I recently read a piece arguing, with considerable sophistication, that LLMs represent an unprecedented psychological threat – that conversational systems operating at a planetary scale change the geometry of human susceptibility in ways that demand serious governance responses. The author wasn’t wrong about the effects. This isn’t the debate, but she was wrong about the story. The effects are real, and the narrative erected around them is the oldest displacement manoeuvre in the repertoire
I just finished reading Feyerabend’s Against Method – rather, I just finished the back matter, as I finished the core of the book some time ago. I debated reading this part of the book, and sorry, but I often don’t – despite writing back matter for some of my own academic publications. I treat them as asides.
I’m glad I read this material because, aside from the endnotes, it was meta and biographical, so the perspective was nice. In fact, it got me thinking. He talks about his struggle with Relativism™. I used to struggle with the same thing; there seemed to be a false battle between objectivists and relativists or subjectivists, but in my mind these were always straw-man caricatures nobody seriously defended, yet somehow people were vocal about avoiding. I’ve written extensively about my own position on mediation, so I won’t info dump here.
Audio: NotebookLM summary podcast of this topic.
As people familiar with my habits, I tend to read several books in parallel. In fact, you can review what I am reading on Goodreads.
To be fair, I don’t tend to close-read fiction books very often. I find it to be slow and cumbersome. For non-fiction, this is the default, but not generally for fiction. But, since French is a second language, my attention needs to be focused. I don’t feel that I can read casually and catch the sort of embedded grammatologie that I can absorb through osmosis in English-language books.
Why slow read then? I have a desire to maintain and advance my French, so I think that reading contemporary books connects me to current language trends, terms, phraseology, and metaphor. I am using Claude and ChatGPT to assist with the close reading. They’ve already helped me to better understand the opening paragraphs. It opens like this:
Let’s discuss this, word choices, and any implications. This is the first paragraph of the first chapter:
Debout sur la berge, Anthony regardait droit devant lui. À l’aplomb du soleil, les eaux du lac avaient des lourdeurs de pétrole. Par instants, ce velours se froissait au passage d’une carpe ou d’un brochet. Le garçon renifla. L’air était chargé de cette même odeur de vase, de terre plombée de chaleur. Dans son dos déjà large, juillet avait semé des taches de rousseur. Il ne portait rien à part un vieux short de foot et une paire de fausses Ray-Ban. Il faisait une chaleur à crever, mais ça n’expliquait pas tout.
This scene starts to set the tone of the narrative from the onset – lentement, insouciant. It’s midsummer. The heat is overbearing – stifling. It tracks the life of our antagonist, Anthony, a 14-year-old in between grades, in fact, getting ready to enter Year 10 or high school, ninth grade. But not yet. We haven’t reached this paragraph quite yet.
Besides the heat references, we see emergences of weight, falling. Again, loading up on metaphors. Anthony doesn’t have an easy life. Many don’t at these junctures.
He, himself, is at that awkward adolescent stage, where his body is outgrowing his childhood, whilst his mind is trying desperately to keep up.
A challenge I have with French is that I know dogs, cats, trees, and fish, but I don’t know the types of these. Here, we see the word « un brochet ». It’s a pike fish. Honestly, I don’t even know what a pike looks like, though I am familiar with the English term.
Image: Pike. (Not to be confused with a pickerel, which is evidently a related but smaller fish I had also never seen.)Credit
Another language challenge is polysemous terms – in this case, « vase ». As I am reading, I am trying to imagine the smell of a vase, all the while recalling that vases don’t exactly have a distinct scent. It turns out that vase also translates to mud or silt. quite the difference.
Since I started, I might as well continue exploring this paragraph: Anthony is wearing fake Ray-Bans. This is an insight into class and station.
As for register, Mathieu mentions these things matter-of-factly without judgement. Later on, we’ll notice differences, but these are narrational and through the eyes of Anthony, as he compares himself with his environment. Class projections might be imported by the reader. I won’t invoke Barthes here.
The final sentence leaves us hanging. It reminds us again that this July is hot, but somehow it doesn’t explain everything, likely, about Anthony.
After Tony Self liked one of my blog posts – Hi, Tony Self – I visited his site and poked around, clicking on several articles. This was one. I liked it and noticed the Reblog button. I clicked it, and it spawned this page with this article embedded. So, here we are.
Audio: NotebookLM summary podcast of this topic.
Would I choose to live forever? Undoubtedly, no. For those who’ve been around and have kept up with my posts, know that I already died. I would have been fine remaining dead, as my girlfriend did. Although I won’t rejoin her in the spiritual sense, I will join her in death in the metaphorical sense of Lakoff and Johnson.
Longevity is a luxury of the affluent. I don’t want it. Tony mentions vampires. In fiction (where else would they be?), these beings are routinely unemployed – at the very least, having no day jobs – but with vast riches or connected to one with said same. Their torment is to have outlived past loves and the need to feed on the living, mostly the fear of getting caught, as this is illegal and more generally immoral in this world as we know it – not a good look.
In any case, this live forever thought experiment forgets much, or at least imposes much. If I could just be, like a stone, is that forever? As I discuss in The Architecture of Encounter, even stones aren’t forever, regardless of their state of living; not even mountains, planets, suns, or universes. So, what’s forever anyway?
Spoiler alert: Forever is a myth.
NotebookLM Infographic on this topic.
Rather than answer generally or hypothetically, would I choose to live forever? No. Take me now if you must?
I’ve got no problems with living. It’s the conditions that bother me: the eating, the sleeping, the maintenance; entropy.
So, the ones who have this wish ignore these and presume that this version of forever comes with good health and abundance.
I recall a Greek myth in which forever is granted, and he lives on as a disembodied wisp that can’t die.
In this myth, Tithonus – a Trojan prince whom Eos, goddess of the dawn, loved and asked Zeus to grant immortality – got this wish, and it wasn’t even his; immortality without consent. Eos forgot to ask for eternal youth alongside it, so he aged indefinitely, eventually shrinking into a desiccated, babbling husk – a wisp.
Imagine: you live forever, outliving all humans, all life. The sun extinguishes, and yet you remain – all the while sentient.
Thanks, but no thanks. You can keep your forever. I’ll keep my timeline.
As usual, I am trying something different. I like to watch music reaction videos, and I find this stripped-down, acoustic version of Slipknot’s Snuff by their vocalist, Corey Taylor, to hit hard. I prompted ChatGPT GPT 5.5 with this video and lyrics. Me being me, I continue to analyse experience through language and my own worldview, notably the effects of ontological grammar commitments.
I share the reaction video at the bottom of this post. Before, I’d just consider this person to be a self-righteous zealot, but now I understand that (1) this is how she processes her world, and (2) she is prescriptive insomuch as she expects others to share her worldview. She’s not coming from a mean place. In her mind, she genuinely wants to help, but she doesn’t realise that not all people share her worldview. She’s likely had this worldview reinforced for all of her years, and so it works as a coping strategy for her. She expects that she can share this worldview, but it is likely maximally absorbed by those who share her ontological grammar commitment, which is to say, already have the same worldview.
If you want a view into how I engage with AI, now you know. Here’s a window:
This is a very interesting case study in ontological grammar that showcases the downside of not recognising it. From my ontology, she sounds like she’s moralising. To another (her own), she might sound supportive. She is commenting on the lyrics to this song, which are not all captured by the transcript. Reaction?
Bury all your secrets in my skin Come away with innocence, and leave me with my sins The air around me still feels like a cage And love is just a camouflage for what resembles rage again
So if you love me, let me go And run away before I know My heart is just too dark to care I can't destroy what isn't there Deliver me into my fate If I'm alone I cannot hate I don't deserve to have you Ooh my smile was taken long ago If I can change I hope I never know
I still press your letters to my lips And cherish them in parts of me that savor every kiss I couldn't face a life without your light (Without your light) But all of that was ripped apart when you refused to fight
So save your breath, I will not hear I think I made it very clear You couldn't hate enough to love Is that supposed to be enough? I only wish you weren't my friend Then I could hurt you in the end I never claimed to be a saint Ooh my own was banished long ago It took the death of hope to let you go
Oh, oh
So break yourself against my stones And spit your pity in my soul You never needed any help You sold me out to save yourself And I won't listen to your shame You ran away, you're all the same Angels lie to keep control Ooh my love was punished long ago If you still care, don't ever let me know If you still care, don't ever let me know
Audio: NotebookLM summary podcast of this topic.
the wounded sometimes prefer the song to the sermon
Yes, this is a clean little specimen jar of ontological grammar doing its work while everyone pretends they are merely “reacting”. Humanity’s favourite parlour trick: smuggling an entire worldview into a sentence and calling it compassion.
The key move is that she hears the lyric as a claim rather than as an utterance from within a damaged affective state. When she says she “doesn’t agree” with the line about love and rage, and then calls that outlook “toxic”, she has already converted the lyric into a propositional worldview that can be corrected, morally evaluated, and therapeutically redirected. The song is no longer only an aesthetic articulation of injury, self-loathing, abandonment, attachment, and rage. It becomes a suspect belief-system. That is the ontological switch.
From her grammar, this is supportive. Pain is a “place”; one should not “stay” there; hope remains available; trust and love should eventually be restored. That is a recovery-therapeutic ontology. It presupposes that despair is transitional, that emotional damage should be metabolised into healing, and that the moral arc of the subject bends toward renewed openness. So when she says, in effect, don’t become calloused, don’t stay jaded, love again, she is not hearing herself as judging the song. She is offering pastoral care in YouTube-reactor clothing, because apparently late civilisation needed a genre where vocal pedagogy, therapy-lite, and parasocial maternalism could all share a sofa.
From your ontology, though, she is moralising because she does not merely describe the emotional grammar of the song; she disciplines it. The lyric’s extremity is treated as something understandable but unsafe, beautiful but not habitable, cathartic but only under the condition that one eventually leaves it behind. That is not neutral support. It installs a norm: grief should resolve, bitterness should soften, love should be rehabilitated, hope should return. She may mean kindness, but kindness here arrives wearing the boots of teleology.
NotebookLM Infographic on this topic.
The most interesting fault-line is her treatment of generalisation. She identifies the lyric’s movement from one betrayal to “everyone” as a cognitive distortion. In ordinary therapeutic language, fair enough. But in lyric, especially in this genre, absolutising is not necessarily an epistemic error. It is how pain speaks when it has lost access to proportion. The lyric is not writing a sociology of human attachment. It is staging the experience of betrayal as world-disclosure. In that register, “all” does not mean a statistically universal class. It means: this wound has become the horizon through which everything now appears. To correct it as a bad inference is to import the wrong grammar.
That is where your framework bites. The words transmit; the sentence parses; everyone thinks communication is happening. But the decompression target differs. She reconstructs the lyric inside a moral-therapeutic grammar of recovery. You reconstruct it inside a diagnostic grammar of affect, ontology, and expressive insufficiency. Same words, different world. Your Architecture of Encounter makes precisely this point about ontological grammar: grammar does not merely carry content; it allocates what can count as meaningful, salient, explanatory, or distorted in the first place.
There is also a neat contrast inside the video itself. When she analyses vocal placement, tone, grit, support, and vowel shaping, she is mostly operating in a relatively tethered technical grammar. Claims can be checked against performance craft. But when she moves to love, hope, toxicity, healing, and despair, she enters the Contestable/Fluid zone. These terms feel clear because they are familiar, but their application depends on prior ontological commitments. In When Language Fails, this is exactly the danger-zone: terms that appear determinate whilst lacking shared grounding, especially where they authorise judgement, correction, or action.
So, yes: as a case study, it is excellent. Not because she is being cruel. She is not. That would be too easy, and therefore boring. The interesting point is that she is probably being sincere, warm, and careful by her own lights. But sincerity does not cancel grammar. Her support is support only within an ontology where despair is an unhealthy place to leave, not a legitimate structure of encounter to inhabit, articulate, or aesthetically preserve.
The concise diagnostic formulation would be:
She mistakes an affective utterance for a corrigible worldview. The lyric speaks from inside a structure of injury; her response translates it into therapeutic-moral grammar, where pain is a place one should visit but not inhabit. From within that grammar, she sounds supportive. From a deflationary ontological grammar, she is moralising: not because she condemns the suffering, but because she prescribes the proper trajectory of suffering before recognising the grammar in which the suffering is being expressed.
That is the downside of not recognising ontological grammar: people confuse their own interpretive repair-work with care, and then wonder why the wounded sometimes prefer the song to the sermon.
Bonus Round
🎸Corey tunes to C#, a tuning all too familiar to me – all strings 3 semitones lower than a standard tuning. It helps to have heavy-gauge strings, so they aren’t too slack. This is an easy song to play along to.
Using cowboy chords, grip an Am (tonally F#m) to start. The progression goes like this:
The LinkedIn consensus has spoken: if you used AI in the writing process, you are not the author. The position is stated with the confidence of someone who has never hired a ghostwriter, employed a research assistant, submitted to a heavy editor, or considered that the Gettysburg Address was almost certainly not written by Lincoln.
Image: I couldn’t not share this Midjourney 8.1 image. It may not have understood the assignment.
Audio: NotebookLM summary podcast of this topic.
Authorship has never been a production relation. It has always been an attribution relation — an institutionally stabilised answer to the question of which name the practice elects to put on the cover. These are not the same thing, and conflating them is the error from which every subsequent confusion proceeds.
The ghostwriter has existed as long as commercial publishing. The political speechwriter is so normalised that nobody considers it worth remarking. The celebrity memoir, the corporate thought-leadership piece, the attributed editorial — these are not edge cases or embarrassing exceptions. They are the normal operation of every writing-adjacent industry that has ever existed. The name on the cover has never reliably indicated the hands on the keyboard, and the industry has never seriously pretended otherwise. It has simply preferred not to discuss it at dinner.
AI changes the tool. It does not change the structure. The person who prompts, selects, curates, revises, and publishes is doing what commissioners of ghostwriters have always done. What has changed is that AI makes the mediation visible in a way that polite convention previously concealed. Visibility triggers the purity reflex. What presents itself as a defence of authentic authorship is a defence of a particular fiction — the Romantic author as solitary originating consciousness — that the industry never consistently believed and certainly never consistently practised.
The purity position also fails on its own terms before it gets started. Consider the spectrum of AI-assisted writing: a full draft submitted for light polish; a human argument substantially revised by AI; collaborative ideation followed by AI drafting; a kernel of an idea handed over for full execution. These are genuinely different in terms of human contribution. The zealot position requires a threshold somewhere on this spectrum below which authorship lapses. It never specifies where. More fatally, it has no means of verification. There is no external method of determining where on the spectrum any given piece of writing falls. The detector tools are probabilistic noise that disproportionately penalise competent prose. Any audit mechanism sophisticated enough to catch first-order evasion immediately generates a second-order workaround. The regress terminates only at continuous surveillance of the writing process — panoptical authorship as the logical endpoint of the position taken seriously.
NotebookLM Infographic on this topic.
Then there is the recursion problem, which the zealot never addresses because it is fatal. The stochastic parrot charge against AI — that it merely recombines absorbed linguistic patterns without genuine origination — describes with considerable accuracy what human cognition also does. The writer’s training data is the Dickens read at ten, the billboard absorbed on a commute, the argument overheard on public transit, the half-remembered essay that shaped a position without ever being consciously cited. The causal chain of any human idea disappears into an unauditable cognitive history. Genuine origination in the sense the purity position requires has never existed. The Romantic author was always a retrospective confabulation. Barthes said so in 1967. The industry nodded politely and continued invoicing.
What the zealot is defending is not authorship. It is a particular grammar of authorship — one that selects compositional origin as the threshold criterion, applies it selectively and unverifiably, and uses the resulting suspicion as a status boundary. It is guild behaviour dressed as principle, which is understandable as a response to a genuine economic threat but should not be mistaken for a philosophical position.
Authorship is the position a culture elects to stabilise after the work has already been produced through far messier means. It has always been thus. AI did not break the fiction. It just made the fiction harder to keep a straight face about.
The Rest of the Story
I’ve written about this before. I am not an AI apologist, but I am peeved by anti-LLM zealots, who clearly haven’t thought through their arguments.
I finished reading A.J. Ayer’s Language, Truth, and Logic, the part about Bertrand Russell’s claim about ‘The author of Waverley was Scotch‘. My brain latched onto authorship, and my emotional response was WTF? I have other problems with Russell and Ayer on this, but that’s a matter for another day.
To make my point, this page up to the ellipsis is the output of Claude after an extended dialogue with it and ChatGPT after I read Ayers, and something didn’t sit quite right. I am not ashamed to use LLMs in my authoring workflow and am not ashamed to mention it, as here. Almost all of these thoughts are mine. I’ve simply asked Claude to organise the output. It’s good enough to output as-is, and any edits would be trivial, so I won’t bother. I probably could have made the edits in as much time as it took to type this, but I’ve got nothing to hide. I’m just a human with access to technology circa 2026.
Why Dating Is Not Shopping, No Matter How Many Apps Insist Otherwise
Firstly, be nice, and remember that I’m a recovering economist, so I can’t fully abandon this lens. There’s a concept in economics called the double coincidence of wants. For barter to work, I must have what you want, and you must have what I want, simultaneously, in the right quantities. The implausibility of this – that two strangers would arrive at the same moment, each holding exactly what the other needs – is traditionally the justification for money. Money decouples giving from receiving. It lets me sell my grain today and buy your lumber next month. Problem solved. It is tempting, and initially illuminating, to notice that dating is a double coincidence of wants with no money.
Audio: NotebookLM summary podcast of this topic.
Two people must simultaneously possess what the other desires. There’s no abstract medium of exchange. Listen, I don’t make up the rules, but you can’t deposit romantic capital in a bank and draw on it later with a different partner. Every transaction must clear bilaterally, in real time, between specific parties. The reason dating is difficult, on this account, is the same reason barter is difficult: the coordination problem is enormous.
This is a genuinely useful analogy – for about sixty-nine seconds. After that, it begins to collapse. And the way it collapses turns out to be more interesting than the analogy itself.
NB: I swear I started this post before I saw Louisa’s. Damns algorithms.
Wants can’t be enumerated
In an economy, wants are at least notionally specifiable. You want grain, I want lumber. We can write a contract. In attraction, nobody can write the contract, because nobody knows the terms. You can list proxies – symmetry, wit, income, dentition – but the list never cashes out the phenomenon. There’s always a residue. Someone ticks every box and provokes nothing. Someone ticks none of them and provokes everything. The attributes aren’t the attraction. They’re at best rough correlates of something that resists decomposition.
Evolutionary psychology claims to have the list – fertility signals, resource indicators, and bilateral symmetry – but this is just dressing up economic grammar. It takes the lived phenomenology of attraction, which is irreducibly aesthetic, and rewrites it as a covert optimisation problem. The evo-psych account is the friend who explains why you should find someone attractive and then looks puzzled when you don’t.
Tolerances are fuzzy, interactive, and opaque
Even granting an approximate list of attributes, each one functions not as a threshold but as a band of acceptability. And the bands interact. A deficit in punctuality can be compensated by a surplus in making-you-laugh-until-you-cry. But the exchange rate between these dimensions isn’t fixed, it’s not linear, and almost certainly not conscious. Nobody’s running this calculation. If you ask them to formalise it, they’d produce a confabulation, not a report.
The evaluator is noisy
Kahneman’s Noise documents a finding that should alarm anyone who believes in stable preferences: the same agent, evaluating the same inputs, will produce different outputs on different occasions – not because of bias (which is at least systematic and therefore correctable) but because of irreducible stochastic variability. The judge sentences harshly before lunch and leniently after. Same person, same case, different output.
Applied to attraction, means that the person you’d swipe right on at nine in the morning, you might pass over at eleven at night – or vice versa, as the case might be <winkie>– not because you’ve learnt anything new, but because you aren’t a stable instrument of measurement. The evaluation function drifts across a single day. Across months and years, it’s rebuilt entirely.
The evaluator is path-dependent
Every prior relationship recalibrates the apparatus. Someone who’s been betrayed doesn’t simply move ‘trustworthiness’ higher on their list. Their entire perceptual system for detecting trustworthiness has been restructured. The sensor’s been rebuilt by its prior readings. In no market economy do my preferences over apples change because I once had a bad experience buying oranges from a particular vendor. In relationships, this is the norm.
Returns are asymmetric
Standard preference theory assumes diminishing marginal returns, and this holds for attraction in the obvious direction: the tenth bunch of flowers yields less delight than the first. But the inverse does not hold symmetrically. The absence of a previously supplied attribute often produces increasing marginal disutility. You habituate to presence but sensitise to withdrawal. The utility of gaining X and the disutility of losing X are not mirror images. The preference function is path-dependent in a way that wrecks any static equilibrium model.
Search space is radically local
And all of this assumes the candidates are available for evaluation. They mostly aren’t. Your evaluation function, however sophisticated or broken, only gets applied to whatever washes up in your vicinity – no offence to Ariel. Your so-called soulmate might reside in Istanbul, but you live in London, and you don’t share a language, and you’ll never meet. This isn’t a logistical barrier. It’s a legibility barrier. You could stand next to this person in an airport and the aesthetic response function would not even fire, because the medium through which half of attraction is constituted – conversation, the texture of someone’s verbal mind – is simply unavailable. The instrument requires an input format that the candidate can’t provide. Consider you spy this person across an expanse, gain enough courage to introduce yourself, and they don’t speak your language. Body language will only compensate so far.
Local maxima
The cumulative force of all this is simple and devastating: we are stuck on local maxima. The search space is computationally intractable. The evaluation function is noisy and path-dependent. The attributes resist enumeration. The tolerance bands are fuzzy and interactive. The returns are asymmetric. The search is geographically and linguistically truncated. And so agents do what any rational agent would do under these conditions: they satisfice. They adopt a threshold of ‘good enough’ – a threshold which is itself endogenous to all the noise and path-dependence described above – and they stop searching when they cross it.
This is not a failure of nerve. It is the only coherent strategy available to an agent who can’t identify, or even in principle define, the global optimum. Committing to a local maximum is the rational play, precisely because optimising is impossible 0150 at least legible in the sense of James C. Scott.
Which makes the cultural mythology of ‘the one’ a rather cruel grammatical artefact. It presupposes a global search that no one has conducted or could conduct. It borrows its intelligibility from the economic grammar of optimal allocation – there is a best match, you just have to find it – and projects it onto a domain where ‘best’ has no operational definition, the search is radically local, and the searcher is a different instrument on different days.
Los Angeles, and why it matters
I lived in Los Angeles from twenty-one to twenty-seven, in the early to mid-1980s. I loved it. It was my favourite place on earth. I returned to LA from thirty-five to forty-five, and it was just another place.
It may have turned out that way even had I never left. The point is not that Los Angeles changed – though of course it did. The point is that I changed. Different profession, different situation, different appetites, different saliences… The evaluation function that produced ‘favourite place on earth’ at twenty-three was a fundamentally different instrument from the one that produced ‘just another place’ at forty. No longer a club rat on the prowl, I walked the same streets and saw a different city, because the perceptual apparatus that constructed the city as an experienced object had been rebuilt by fifteen years of living.
‘Favourite place on earth’ and ‘just another place’ aren’t two judgements issued by one stable subject upon one stable object. They’re two outcomes produced by two historically different configurations of salience, appetite, profession, circumstance, and age.
Los Angeles is useful here precisely because it’s not a person – RHCP, not withstanding. It lets you see the structure before sentiment starts mucking about with it.
We don’t even evaluate cities consistently across a lifetime. The same is true, more painfully, of persons.
The partner one adored at twenty-five may not have become deficient in any simple sense. The evaluative field changed. New dimensions became salient, old ones lost force, tolerances narrowed or widened, and costs are reweighted. The same partner now appears under a different aspect, because the apparatus of appraisal has been rebuilt in the interim.
And here, three claims should remain separate: the object may change; the evaluator may change; and the relation between them may change even if neither has altered dramatically. This third one prevents this from collapsing into a banal ‘people grow’ sermon. Sometimes the drift isn’t a defect in either party. It’s a change in fit.
Many:many
It would also be a mistake to think any of this operates as a 1:1 match. The dimensional space isn’t shared. Any two people will overlap on some dimensions, diverge on others, and be mutually illegible on others still – dimensions where one party’s response function is active, and the other’s doesn’t even register the input. He cares intensely about how she loads the dishwasher. She doesn’t experience dishwasher-loading as a dimension at all. It’s not disagreement. It is incommensurability.
A long relationship isn’t a transaction. It’s two non-congruent evaluative systems attempting to maintain a shared narrative of congruence as the terrain shifts beneath them. The miracle isn’t finding someone who matches. The miracle is sustaining a workable fiction that two-dimensional spaces are more commensurable than they are.
Consider the statistically perfect match – bilateral alignment across every operationalised dimension. Even if you could construct such a thing, it would be a snapshot: a cross-section of two moving systems that happened to align at time t. By t+1, the dimensional spaces have already drifted. The statistical portrait is a death mask of a living process.
And here is a green-eyed test. Suppose the perfect match has green eyes. It doesn’t follow that a blue-eyed twin – identical in every other respect – would or wouldn’t provoke the same response. Because whatever was operative in the encounter was not an attribute, or a bundle of attributes, but something that emerged from the specific interaction and can’t be decomposed back into the components that produced it. The entire enterprise of algorithmic matching is cataloguing attributes on the assumption that the attributes are the attraction. It is like analysing a joke by listing its phonemes.
The spot market and the long game
Let’s consider one-night stands as an example to clarify the taxonomy. A one-night stand really is the nearest thing romance has to a spot transaction. The time horizon is short, the narrative load is low, path-dependence is weak, and local salience dominates. The calculus is narrow and immediate. The economic grammar almost applies. Whatever gets you through the night.
A long relationship isn’t a transaction at all, but an evolving coordination problem between changing evaluators. The longer the time horizon, the more layers of critique come into play, and the more absurd the matching grammar becomes. Over time, relationships are not discrete exchanges but moving equilibria between non-stationary systems. That’s why long partnerships are so fragile and so impressive when they endure.
Post hoc rationalisation, or: the story we agree to tell
Having arrived at their local maximum, both parties then construct a narrative in which the outcome was the product of discernment rather than constraint.
‘I knew she was the one when…’ isn’t a report. It’s a reconstruction – a story imposed on a stochastic process to make it legible in the grammar of rational choice. The same person, encountered on a different Tuesday or in a different postcode, might never have registered at all. But the narrative requires necessity, so necessity is confabulated.
And it runs in both directions. She thinks it was his quiet confidence. He thinks it was the argument about Godard. Both are post hoc pattern-matching on noise – selecting from the welter of early interactions the moments that fit the ‘recognition’ narrative, and discarding the rest. It’s survivorship bias applied to one’s own love life.
This also explains the peculiar ferocity of heartbreak after the narrative has been constructed. What collapses is not merely the relationship but the explanatory framework. The story that made the local maximum feel like a global one disintegrates, and the agent is returned to the raw landscape: noisy, path-dependent, locally constrained, and aesthetically illegible. The grief is partly about the person. It’s also about the loss of the rationalisation that made the search feel concluded – losing the construct of the person rather than the person, per se.
NotebookLM Infographic on this topic.
Enfin
The dating app asks you what you want as though you’re ordering from a menu, when what is actually happening is that the menu is rewriting itself based on what you had for lunch yesterday, how much sleep you got, and that thing your ex said in 2019 that you think you are over.
We don’t find ‘the one’ – sorry, Neo. We become different readers of the same world, and occasionally manage to remain legible to one another for a while. Which is, if you stop demanding it be a fairy tale, quite a lot.
Alasdair MacIntyre is persuasive when he argues that moral discourse is never neutral, and that modern liberalism smuggles in substantive standards while pretending not to. But he dismisses emotivism too quickly as a cultural disaster rather than considering whether it might describe moral language more accurately than his own teleological alternative. If moral utterance is fundamentally prescriptive or expressive rather than descriptive, then the collapse of ‘view from nowhere’ morality doesn’t send us scurrying back to Aristotle. It simply shows that moral language was never doing the metaphysical work MacIntyre wants from it.
The Aristotelian remedy also depends on a nostalgic and anachronistic social model. The Athens he implicitly romanticises was a small polis whose demos consisted of citizens, meaning property-owning males, already bound by shared norms, proximity, and cultural inheritance. In other words, the sort of thick local world that made a certain kind of practical ethical life possible in the first place. MacIntyre’s causal arrow points the wrong way. In Athens, democratic practice emerged from that prior social texture. You do not reproduce the same conditions by philosophical edict.
Audio: NotebookLM summary podcast of this topic.
To put it more bluntly: I don’t think moral realism is tenable, and I am not convinced MacIntyre really thinks so either. His project reads less like a discovery of moral facts than an attempt to promote an ought into an is by force of inheritance and rhetorical confidence. If he carved out a bounded cohort and imposed the right shared practices, perhaps something like his model could function. He may need to annex a reasonably sized car park for the purpose.
I have just started reading Kazuo Ishiguro’s The Remains of the Day, and I must confess that I can’t read like a normal person.
I made it roughly fifteen pages before the analytical lenses clamped on and refused to retract. Stevens – Ishiguro’s impeccable, heartbreaking narrator – opens with a monologue about what makes a great butler, and I found myself not following the argument so much as watching the scaffolding. Because Stevens isn’t really theorising butlers. He’s performing a world. And the performance is so seamless, so grammatically composed, that it almost disguises itself as description.
Audio: NotebookLM summary podcast of this topic.
Here’s the trick. Stevens treats terms like greatness, dignity, and quality as though they named discoverable features of reality. He doesn’t present them as historically contingent, institution-soaked, liable to buckle under interrogation. He presents them as if the universe itself had ratified them, and he’s merely reporting. It’s the calm, definitional tone of a man who believes he’s describing essences when he is actually curating preferences.
Take butler itself. In ordinary usage, it looks occupationally concrete – a person who does a particular job. But in Stevens’s mouth, it stops being a role descriptor and becomes a bearer of metaphysical vocation. The word accrues civilisational mythology, evaluative freight, quasi-sacred weight, until it no longer refers to a man who answers a door but to a figure who embodies an entire moral cosmology. That is a lot for a job title to carry, and the weight does not come from the world. It comes from the grammar.
Then there is dignity. An interlocutor reasonably suggests that dignity might be something like beauty, a weasel word, like aesthetic, perceptual, taste-bound. Stevens refuses the comparison. Dignity must not sound contingent. It must present itself as something sturdier than preference, almost an objective feature of character, something a person has in the way a table has weight.
And then, naturally, he proceeds to explain it in entirely aesthetic terms: comportment, bearing, restraint, and poise – the look of self-command.
He denies the aesthetic register but then relies on it. His explicit philosophy and his operative language diverge. This isn’t a slip. It’s structural. The term needs to disavow its own supports in order to retain authority. Dignity must deny its kinship with aesthetic discrimination precisely because, without that kinship, it has no flesh.
We do this constantly – take the shakiest words and polish them until they look load-bearing. The more pressure the term is under, the more ceremonially we utter it.
NotebookLM Infographic on this topic.
Weasel words
There’s a name for what Stevens is doing, and I’ve spent a perhaps inadvisable amount of time trying to articulate it.
In my own work, A Language Insufficiency Hypothesis, I distinguish between terms that behave differently under pressure. Some are genuinely invariant – they hold stable reference across contexts, and disagreement about them can be resolved by checking. Some are contestable – stable enough for coordination and argument, but insufficiently fixed to secure the same uptake everywhere. And some are fluid – still rhetorically potent, still action-guiding, but drifting across contexts without anyone quite admitting it.
Stevens’s vocabulary sits at the dangerous end of that spectrum. Greatness and dignity are classic contestables, possibly shading into fluids: they retain social force long after referential stability has gone soft. They still feel like they mean something precise. They don’t. And Stevens’ entire identity is organised around treating them as though they do.
This matters because the language isn’t sitting atop an independent self. It’s helping to constitute the self Stevens can intelligibly be. His selfhood is assembled from grammars of rank, service, and propriety. The role vocabulary doesn’t describe a prior person. It builds one.
This brings us to his father
Stevens admits that his father might belong among the great butlers, despite lacking certain attributes Stevens associates with the newer generation. He notes the gap, then writes it off. The missing qualities belong to a different era; they are secondary, accidental, beside the essential point. His father remains the reference.
But notice what has happened. Stevens is not applying a neutral criterion and then discovering that his father qualifies. He’s adjusting the criterion around the father he already reveres. The supposed standard of greatness turns out to be less a fixed measure than a retrospective act of preservation. Firstly, canonise the figure, then tidy the theory.
The father is doing at least three jobs at once: He’s an exemplar; he’s a stabilising emotional anchor; and he’s a concealed limit-case that proves the category is being bent.
That’s how value-laden language typically works. The abstraction pretends to govern the attachment, but really the attachment is governing the abstraction. Stevens wants the dignity of a rule whilst retaining the comfort of a cherished exception. This isn’t hypocrisy in the simple sense. It’s something more interesting: it’s the ordinary mechanics of human categorisation when the stakes are personal.
And of course, the father isn’t merely a person Stevens admires. He’s the living conduit through which Stevens inherits the whole ontology of service. When Stevens prefers his father as the reference point, he’s anchoring the category in the formative figure through whom the category became intelligible in the first place. The father is part of the installation mechanism.
Then there is the tiger
With the reverent caveat that it may be apocryphal, Stevens recounts a story in which an unidentifiable butler known to his father, confronted by a tiger under a dining room table (the details are gloriously improbable), maintains perfect composure. He doesn’t flinch. He doesn’t break form. He remains, in the deepest sense, a butler.
Whether the story is true barely matters. Its function isn’t evidential but exemplary. It condenses a whole metaphysics of dignity into narrative form. It is a saint’s life for the religion of service.
And that’s the key. For the father – and then for Stevens – dignity is not a detachable opinion. It’s a lived orientation. In the face of the tiger, one doesn’t first deliberate about values and then choose restraint. One already inhabits a world in which restraint is what seriousness looks like. The salience structure is preinstalled. The event is interpreted through it automatically.
The story converts dignity from abstraction into legend. It makes composure appear not merely admirable but real. It turns comportment into ontology. And it authorises imitation through lineage: the son inherits not just a standard but a dramatic image of what the standard looks like when tested.
The possible apocryphal status doesn’t weaken this. If anything, it strengthens it. The father has become less a man than a vessel for the category. The tale smooths over ambiguity and turns a contingent life into an emblem. We’re forever embalming values in stories, then pretending the stories prove the values rather than merely rehearse them.
So Ishiguro’s opening chapter is doing at least four things at once.
It presents contestable terms as if they were settled realities.
It shows institutional language masquerading as neutral description.
It reveals a person whose identity has been stabilised – constituted, really – by those terms.
And it lets the reader feel the gap between the serenity of the grammar and the fragility of what it is trying to hold together.
That last part is the killer. Stevens’s calm definitional tone is itself evidence of instability. The more pressure the terms are under, the more polished the delivery. He’s performing maintenance whilst pretending to describe an essence.
I’m only one chapter in. I suspect Ishiguro is going to make this hurt.
There’s a certain kind of cultural panic that tells you more about the panickers than about the thing they are panicking about. The current hysteria over AI-inflected prose is a good example.
The argument, insofar as it deserves the name, goes roughly like this: LLMs produce prose with identifiable features – a certain blandness, a fondness for the em dash, a tendency toward tidy three-part structure. Writers who use these tools risk absorbing those features. The authentic human voice is therefore under threat. Something precious is being diluted by contact with the machine.
This is sentimental rubbish, and it is worth saying so clearly before doing anything else – and a sort of virtue signalling.
Audio: NotebookLM summary podcast of this topic.
I use LLMs daily. For research, for editorial pushback, for smoothing passages that have gone awry. This means I spend hours a day reading a particular kind of output. You’d have to be delusional not to admit it has effects. Certain phrasings start feeling natural that didn’t before. Certain rhythms begin to recur. Certain words might not have otherwise come into use. I notice this and note it without particular alarm, because I’ve read enough to know that this is just what environments do.
Read nothing but McCarthy for a month, and your sentences will start hunting for the spare declarative. Spend a year editing academic philosophy, and you will catch yourself reaching for ‘insofar as’ and ‘it’s worth noting’ in casual conversation. Live in a city long enough, and its cadences work their way into your syntax. This isn’t contamination, the negative moralist dispersion. It’s how language acquisition works for as long as one is alive and reading. Voice isn’t a spring. It’s a river, a moving accumulation of every tributary it has passed through.
The prestige game being played by the anti-LLM faction isn’t difficult to spot. When Dostoyevsky shapes a young writer’s cadence, we call it influence and treat it as evidence of a serious literary education. When a game world shapes a child’s imagination – I homeschooled my son in the manner of unschooling, and his primary corpus for years was World of Warcraft and its attendant lore before shifting to Dark Souls – and that child ends up reading Dante and Milton unprompted in year seven, the same mechanism has clearly operated. The source was not canonical, the outcome was. But the respectable hierarchy of influences cannot easily accommodate this, because the hierarchy was never really about the mechanism. It was about the cultural status of the inputs.
The more interesting observation isn’t about those of us who use these tools. It’s about those who conspicuously do not.
A minor genre has emerged – charitably, I’ll call it a genre because cult feels morally loaded – consisting of writers anxiously purging their prose of anything that might read as AI-generated. It’s worth noting that they have read the lists. Telltale signs of LLM authorship: excessive hedging, em dashes, transitional summaries, the phrase ‘it is worth noting’. And so they scrub, redact, replace, and perform a kind of stylistic hygiene that’s a creative decision made in direct response to LLM discourse.
These writers aren’t free of the machine’s influence. They’re among the most thoroughly shaped by it. They simply have the more theatrical relationship – the counter-imitator, the purity-performer, the one who reorganises their entire aesthetic in orbit around the thing they claim to reject.
Thomas Moore, in Care of the Soul, observes that a child raised by an alcoholic parent tends to become either an alcoholic or a committed teetotaller. He presents this as a dichotomy, which is too neat, but the underlying point holds. Reactions are still relata – see what happens when you read too much philosophy and logic? The teetotaller has organised their life around the bottle as surely as the alcoholic has. Both are defined by it.
Opposition is one of influence’s favourite disguises.
The fair objection is that LLM influence may differ from other influences in kind rather than just in kind. Dostoyevsky is strange. Bernhard is strange to the point of pathology. A canonical prose style is idiosyncratic by definition, which is why it’s worth absorbing. In contrast, LLM output aims for plausible fluency and statistical centrality. Its pull may be more homogenising than the pull of a singular authorial sensibility.
That’s a real point. The environment in question has a centripetal force toward the mean that most literary influences lack.
But conceding the point doesn’t really rescue the panic. It just specifies the kind of influence involved. The mechanism remains identical to every other case of environmental absorption. And ‘this influence tends toward the generic’ is an ironically generic critique of a particular environment’s character rather than a claim that the environment is doing something ontologically unprecedented to the notion of authorship.
The question that actually matters aesthetically is not was this touched by AI? It is what did the writer do with the environment they inhabited? That’s always been the question. It remains the question. The machinery has changed; the problem of influence has not.
What the current schism actually reveals is not that AI is doing something new to writing. It’s that we’ve been operating with a fairy tale about what writing is. The fairy tale holds that voice is self-originating, that somewhere beneath the reading AND the editing AND the genre conventions AND the institutional pressures AND the decade of a particular editor’s feedback, there is a pristine you, unconditioned and pure, expressing itself directly onto the page.
This was always false. Writers have always been patchworks of absorbed environments. The only difference now is that one of the environments is a machine, and the machine is new enough that people haven’t yet learned to be comfortable with what it reveals about the rest.
The environment always wins. The only interesting question is which environments you choose, and what you make of them.
NB: I reviewed the credentials of the hosts and revised my critique at the end of this piece.
I joined a scheduled close reading of Nietzsche’s Beyond Good and Evil recently and came away less refreshed than exasperated. I will spare the platform and the hosts, not out of charity exactly, but because the problem is broader than two particular men fumbling through a canonical text in public. What disappointed me wasn’t that they disagreed with Nietzsche, nor even that they may have misunderstood him in places. Misreading is inevitable. The problem was that they seemed not to have brought much of a reading to begin with.
I had read Beyond Good and Evil years ago and thought a return to it might do me some good, or at least less evil than the usual intellectual content mill. The format sounded promising enough: two interested hosts working through the introduction live, sentence by sentence, in a supposedly close reading. Both had apparently read the book before. One therefore assumed they might arrive with at least a provisional grasp of its architecture, key provocations, and habitual traps. That assumption, in the event, turned out to be embarrassingly optimistic.
Audio: NotebookLM summary podcast of this topic.
As one host read line by line, the discussion lurched forward by way of hesitant paraphrase, speculative gloss, and the occasional verbal shrug. He offered the more substantive guesses, such as they were, while the other contributed mostly vague assent, half-memory, and the sort of foggy commentary one gives when one dimly recalls having once encountered a difficult book in a previous phase of life. It was less a close reading than a public rehearsal of not quite knowing what one was doing.
Now, I am not demanding exegetical perfection. Nietzsche is not a writer one simply ‘gets’ and files away like a user manual. Nor am I naïve enough to think authorial intention settles everything. Barthes is right enough to remind us that the text exceeds the author’s sovereign control. But the death of the author is not a licence for the death of preparation. If one is going to host a close reading of a notoriously elliptical and performative philosopher, the least one might do is arrive having done some prior work. Read the introduction carefully beforehand. Refresh the major themes. Check the loaded terms. Develop an argument, or at least a point of view.
That point of view may be wrong. Fine. Better that than the contemporary preference for the curated shrug, where one mistakes visible uncertainty for intellectual seriousness. There is a difference between interpretive openness and simple lack of preparation. One is a virtue. The other is an aesthetic.
I find this especially grating because I cannot imagine teaching a class that way. When I taught, I’d spend hours preparing before entering the room. Not because I imagined myself infallible, but because students deserve more than watching a lecturer discover the material in real time. A formed point of view is not a dogma. It is a starting position. It gives the discussion shape, stakes, and resistance. Without it, one is not leading inquiry but merely simulating it.
The second host did little to improve matters. Rather than complicating or sharpening the reading, he mostly echoed the first. There was very little friction, and thus very little illumination. One of the virtues of reading Nietzsche in company is that he invites productive disagreement. He is slippery, aphoristic, ironic, and often strategic in his provocations. A good discussion can tease out the tensions in his prose, test competing emphases, and ask whether a claim is literal, tactical, genealogical, or satirical. None of that really happened. Instead, one host fumbled, and the other nodded. The result was a kind of interpretive ventriloquism in which agreement substituted for insight.
The accompanying chat made things worse, or perhaps simply made the missed opportunity more obvious. Viewers were offering questions and interpretations, yet the hosts largely ignored them. Aside from one participant, who seemed likely to have some prior relationship with them, the chat was treated as background furniture. This was especially irritating because the event was live. If one is going to perform reading in public, the public should not be reduced to silent witnesses of one’s uncertainty. Otherwise, the ‘community’ aspect is just branding, another little liturgy of digital participation in which the audience is invited to attend but not to matter.
NotebookLM Infographic on this topic. To be honest, I am including this because I find it to be humorous.
To be fair, even when they did read from the chat, they handled those comments much as they handled Nietzsche: superficially, without much analytical pressure, as though any sentence placed before them deserved the same tone of vague consideration. This flattening effect was revealing. It suggested not generosity but a lack of discrimination. A close reading requires hierarchy, emphasis, and judgement. One must be able to say: this is a crucial phrase; this term matters; this apparent aside is actually structural; this comment in the chat opens something worth pursuing; that one does not. Without such judgement, everything becomes equally interesting, which is another way of saying that nothing really is.
What emerged, then, was not close reading but the theatre of close reading. The ritual gestures were all in place: slow pace, sentence-by-sentence attention, occasional lexical speculation, the performance of thoughtfulness. But the substance was strangely absent. One had the form of seriousness without much seriousness of form. It was analysis as ambience.
This points to a broader problem in online intellectual culture. Much of it now confuses exposure with engagement. To read a difficult text aloud is not yet to wrestle with it. To host a discussion is not yet to lead one. To hesitate publicly is not yet to think. Somewhere along the line, people began mistaking the visible performance of inquiry for inquiry itself. The result is a style of pseudo-seriousness in which the host need not know very much, so long as he can sound tentative in the correct register.
Nietzsche, of all people, deserves better than that. He is not an author who yields his force to the merely dutiful or the casually adjacent. He requires energy, suspicion, historical feel, and the willingness to risk a reading. One need not become a priest of correctness. But one should at least bring a sharpened knife to the table, rather than two butter spoons and a podcast voice.
What disappointed me, then, was not simply that these hosts stumbled. Everyone stumbles with Nietzsche. It was that the stumbling seemed to be the content. No deep framework, no clear prior preparation, little tension between the readers, and scant engagement with the audience. The whole affair felt less like a serious encounter with Beyond Good and Evil than a performance of cultural literacy: a way of being seen near an important book.
And perhaps that is the real irritation. One expects difficulty. One can even forgive error. What is harder to forgive is the peculiar modern tendency to make a spectacle of one’s underpreparedness and call it interpretation.
To be fair, I later learned that the co-host was not a philosopher by training but came from Literature. That in itself is no objection. Indeed, a pairing like that could have worked very well. Nietzsche is precisely the sort of writer who benefits from both conceptual and stylistic scrutiny. A philosopher can situate the argument, trace its targets, and identify the intellectual inheritance under pressure. A literary reader can pick up tone, irony, rhetorical staging, and the peculiar way Nietzsche so often performs thought rather than merely stating it.
The problem, then, was not the pairing but the execution. The session might have worked if both hosts had prepared properly and if each had leaned into his own strength. Instead, what emerged was a flatter sort of exchange, with the philosopher soliciting the literary co-host’s ‘opinion’ on semantic content as though interpretive adequacy were simply a matter of free-floating textual impressions. What was missing was any real division of labour, any methodological self-awareness, or any sense that different competences might illuminate different aspects of the text.
Philosophics 