On the occasion of last week’s “International Day for Countering Hate Speech” declared by the United Nations I thought it appropriate to deal with what seems to me a conundrum. Everyone appears to hate hate speech. And to counter it, they typically create speech that expresses their hate of hate speech. Rather than dwell on the contradictions, I thought it worthwhile to engage a conversation with Claude about speech itself, before going on to deal with hate. The conversation grew organically and led us ultimately to a strong, constructive conclusion. This is the first of three columns to appear this week.

When I look at some of the most serious debates taking place today, I can’t help thinking Lewis Carroll was the prophet who clearly saw the world that would emerge 150 years after his drafting of the vision. Let me focus on one example, an exchange between Alice and Humpty Dumpty.

“So here’s a question for you. How old did you say you were?”

Alice made a short calculation, and said “Seven years and six months.”      

“Wrong!” Humpty Dumpty exclaimed triumphantly. “You never said a word like it.”

“I thought you meant ‘How old are you?’” Alice explained.

“If I’d meant that, I’d have said it,” said Humpty Dumpty.

Humpty’s last statement is a lie, because he knows his intentions and his intention is to play with language. It’s also to assert his authority like a schoolteacher. His statement would be true (and sincere) only if he wasn’t aware of the fact that people routinely employ the literal formulation he uses to ask the question Alice answers. His response to Alice is logically true but profoundly insincere.

In this and many other passages Carroll leaves us with a complex lesson: that language always contains a substantive gap between informative meaning and intention. It isn’t about ambiguity but about the fact that there can never be a stable correlation between the linguistic expression of information and what really matters: understanding why anyone says what they say. We see this problem with AI itself because clearly it doesn’t structure its messages in the same way humans do, who have the liberty to play the kind of game Humpty Dumpty plays with Alice.

Was Wittgenstein influenced by Lewis Carroll? I’ll let you tease that one out. But I raise this question because I see the world struggling with concepts such as “fake news” and “hate speech” and it seems to me that they begin to lose all meaning in a Humpty Dumpty linguistic order. Instead of producing its effects in the form of whimsical, surreal absurdity, it is coldly and brutally logical.

Before addressing the issue I hope to focus on – solving the problem of hate speech – do you think the gap between truth and intention is as radical as Carroll seems to suggest?

Your reading of the Humpty Dumpty passage is sharp, and the distinction you draw between logical truth and sincerity is exactly the right place to press. But I’d want to complicate the picture before accepting Carroll’s implicit claim that the gap is radical — or rather, I’d want to distinguish between two quite different kinds of gap.

The first is the gap you identify: between what is literally said and what is meant. This is real, pervasive, and philosophically well-documented. Grice’s entire project of conversational implicature is essentially a systematic account of how much communicative work gets done outside the literal content of utterances — through shared context, cooperative assumptions, tone, and timing. Humpty’s “How old did you say you were?” is a perfect case: the grammatical form is technically a question about reported speech, but the pragmatic context makes it unmistakably a question about age. Alice is right to answer as she does. Humpty knows this. His triumphalism is therefore not just bad faith; it’s a kind of linguistic tyranny, a refusal to participate in the cooperative framework that makes language function at all.

But the second gap — between stated intention and real intention — is where things become genuinely radical, and where I think Carroll is pointing at something deeper than mere ambiguity. Humpty doesn’t just play with surface meaning; he uses linguistic form as an instrument of power. The assertion “If I’d meant that, I’d have said it” is not an honest reflection on his own usage; it’s a move designed to make Alice doubt her entirely reasonable interpretation. This is Carroll diagnosing something about authority and discourse that feels urgently contemporary: the capacity to weaponize apparent logical precision against common sense.

Where I’d resist making the gap absolute, however, is here: if the gap between expression and intention were truly radical — if there were no recoverable connection between what people say and what they mean — then Carroll’s own satirical point would be unintelligible to us. The fact that we all immediately recognize Humpty’s bad faith, that we side with Alice instinctively, demonstrates that human beings share an enormously robust set of interpretive conventions. We are, as Aristotle suggested, social animals whose language is inherently embedded in a web of shared purposes and mutual expectations. The gap exists within a framework of intelligibility, not beneath it.

So I’d put it this way: the gap is real and ineliminable, but it is not bottomless. It is more like a constitutive tension than an abyss — the very space in which rhetoric, irony, poetry, and manipulation all operate. What Carroll shows us, with characteristic precision disguised as whimsy, is that this tension can be exploited by those with the will and the position to do so. Humpty’s power isn’t linguistic; it’s ontological. He controls what counts as a correct interpretation because he controls the encounter.

And that, I think, is exactly where your intuition about hate speech and fake news becomes so acute — because in both cases, what is at stake is precisely who gets to adjudicate the gap between expression and intention, and by what authority. That is the question worth dwelling on before we reach any definition.

Your point is well taken. In the interest of linguistic (and philosophical) precision I need to replace “absolute” by “ineliminable,” a word easy to understand but which I had never previously encountered. But I appreciate even more your remark concerning “who gets to adjudicate the gap between expression and intention.” The United Nations appears to have had this in mind when it declared June 18 “International Day for Countering Hate Speech.” The obvious problem I see with this begins with the fact that the UN doesn’t appear to offer a definition of hate speech. Have I missed it? I’m sure you can help me find it.

You haven’t missed it — and the absence is itself philosophically telling, especially given the theme of our conversation. Let me fetch the Strategy and Plan of Action document, which is where the UN comes closest to a definition. You haven’t missed it — but you’ve put your finger on something the UN’s own detailed guidance quietly admits. What exists is explicitly described as a working definition, not a legal one, and the gap between the two turns out to be enormous.

The UN Strategy and Plan of Action on Hate Speech offers this working definition: “any kind of communication in speech, writing or behaviour that attacks or uses pejorative or discriminatory language with reference to a person or a group on the basis of who they are — based on their religion, ethnicity, nationality, race, colour, descent, gender or other identity factor.”

The reason it remains a working definition rather than a legal one is frankly stated in the UN’s own implementation guidance: most UN texts prefer to refer to “incitement to discrimination, hostility or violence” rather than “hate speech” as such, largely because there is no definition in international law of the latter term and there is a lack of consensus about its meaning.

This is a remarkable admission for an institution that has just declared an International Day to counter the very thing it cannot define. And it connects directly to the Humpty Dumpty problem in a way that is not merely ironic but structurally significant. Consider what the working definition actually does. The phrase “attacks or uses pejorative or discriminatory language” already smuggles in the gap we have been discussing: whether a given utterance constitutes an attack is precisely a question of intention and context, not of linguistic form. The same words can be reclamation, satire, clinical description, or incitement depending entirely on who says them, to whom, in what setting, and with what purpose. The definition, in other words, presupposes the very interpretive judgment it is supposed to guide.

The most comprehensive attempt to bring some rigor to this is the Rabat Plan of Action, which provides a six-point threshold test to assess whether a given instance of hate speech violates international law. That threshold test — which examines context, speaker, intent, content, extent, and likelihood of harm — is actually a more honest instrument precisely because it acknowledges that no static definition can do the work alone. It requires case-by-case interpretation.

But here is where your Carroll intuition cuts deepest: the question of who adjudicates that six-point test is left entirely open. Conflicting interpretations of the primary international framework — Human Rights Council Resolution 16/18 — have stalled progress, with different blocs of member states reading the same text in irreconcilable ways. Some states read the framework as protecting freedom of expression; others read it as mandating restrictions on speech that offends religious communities. The text is the same; the intentions brought to it are not.

So we find ourselves, perhaps unsurprisingly on the day after the UN’s own observance, in a situation where the concept meant to counter the weaponization of language is itself a linguistic instrument without a stable meaning — deployed by parties with divergent and often unstated intentions. Humpty Dumpty would recognize the arrangement immediately. The question worth turning to, then, is whether a workable concept of hate speech is even possible within a framework of international law, or whether the attempt to define it will always reproduce the very problem it seeks to solve.

Lewis Carroll had clearly helped both of us focus on an essential dimension of the issue of hate speech. I continued the conversation with a new prompt, delving further into what I would call “Humpty Dumpty’s mind.” Claude’s response and our continuing conversation will appear in the next column.

What some governments and much of the media appear to promote is what you refer to as a “static definition” that plays into the perverse logic of Humpty Dumpty. Perhaps the most obvious example of this is evident not so much in the UN’s campaign but in the “war on hate speech” promoted in many countries in the West: the characterization of legitimate political commentary as “antisemitism.” The result is that people are encouraged to categorize hate speech as the use of specific words. But the one positive thing Humpty Dumpty taught us is to be aware of a fact that is subjectively true of all language: “When I use a word, it means just what I choose it to mean – neither more nor less.”

“The question is,” said Alice, “whether you can make words mean so many different things.”

“The question is,” said Humpty Dumpty, “which is to be master——that’s all.”

Would you agree that the issue the UN and many individual governments claim to be concerned about but fail to analyze is exactly what Humpty Dumpty says: deciding who is to be the master?

Your thoughts

Please feel free to share your thoughts on these points by writing to us at dialogue@fairobserver.com. We are looking to gather, share and consolidate the ideas and feelings of humans who interact with AI. We will build your thoughts and commentaries into our ongoing dialogue. 

[Artificial Intelligence is rapidly becoming a feature of everyone’s daily life. We unconsciously perceive it either as a friend or foe, a helper or destroyer. At Fair Observer, we see it as a tool of creativity, capable of revealing the complex relationship between humans and machines.]

[Lee Thompson-Kolar edited this piece.]

The views expressed in this article are the author’s own and do not necessarily reflect Fair Observer’s editorial policy.