Introduction: The Said and the Unsaid
One of the more striking features of the latest technology to capture our collective attention – namely the use of Large Language Model (LLMs) – is the extent to which it relies on, and is in fact animated by, what is perhaps our oldest technology, namely language. Despite all the anxieties that Artificial Intelligence (AI) in general, and LLMs in particular, have generated among those of us who engage with language for a living, one cannot deny that they focus our attention very sharply on language, and that we increasingly relate, not just to other humans, but to mechanical objects of all kinds using these clunky old things called words. What is more, none of the mechanisation of language carried out by LLMs does anything to eliminate the immense ambiguities, uncertainties and complexities that literary scholars in particular have always found in their object of investigation. Quite the contrary, one could argue that it is precisely now, as LLMs make it possible for natural language to replace machine code as the primary means for human-machine interactions, that the tools of literary analysis become most urgent and germane (for examples, see Hristova et. al.; Magee et. al.). For surely anyone who wants insight into how LLMs work (or how they are reshaping human experience) would want to begin with a rich understanding of how language works, and perhaps especially literary language, if by that we mean language without a fixed and uncomplicated referent in the world.
An intriguing example of what I am pointing towards here is a longstanding effort among computational scientists to create machines capable of detecting irony. The puzzle is an intriguing one, of course, because language-based AIs – including older semantic-based models and more recent transformer-based LLMs – create meaningful responses to prompts through a statistical analysis of colossal databases of explicit statements. But irony involves precisely meaning something other than what one explicitly says. As the tradition going back to Quintilian tells us, in irony ‘something contrary to what is said is to be understood (contrarium ei quod dicitur intelligendum est)’ (Quintilian 401). How, then, could a probabilistic analysis of what is explicitly said ever capture its meaning – even if its data set were literally everything that has ever been said?
In essence, the answer that computational scientists have come up with involves placing the question of meaning in brackets, and simply looking for external signs of irony. Before the advent of transformer-based LLMs, this meant gathering together huge caches of ironic statements (from Twitter posts labelled #sarcasm or Reddit posts tagged with a /s, indicating the author’s sarcastic intent) and searching for what was called ‘syntactical fingerprints’, or tell-tale regularities that allowed machines to predict that a speaker was not serious or intended their words in a sarcastic fashion. And in terms of detecting sarcasm, or the crudest form of irony, these systems were remarkably successful (Ghosh and Veale; Veale 15–19). The so-called self-attention mechanism of LLMs and their ability to process data sets across multiple levels at once has resulted in exponentially more subtle irony-detection tools (Gole et. al; Zhang et. al.; Lin et. al.). But the same issue remains. The machines might be able to determine from outward expressions that a speaker does not intend what they say. But how could a machine possibly intuit the meaning that is not said? By what set of calculations, no matter how expansive and complex, could it conjure the unsaid up from the said?
The fact that we humans are capable of such conjurations – that we can convey meaning by saying something other than what we mean, and that we actually do so quite frequently – is something that, I think, should surprise us more often than it does. How is it that words can mean other than what they mean? How is it that others can recognise or understand that, in a hidden place behind or beneath my words, in the pitched chamber of my mind, I want to say something else? And how is it that the very thing that seems to make language such a poor tool for communication – including the ambiguities, uncertainties and complexities mentioned above – can simultaneously make it an effective one?
Most recent discussions of irony among philosophers have tended to focus on its normative-ethical dimensions and consider whether it represents a responsible attitude towards the world. The seminal text here is undoubtedly Richard Rorty’s Contingency, Irony, and Solidarity, which characterises irony as a healthy repudiation of foundational metaphysical truths and recognition of the contingency and multiplicity of worldviews. Alternatively, Jonathan Lear’s A Case for Irony argues that irony is one crucial element within a process of ethical self-formation or ‘becoming human’ (Lear 3–5). On Lear’s account, it is not so much an attitude of detachment or a rejection of foundational truths as it is an experience of radical uncertainty – an ‘uncanny disruption’ of one’s ‘practical identity’ (56) or a ‘vertigo’ (17) that exposes a gap between the aspirations and the actuality of one’s chosen place in the world. Richard Bernstein’s Ironic Life responds with a comparative analysis of Rorty’s and Lear’s respective positions, siding largely with the former’s pragmatism against the latter’s perfectionism. And it ultimately suggests we treat irony as an aspect of a philosophical persona, and especially as a means for negotiating the relationship between philosophy as a technical theoretical discipline, on one side, and as a practical way of life, on the other.
But the puzzle of how LLMs might recognise, much less produce, ironic discourse cannot be addressed by considerations of the normative-ethical significance of an ironic disposition. Far more fundamental issues seem to be at stake. Indeed, beyond the already complicated task of defining irony, here we might need to explore the genuinely critical-transcendental question: what are irony’s conditions of possibility? What must things like language and meaning, privacy and publicity, subjectivity and community, or our psychic and social lives be, and how must they operate, for something like irony to exist? The most comprehensive exploration of irony in this more expansive sense with which I am familiar is the Australian literary critic Claire Colebrook’s Irony, which traces the history of the concept from ancient through medieval and early modern times, to the crucial watershed moment among the Romantics in the nineteenth century (when, as Colebrook explains, irony shifted from being an identifiable rhetorical trope to being an almost metaphysical category, and a privileged way of configuring the human capacity for creativity), to twentieth century New Critics, post-structuralists and postmodernists. In the shadow of this monumental project, I propose to probe the issue across the axis of just two contrasting approaches: the neo-Romantic deconstructive investigation of irony in the work of Paul de Man; and a pragmatic speech act theory model of irony promoted most explicitly by the analytic philosopher John Searle. Between these two positions, I maintain, we can develop a conception of irony as a fundamental aspect, not only of language, but of any social relation whatsoever.
My argument unfolds in three sections. The first is organised around a close conceptual analysis of de Man’s crucial essay on ‘The Concept of Irony’. I show how, for de Man, it is not possible to place limits on the operation of irony, and how what he calls its ‘permanent parabasis’ interrupts or undoes all language and all efforts to communicate meaning. Following Friedrich Schlegel, de Man concludes that, owing to the unavoidable possibility of irony, an experience of nonunderstanding is as fundamental as understanding. The second section pieces together a speech-act theory inspired response to this line of thought. Here irony might be represented, not as an irreducible experience of nonunderstanding, but as a conventionally determined and rule-governed language game. At the same time, I suggest, while there is an inescapably public aspect to irony, there is an equally essential private or secretive one. Irony thus entails what (echoing Kant but also bending his meaning) I propose to call an ‘asocial sociability’ or being held together by the very thing that holds us apart. The final section explores this concept of asocial sociability by drawing on the work of Georg Simmel and Byung-Chul Han, both of whom maintain that our ultimate ignorance of one another’s inner life is the condition of trust, and that secrecy structures the social bond. I conclude with some comments on the implications of this approach for constructing LLMs capable of mimicking human interaction, and the irreducible place of error, ignorance and the unknown in the latter.
Permanent Parabasis
De Man begins ‘The Concept of Irony’ by noting the irony of his title. For ‘irony’, he says, ‘is not a concept’ (163). If it were a concept, he continues, it would be possible to give it a definition. But this has proved surprisingly difficult throughout the history of reflections on the theme, including among those who de Man believes have explored the issue most rigorously, namely the German Romantics. One could say, with Northrup Frye, that irony is a rhetorical trope – one that involves ‘a pattern of words that turns away from … its own obvious meaning’. But all tropes involve a ‘turning’, for turning is the very meaning of trope. Thus, Frye’s definition, if it can even be called a definition, is too ‘all-encompassing’ to be of much use. Indeed, de Man suggests, ‘definitional language seems to be in trouble where irony is concerned’ (164–65). It is as if every effort to define irony or fix its meaning is haunted by the possibility of irony, or the possibility that the meaning of words is not fixed.
From here, de Man develops an opposition between what he calls ‘American’ and ‘German criticism of irony’. As a representative of the former, he chooses the literary critic and rhetorician Wayne Booth, specifically Booth’s A Rhetoric of Irony. Initially, Booth is praised for what de Man calls his ‘eminently sensible’ approach to ‘practical criticism’, and for prudently avoiding any effort to define irony in favour of attempting to identify it (165). His question is thus not ‘what is irony’, but ‘how do I know that the text with which I am confronted is going to be ironic or is not going to be ironic? … By what markers, by what devices, by what indications or signals in the text can we decide that a text is ironic or not?’ This, not incidentally, is the exact same question asked by those interested in training LLMs to detect irony – what external features, what ‘syntactical fingerprints’, or quantifiable regularities, characterise it? By what markers, what devices, what indications or signals will the LLM know it? But just as soon as Booth puts the question on the table, he realises he has a problem, one that he is anxious to solve straight away. Because every text (in fact, every discourse, every utterance, every representation of any kind) is at least potentially ironic. They could all be either intended or interpreted as ironic. And this means that every decision concerning irony remains open to doubt – a doubt that has no inherent limit but could conceivably be pursued to infinity. Thus, in a footnote that de Man quotes at length, Booth both acknowledges the danger and attempts to stamp it out. ‘How do you know that Fielding was not being ironic in his ostensibly ironic attack on Mrs. Partridge?’ Booth writes, referring to Henry Fielding’s Tom Jones:
If I answered this with a citation or other ‘hard’ data in the work, I can of course claim that Fielding was ironic in his use of them. But how do I know that he was not really pretending to be ironic in their use, not in fact ironically attacking those who take such data without irony? And so on. The spirit of irony, if there is such a thing, cannot in itself answer such questions: pursued to the end, an ironic temper can dissolve everything, in an infinite chain of solvents. It is not irony but the desire to understand irony that brings such a chain to a stop. And that is why a rhetoric of irony is required if we are not to be caught, as many men of our time have claimed to be caught, in an infinite regress of negations. And it is why I devote the following chapters to ‘learning where to stop’ (Booth qtd. in de Man, 166).
For Booth, that is to say, no amount of evidence or ‘hard’ data, no accumulation of empirical examples or facts, can put a limit on irony. For every bit of ‘data’ could itself become ironic or swallowed up by ‘the spirit of irony’. The only way to avoid the ‘infinite chain of solvents’ or the ‘infinite regress of negations’ that irony sets in motion is to ‘understand’ it.
Booth’s manoeuvre, then, has an implicitly Kantian flavour. It involves shifting from the sensible to the intelligible, from intuition to the understanding. We might not be able empirically to prove something is ironic using ‘hard’ data. But, on the level of the intelligible, those who understand it – those who are intelligent – will know it when they encounter it, and thus know when to ‘stop’ looking for it. It is precisely on this point in Booth’s argument that de Man is able to pounce. For ‘irony’, he explains, ‘is always the irony of understanding’ and ‘what is at stake in irony is always the question of whether it is possible to understand or not understand’ (166–67). Exactly as Booth describes, irony cannot be reduced to ‘hard’ data, intuition, or experience. But that is because it involves a ruse of the understanding, or an irreducible undecidability of the understanding. Irony is possible precisely because we can never be certain whether we have understood.
In order to follow this line of thought further, de Man refers to the German Romantic writer Friedrich Schlegel’s discussion of irony in his ‘Über die Unverständlichkeit’, which he variously translates as 'On the impossibility of understanding’, ‘On incomprehensibility’, and ‘On the problem of the impossibility of understanding’. If ‘irony is tied with the impossibility of understanding’, de Man maintains, ‘then Wayne Booth's project of understanding irony is doomed from the start’. For ‘no understanding of irony will ever be able to control irony and to stop it’. And what goes for ‘the possibility of understanding’ would eo ipso go for ‘the possibility of reading, the readability of texts, the possibility of deciding on a meaning or on a multiple set of meanings or on a controlled polysemy of meanings’ (167). Irony, then, is not just a trope, or one use of language among many. It threatens the operation of language, meaning, and communication as such.
What follows is a long and involved discussion of ‘the German criticism of irony’ mentioned above, focusing particularly on Friedrich Schlegel and his most important influence, Johann Fichte. De Man’s aim is to show how Fichte’s philosophical system of subjective idealism as set out in his Wissenschaftslehre or Science of Knowledge forms the foundation for Schlegel’s literary-poetic conception of irony. The essential point is that, for both Fichte and Schlegel, the system of logic cannot be distinguished from a system of tropes, or substitutions and displacements, at the core of which sits ‘catachresis’, or the act of naming by misnaming (173). Here irony (which in its simple form is catachrestic misnaming or saying something while meaning something else) becomes the condition of all tropes, all logic, and all language in the widest possible sense. Thus, and as de Man notes, Schlegel ultimately maintains that ‘philosophy is the true home of irony’ which is best defined as ‘logical beauty’. And ‘poetry’ has the capacity to ‘rise to the height of philosophy’ because in poetry ‘irony’ is not limited to ‘ironic passages’ but can be found ‘everywhere’ (Schlegel qtd. in de Man, 177). In poetry, in other words, it is always possible that the words mean something other than what they say, and thus impossible to fix outward expressions to inward intentions.
For de Man, the radical implications of this line of thought would be difficult to overstate. From the philosophical analysis of irony in poetry we arrive at the conclusion that language does not convey meaning but, in de Man’s terms, ‘interrupts’ it. Or rather, the paradoxical condition for meaning is its interruption. A little earlier in the essay, de Man pre-emptively itemises three strategies that various thinkers have used to ‘defuse’ irony in this sense or avoid its destructive consequences. First, there is an effort to ‘reduce irony to an aesthetic practice or aesthetic device’ or treat it as a playful distancing from the truth that nevertheless leaves the truth intact. Second, there is the suggestion that irony concerns ‘a dialectic of the self as a reflexive structure’ and that the multiple possible meanings of any ironic utterance can be comprehended and contained by a larger, more stable subjectivity. Finally, and perhaps most complexly, there is an attempt ‘to insert ironic moments or ironic structures into a dialectic of history’ or, as we see in Hegel and Kierkegaard, to subsume irony into a teleological pattern in which truth and meaning are ultimately destined to win out (de Man, 169–70). From de Man’s perspective, all three of these models must be militated against. Anyone serious about irony must come to treat it, not as a moment within a larger structure or process (whether it be aesthetic, subjective or historical) but as what de Man calls ‘permanent parabasis’ or the permanent interruption or disruption of every attempt at meaning.
The final part of ‘The Concept of Irony’ is designed to reinforce this central proposition concerning the ‘permanent parabasis’ of irony and to generalise its consequences. After briefly analysing studies of Romanticism and irony by Walter Benjamin and Peter Szondi (and suggesting that both ultimately follow a Hegelian line and reduce irony to a ‘dialectic of history’), de Man returns to Schlegel’s ‘Über die Unverständlichkeit’. He points specifically to a key passage in which Schlegel argues for the positive social and moral value of nonunderstanding, ignorance and concealment. ‘But is nonunderstanding, then, something so evil and objectionable?’, Schlegel writes. ‘It seems to me that the welfare of families and of nations is grounded in it’, he continues:
An incredibly small portion [of nonunderstanding] suffices, provided it is preserved with unbreakable trust and purity, and no restless intelligence dares to come close to its holy borderline. Yes, even the most precious possession of mankind, inner satisfaction, is suspended, as we all know, on some such point. It must remain in the dark if the entire edifice is to remain erect and stable; it would lose its stability at once if this power were to be dissolved by means of understanding. Truly, you would be quite horrified if your request were answered, and the world would all of a sudden become, in all seriousness, comprehensible. Is not this entire infinite world built out of nonunderstanding, out of chaos, by means of understanding? (Schlegel qtd. in de Man, 183).
The strange virtue of irony, then, is that it preserves a certain darkness or negativity – an inescapable experience of nonunderstanding. The genuine ‘horror’, as Schlegel calls it, would not be this experience of nonunderstanding, but a thoroughly transparent world of pure understanding. And if there is order or meaning, if anything is understood, it relies on this more primordial ‘chaos’ of nonunderstanding. ‘Any attempt to construct – to narrate – on no matter how advanced a level, is suspended, interrupted, disrupted, by a passage like this’, de Man concludes (184). There is no truth or meaning beyond the reach of irony and nonunderstanding. There is no way to ‘stop’ it. There is only its infinite operation.
Indirect Speech
De Man seems to reach the outside boundaries of the transcendental treatment of irony, where irony becomes the condition for the possibility, or rather impossibility, of meaning in the widest imaginable sense. He thus quite deliberately pushes up against the limits of common sense. The most obvious alternative to this approach – which is extravagant by any measure – would appeal to something like Ludwig Wittgenstein’s pragmatic conception of language or J. L. Austin’s theory of speech acts. The argument here would be that figures like de Man have allowed themselves to get lost in the problem of meaning (and thus wandered into any number of metaphysical labyrinths) when the real function of language has less to do with meaning than it does with use (or rather, in language, meaning is the effect of use). On this model, irony could be presented, not as a nonunderstanding at the foundation of all understanding or a mysterious encounter with the unknown, but as one possible move in a conventionally structured language game or one kind of speech act among many – one thing that we might do with words. What is more, the problem of the ironist saying one thing explicitly but intending another implicitly could be solved by noting that the intention of a speech act is not buried in the dark and inaccessible recesses of the speaker’s mind but perfectly legible in what Austin calls the ‘illocutionary force’ of their words, or the specific action those words are designed to accomplish in the world.
This speech-act theory conception of irony is best represented by John Searle, notably in his long essay ‘Literary Theory and Its Discontents’. There Searle characterises irony, metaphor, and other forms of ‘indirect speech’ as products of the difference between what he calls ‘sentence meaning’ and ‘speaker meaning’, or the literal meaning of what a speaker says, on the one hand, and what the speaker intends to do with their words, on the other (645). To use Searle’s example, if I say, ‘the window is open’, I might simply mean that the window is open, in which case sentence meaning and speaker meaning are identical. But I might mean that it is cold outside. Or I might be asking you to close it. Or, if I say these words in the context of a diplomatic meeting, I might mean there is an opportunity for negotiation. The key is that the relationship between sentence meaning and speaker meaning is conventionally determined or the effect of an identifiable set of rules. It depends, as Searle puts it, ‘on a set of principles and strategies by means of which speakers and hearers can communicate with each other in ways that enable speaker meaning to depart from sentence meaning’. In irony and in all forms of indirect speech, there is ‘a systematic set of relationships between the conventional meaning of a sentence and the particular historical speaker’s meaning, as determined by the speaker’s intentions on particular historical occasions’ (647).
Treating irony as a speech act in this fashion would certainly go a long way towards clarifying how an LLM might identify examples of it. For here irony would be a purely public and rule-governed phenomena, entirely susceptible to probabilistic calculations. But Searle’s approach also seems to require a considerable amount of confidence in what he calls the ‘Background’ conditions of speech acts – the shared community of understanding or sensus communis in the Kantian sense. And Searle seems much more interested in establishing what kind of linguistic acts such a Background allows for than in explaining how it is generated or transmitted. Searle is happy to admit that it is entirely contingent, and there is nothing necessary or natural about the rules and conventions governing linguistic interaction. But both the process of learning these rules and the process of following them would have to be largely unconscious or beneath the order of explicit linguistic expression. And here we broach the possibility that irony might not only be an effect or product of the Background, but a condition of it as well. Put differently, it may be the case that the social norms and conventions that Searle believes make irony legible or intelligible are structured by the opacity and uncertainty – the nonunderstanding as Schlegel and de Man put it – that irony entails.
We might get at this issue by considering the everyday operation of what Austin would call a ‘felicitous’ or successful ironic speech act. Like all tropes, indeed all rhetoric and language as such, irony involves a set of social relations or a scene of interaction. To be ironic, it is not enough for me as an individual to say something other than what I mean to myself. Along with the ironic speaker, there must also be at least two other actors on the stage. First, there must be an, as it were, epistemologically higher audience who recognises that what I mean is something other than what I say. I must successfully convey meaning to another by saying something other than what I mean. Second, however, and just as importantly, there must also be a second, epistemologically lower audience who takes my words at face value, who does not penetrate beneath the surface or understand my actual meaning but remains (usually comically) in the dark. Of course, the second audience does not need to be present. It does not even need to be real and can be entirely a product of the imagination. But it is one of the implied presuppositions for any ironic speech act. In its most elementary form, irony works when someone understands my true meaning while watching or imagining someone else mistake it.
Thus, just as Searle says, there is something essentially social and public about irony. It is a conventionally structured speech-act part of the meaning of which is found in its effect on an audience, or what speech-act theorists call its ‘perlocutionary force’. But at the same time, it is a sociality and a publicity that necessarily entails something asocial or private. For irony only operates as a social relation insofar as the epistemologically higher audience that understands what is meant also understands that some aspect of the one who is speaking remains fundamentally hidden from view – that the speaker has an inner life that cannot be intuited by others, and that they are therefore capable of holding a representation and not divulging it, potentially forever. Indeed, this capacity for secrecy is one way of configuring the dangerous infinity of irony that Booth warns us against, and de Man wants to embrace. And this suggests that irony is a social relation that paradoxically relies on that which withdraws from all social relations. It is a kind of asocial sociability, a relation through separation, a public appearance or manifestation of what ultimately disappears into privacy. Or put differently again, in irony we engage with the world on the basis of our capacity to disengage or keep secrets from it. We share what divides us and are held together by the very thing that holds us apart.
Asocial Sociability
Every act of irony, then, or ironic speech act, involves a combination of publicity and secrecy, sociability and its opposite. While strange at first glance, this notion of asocial sociability sits within a larger intellectual tradition. Along with being one way of expressing what Schlegel meant by the experience of nonunderstanding considered above, it was in many ways central to the work of the social theorist Georg Simmel, who set it out in systematic terms and incorporated it into his unique methodology. Simmel characterised society, not as a unified object of investigation, but as an amalgamation of everyday interactions. More accurately, he sought to derive what he called the general ‘forms of sociation’ from detailed analyses of everyday interactions, or moments when individuals revealed some aspect of themselves to one another while concealing others. Thus, in his programmatic statement ‘How is Society Possible?’ Simmel wrote: ‘The fact that in certain respects the individual is not an element of society constitutes the positive condition for the possibility that in other respects he is. The form of his socialised being is determined or codetermined by that of his unsocialised being’ (18). For Simmel, in other words, it was not only true that humans are equipped with the capacity to keep their motives and intentions to themselves. This capacity was also a necessary feature or ‘positive condition’ of relations with others. We can only reveal because we conceal, Simmel maintained. Our ‘socialised’ and our ‘unsocialised’ being are strictly inseparable.
This idea was developed most extensively in Simmel’s essay ‘Sociology of Secrecy and of Secret Societies’. On face value, ‘Sociology of Secrecy and of Secret Societies’ looks like a prolegomenon to an empirical study of the phenomenon of secret societies. It thus sets out a broad theoretical frame within which it proposes to examine societies like the Illuminati, the Pythagoreans, the Gallic druids, the Freemasons, the Omladina, the Carbonari, and so forth. In fact, Simmel’s central claim is much more capacious. For, in essence, he argues that secrecy is not one type of social interaction among many but a structure or form of all interaction. Indeed, on Simmel’s account, this relationship between secrecy and social interaction is deeply rooted in the human way of being in the world. ‘In the presence of the total reality on which our knowledge is founded’, Simmel declares, ‘our knowledge is marked by peculiar limitations and aberrations’. We ‘preserve only so much truth’ and ‘so much ignorance’ or ‘error as is useful for our practical actions’ (444). Human knowledge, in other words, is partial and incomplete, and ignorance is as fundamental as truth. At the same time, while all beings might remain hidden from us at various times in various ways, the other human being is a unique case. For the human being is equipped with an ‘interior’ world, which it can either ‘intentionally reveal’ or keep hidden with ‘a lie or concealment’. ‘No other object can reveal itself to us or hide itself from us in this way as a person can’, Simmel claims, ‘because no other modifies its behaviour through consideration of its becoming known’ (444–45). But Simmel’s point is not merely that humans can conceal their inner world. It is that, in order to interact in any robust sense, they must. Simmel explains:
Whatever we say, as long as it goes beyond mere interjection and minimal communication, is never an immediate and faithful presentation of what really occurs in us during that particular time of communication, but is a transformation of this inner reality, teleologically directed, reduced, and recomposed. With an instinct automatically preventing us from doing otherwise, we show nobody the course of our psychic processes in their purely causal reality and from the standpoints of logic, objectivity, and meaningfulness complete incoherence and irrationality. Always, we show only a section of them, stylized by selection and arrangement. We simply cannot imagine any interaction or social relation or society which are not based on this teleologically determined nonknowledge of one another (464).
That is to say, to interact in any fashion whatsoever is to conceal infinitely more than we reveal. And if this were not the case, if we sought to be perfectly transparent or open with one another, then social life would devolve into a cacophony.
But Simmel wants to push this notion of asocial sociability even further. For, on his account, it is not simply a matter of pragmatics, or the need to be selective in order to be coherent. Simmel’s more powerful claim is that, precisely because we must conceal to reveal, every interaction relies on a mutual act of faith in one another. In ‘Sociology of Secrecy and of Secret Societies’ Simmel refers to this as the ‘credit-economy’ (446) of social relations, by which he means an unspoken exchange of trust required for even the most minimal understanding. More recently, the contemporary cultural critic Byung-Chul Han has picked up on this aspect of Simmel’s work and placed it at the core of a critique of what he calls ‘the transparency society’. On Han’s account, the dominant value of the modern world is transparency, and the almost universal assumption is that unfiltered openness generates social trust. In fact, Han proposes, the exact opposite is the case. There is no need for trust in a world of clarity or translucence. On the contrary, trust – and by extension the social bond as such – first appears amid opacity and concealment. As Han puts it, ‘trust is only possible in a state between knowing and not knowing’. Transparency, on the other hand, ‘is a state where all not knowing is eliminated’. Thus, ‘where transparency prevails, no room for trust exists’ (47–48). Indeed, from Han’s perspective, ‘the society of transparency is not a society of trust but a society of control’ (vi). Put differently, ‘human beings become sociable’, not when they are perfectly open or ‘intimate’, but ‘when they preserve distance from one another (36)’. Distance, then, is a paradoxical condition of proximity. And secrecy is a condition of trust.
Conclusion: Negative States
Among the stranger ideas circulating today must be the idea that we make machines more human, and perhaps even more human than human, by making them increasingly smart, and that their intelligence will somehow be most convincingly indexed by their ability to produce language. For while language certainly makes humans smart, it would not be too hard to show that it also and in equal measures makes them dumb, and that it leaves us dumbfounded as often as it leaves us informed. With a similar line of thought in mind, Gilles Deleuze often took issue with what he called ‘the dogmatic image of thought’. He meant an image of thought as a simple mechanism for distinguishing between truth and error or meaning and nonsense. Here, Deleuze explained, thought is presented as inherently capable of and receptive to truth. And error always denotes thought’s failure – an external corruption or internal malfunction of its otherwise immaculate realm. Following Nietzsche, Deleuze attacked this image of thought as myopic and naïve. He thus insisted that ‘mature, considered thought’ is never exclusively oriented towards truth but includes ‘negative states which are profound in entirely different ways’ (Nietzsche 105). Thought thus cannot be separated from ‘stupidity, malevolence, and madness’ or ‘cowardice, cruelty, baseness and stupidity’ (Difference 151). On the contrary, for Nietzsche and for Deleuze, these represent some of thought’s fundamental possibilities. And each of them thus needs to be approached, not as a breakdown, but as what Deleuze calls ‘a structure of thought’ (Nietzsche 105).
Surely something similar could be said about the human capacity for irony and secrecy, as well as their close cousins like deception, distortion and prevarication. These are not corruptions of thought or tricks that we play on thought in order to prevent it from arriving at its natural home of truth. Nor can they be understood in exclusively ornamental terms, as aesthetically pleasing or rhetorically convincing detours on the way to the truth. They are as much as anything else ‘structures of thought’ fundamental to human experience and to our relationships with ourselves and our worlds. And while it is hard to imagine how these sorts of things might be mimicked by way of probabilistic calculations and statistical regularities, this is not to suggest that they are somehow uniquely human, or that LLMs or some similar device will never be able to perform them. It is only to suggest that, if we ever are going to replicate them mechanically, those charged with the task of doing so will have to reorient significantly their established and deeply entrenched understanding of what it means to be human, and what it means to live in language and with others. And here the practices of literary criticism and literary theory would remain impossible to circumvent.