutual assurance of dignity and equality, ought to be taken further. By invoking Rae Langton’s illocutionary silencing, one can easily make the argument that the erosion of this assurance and the heightening of racially motivated hate and prejudice, as a result, may come to silence minority groups. Thus, the fundamental core value of individual right to self-disclosure would be undermined if hate speech against the targeted group is pervasive and prevalent. The argument, therefore, is that the instantiation of individuality that is provided by democracy is itself undercut by the hate speech. The mutual assurance that we afford one another in society is degraded such that an individual’s self-disclosure, their expression itself, is silenced. The argument being that hate speech illocutionarily, and practically silences individuals based on their identity.
Further, another prong of Waldron’s argument come to directly undermine the five of the six concerns raised by Baker. Namely, that pervasive hate speech leads to emboldening of other like- minded hatred and may build up prejudice and racism as a result. Hate speech as a result reinforces racist views and, as argued by McGowan and Gelber, shifts the permissibility in the accommodation game to allow for more and greater forms of racism. Contra Baker, Waldron seems to correct here. Ever since Elon Musk’s take over of Twitter and the following removal of hate speech banning on the platform, racism, slurs, misogyny and other forms of hate speech rose drastically. Thus, in this example, the removal of censors and more speech led to greater numbers of hate speech and bigotry – affirming Waldron’s view.
Though I align myself more with Waldron’s view concerning hate speech, Baker’s final worry of the over-extension of state power remains the greatest worry. Perhaps so much so that it may, by itself, make the prohibitions of hate speech unjustifiable. Mentioned in Gelber’s paper are few examples of the state using hate speech legislation to justify the removal political dissidents. This possibility may pose a greater threat to democracy than the effects of hate speech, and thus some form of Baker’s free speech absolutism may remain defensible.
5 Arendt and Epistemic Arguments against Free Speech
Euclide est un véritable despote ; et les vérités géométriques qu'il nous a transmises sont des lois véritablement despotiques – Mercier de la Rivere
In a characteristically remarkable article in the New Yorker titles Truth and Politics, Arendt exposes the historical relationship between truth and politics, and comes to investigate its relational status during her time; an analysis which I find more pertinent today.
Arendt distinguishes between two types of truths factual truths and rational truths. Rational truths are truths which are the result of philosophical, mathematical, and scientific rigour. Thus, in a sense they are eternal. If a mathematical axiom, or a scientific discovery were to be lost, we would not lose them for ever; as through rationality, it is plausible that these truths can be found/ discovered again. In contrast, factual truths are fragile and contingent. These truths are essentially historical, and political facts which are susceptible to loss and manipulation. Factual truths cannot be rationally pieced together like rational truths. If erased, the only hope is that we find another account of it in the future.
Arendt describes her contemporary political environment as such,
While probably no former time tolerated so many diverse opinions on religious or philosophical matters, factual truth, if it happens to oppose a given group’s profit or pleasure, is greeted today with greater hostility than ever before.
But why is it that factual truths are met with such hostility? Why does Arendt assume this crucial proposition? Here lies another insightful analysis given by Arendt; namely, that truths are inherently coercive. Foloowing from this is the insight of the fundamental differences between truths and opinion. Opinions thrive on agreement and consent, their power correlates with the number of people who hold them. This needn’t be the case for truth. As truths, themselves, carry an epistemic obligation. To paraphrase Shah, to find out seomthing is true, is to get very good reason to believe it. Thus, seen from a political view, truth is despotic. Thus, truth in itself threatens tyrants, it is a power source that is almost undefeatable.
Moreover, a point of irritability of factual truths is its contingency. Historical facts could always have not been the case. This contingency makes it so that factual truth is no more self-evident than opinion. This may be among the reasons that opinion-holders find it relatively easy to discredit factual truth as just another opinion – their mode of transmission can always be suspected as forgeries.
Factual truth’s inherent contingency had within its arsenal one crucial point of defense, the fabric of factuality. The fabric of factuality, as Arendt describes, is an interconnected system of other factual truths that depend on another, and as they reflect reality, they reflect the reality’s coherence. Thus, when a lie replaces a factual truth, a hole is ripped into this fabric. This a methodology historians can, and have, utilised. When analysisng vast records a lie may come to stand out due to its incoherency with other stated propositions. When this hole is located, a historian or a political analyst may point to it and say ‘this is likely to be false.’ As long as the context itself is kept intact, the lie always comes forth in its own accord.
However, this powerful, and possibly only, defense that factual truths have are now under threat by mass media. As lies are threatened due to their incoherency with the fabric of factuality; if this fabric were to be replaced with a fabric of lies, it ‘will fit without seam, crack, or fissure, exactly as facts fitted into their own original context.’ This comes to threaten the very nature of truth and opinion in political discourse. As if so, ‘what prevents these new stories, images, and non-facts from becoming an adequate substitute for reality and factuality?’ This harrowing warning is realised today, with the closest parallel to this new ‘image’, this icon, being the MAGA movement under Trump; where the factual truths are accused of being icons of themselves, hence distrusted. The despotic nature of truth is being lost; the convincing nature of truth itself is disappeared. This erosion of truth has undoubtedly been crucially assisted by social media – the maximilised mass media – where falsehoods thrive and new fabrics can come to be coherently assembled without restriction.
6 Social Media
The theoretical landscape outlined so far finds its modern testing ground in the digital sphere: a communicative environment in which the epistemic conditions Mill presupposed, and the structural inequalities Gelber described, are radically transformed. Platforms like Twitter (now X), Facebook, and TikTok constitute the new “public square” of discourse, yet their architecture amplifies precisely the distortions and asymmetries that render both free and hate speech deeply problematic. The recent reaction to the assassination of Charlie Kirk epitomizes this inversion. In the immediate aftermath, MAGA-aligned figures (previously stalwart defenders of “free speech absolutism”) began invoking the language of “hate speech” and “dangerous rhetoric.” The irony here reveals not just political hypocrisy, but a deeper structural instability in how speech, power, and digital mediation now intersect.
Empirically, there is a growing consensus that social media has intensified the reach and virality of hate speech. A 2023 Brookings Institution report found that online hate speech on major platforms increased by over 50% globally since 2019, with Twitter (post-Musk acquisition) showing the most dramatic escalation. Similarly, the Center for Countering Digital Hate documented a 202% rise in the use of racial and homophobic slurs on X(formerly known as twitter) in the first six months following Musk’s ownership and the reinstatement of previously banned accounts. This data suggest that “more speech,” contra Mill and Baker, may not lead to greater epistemic robustness or democratic resilience, but to a polluted discursive environment in which misinformation and hate mutually reinforce each other. This is no mere theoretical problem: the algorithmic infrastructure of social media thrives on outrage. Hate speech, misinformation, and emotionally charged content generate higher engagement, and therefore higher revenue. The “marketplace of ideas,” once conceived as a rational exchange of arguments, has become what Arendt would call a “marketplace of images”; self-reinforcing feedback loops of emotional and ideological affirmation.
The justification for free speech, that the contest of ideas purifies truth, thus collapses under the weight of virality. Mill’s deliberative ideal presupposed not a platform designed to monetize attention, but a civic culture that values rational inquiry. Social media inverts this: attention replaces truth as the operative currency. The value of speech, then, becomes subordinated to its algorithmic value.
The digital landscape also complicates Gelber’s systemic model of hate speech. On social media, hate speech no longer simply “reinforces” systemic discrimination, it becomes a system in itself. The anonymity and reach of online platforms make possible new forms of harassment that are both horizontally distributed and structurally embedded. The “pile-on” phenomenon, in which marginalized users are targeted, transforms ordinary speech into a collective weapon of exclusion. Yet, paradoxically, the same logic fuels “cancel culture”; the mass denunciation of individuals (often for transgressive or unpopular speech) that functions as a quasi-democratic form of moral regulation. Both dynamics rely on the same technological affordances: virality, anonymity, and the collapse of public/private boundaries. Here, again, the rhetorical inversion surrounding MAGA discourse becomes telling. The movement that once denounced cancel culture now decries “hate speech” against its own figures. When conservative commentators invoke hate-speech rhetoric to defend figures like Kirk, they implicitly concede that speech has real harm-producing power, precisely the claim they long rejected. What has changed is not the principle, but the position of power. The internet has flattened public discourse to such an extent that everyone now fears being silenced: the progressive feminist of 2015 and the conservative influencer of 2025 are, in a structural sense, playing the same game.
This leads to the central paradox of digital free speech: where to draw the line, and who should draw it. As of 2024, most major platforms use a hybrid model of content moderation, blending algorithmic detection with human oversight. Yet these systems remain deeply flawed. Algorithms fail to capture context and irony; human moderators face impossible scales of content and cultural variation. Moreover, attempts at regulation (whether through the EU’s Digital Services Act or U.S. congressional hearings on Section 230) confront the same philosophical impasse Baker warned of: any mechanism strong enough to suppress hate speech could easily be repurposed to suppress dissent. The danger, therefore, is twofold. Without regulation, hate and misinformation corrode the fabric of democracy, producing Arendt’s “fabric of lies.” With regulation, the state (or platform owners) acquire powers of speech arbitration that threaten autonomy and dissent. No one knows who speaks truth anymore, or who should decide what truth is.
Social media thus transforms the classical liberal question “how much freedom can speech tolerate?” into a new digital tragedy. Platforms amplify hate and falsehood not because of ideological bias, but because of their structural incentives. Through the vicious cycle of trends, every group becomes both oppressor and oppressed in turn. What remains unresolved is not whether hate speech should be censored, but whether the digital public sphere can sustain the epistemic and moral preconditions of free speech at all.
The challenge for any governing body, then, is not simply enforcement, but legitimacy:
Can regulation be executed in a way that distinguishes between harmful speech and legitimate dissent without reproducing the very power asymmetries it seeks to correct?
Questions to Consider for this Prompt:
Can we maintain Mill’s deliberative ideal when digital discourse rewards emotional outrage over rational argument?
Should hate speech laws be historically contingent, changing as power dynamics shift, or should they apply symmetrically to all citizens?
Is Baker’s fear of state overreach justified in the digital age, where much of speech regulations is now corporate rather than governmental?
Could the censorship of falsehoods, in order to maximizes the Millian truth-tracking prerogative, be the answer?
Can we preserve the sanctity of truth without such censorship? Or is censorship of falsehoods necessary to preserve the very nature of truth itself?
How can “cancel culture” and state censorship be distinguished in practice, if both silence dissent through public sanction?
Can hate speech and free speech co-exist within the same communicative structure, or does one always undermine the other?
How salient is Waldron’s worry that hate speech erodes the assurances that we give to each other in society? Is this threat existential?