February 2018 to now · 201 posts · still running (latest August 2026) · the hinge is the eight months of silence from March to November 2025, which he calls the wilderness · this page is based on 18 posts read in full, spanning February 2018 to June 2026
On 12 December 2025, the Purdue University board of trustees approved a rule: from now on every undergraduate, in every major, must demonstrate an "AI working competency" in order to graduate. Six months earlier Ohio State had launched something similar.
A middle-aged man in Gainesville, Florida read the news and noticed the thing nobody else was noticing:
If something is genuinely inevitable, you don't have to make it a graduation requirement.
Rivers do not need legislation to run downhill. Everyone telling you that AI is unstoppable and that you'd better get on board is, at the same moment, spending enormous money and power making sure the unstoppable thing actually happens. He gave the pattern a name, with a nod to Herman and Chomsky's Manufacturing Consent: manufactured inevitability.
His name is Michael Sacasas, and he writes as L. M. Sacasas. He has been writing technology criticism for seventeen years, the last eight and a half as a newsletter called The Convivial Society: 201 installments, all of it free, no paywall.
Not "which technologies are good and which are bad." He has one question, and he has stated it plainly: the question of technology, pursued to any depth, simply becomes the question of the human.
The line he returns to most often is Ivan Illich's: contemporary man tries to build a totally man-made environment in his own image, and then discovers he can only do it on the condition of constantly remaking himself to fit it.
Following that thread, he has dug out four genuinely original conclusions.
Until 2021 he said what everyone says — attention is scarce, so spend it wisely. Then he read the title of a book Illich announced in 1980 and never wrote: a history of scarcity. History. Meaning scarcity is not a fact about the world; it is a certainty we were taught.
So Sacasas turned the lens on his own field. Every conversation about attention begins by calling it scarce. And the moment you call something a resource, you have already conceded that it can be extracted. In his words: the framing of the problem is itself part of the problem.
His counter-proposition: you and I have exactly as much attention as we need. When it feels insufficient, what is missing isn't attention — it's the judgment of what would be good to do right now.
Three years later he ran the thread to its end with the biggest analogy in his work: the enclosure of the human psyche. When the English commons were fenced, the deepest loss wasn't that the poor lost their cut of the profits — it was that "commons" got rewritten as "resource." What is being fenced now is the mind, and the hedgerows are screens. Distraction isn't a bug; it's the enclosure working. An idle mind, like idle land, reads as waste.
Not "it's hard." Literally impossible.
In December 2012, after the Sandy Hook shooting, he opened Twitter and felt the urge to say something — and immediately recognized the urge as a temptation. Why? Because on social media I am neither alone nor in anyone's presence. Saying nothing is not the same as silence. Silence is felt, it is meaningful, and it requires a body that is actually there.
The same insight explains his most-read post ever, written in April 2020: why is video-conferencing so exhausting? "The problem with video-conferencing is that the body is but isn't there."
This is what he most wants people to hear now: "Your AI is not a tool. It is an environment, and you are in it."
"It's just a tool, what matters is how you use it" is dangerous because a tool can be put down and an environment cannot. What forms you is never the content; it's the tiny motions you repeat every day. His own example: the instant a sentence gets stuck, his hands have already opened Twitter. That isn't a failure of will. It's a will that was trained successfully — by someone other than him.
He named the biggest problem himself. Asked what should actually be done, he wrote: "I get some variation of that question a lot, and I confess that I don't ever feel as if I give an adequate answer."
And one more thing worth holding onto. In March 2025 he published an essay on lethargy — a word rooted in the Greek lēthē, forgetting — and closed by saying he'd probably pick up the pace. Then he vanished for eight months. The post that broke the silence in November was called "Out of the Wilderness," and it said the thing that had broken down was "the labor of thinking well and writing well."
A man who has written about attention for seventeen years turned out not to be immune either. He didn't hide it. He wrote it — and then quoted the line a son says to his despairing blind father in King Lear: "Thy life is a miracle. Speak yet again."
On 12 December 2025, the Purdue University board of trustees voted through a new rule: every undergraduate, whatever their major, must satisfy an "AI working competency" requirement to graduate. The university retooled 22 courses and updated more than 300 plans of study to make it happen. Six months earlier, Ohio State had announced a comparable "AI Fluency" initiative.
Coverage of this sort of news comes with a standard vocabulary: forward-looking, preparing students for the future, fall behind if you don't.
One person read it and saw something else. In his newsletter he noted a curious fact — the very people who are ostensibly convinced of the inevitability of AI seem to lack the confidence you would expect to accompany such a conviction, and instead appear bent on using their power and wealth to make certain AI is imposed on society.
Put bluntly: a genuinely unstoppable thing does not need a graduation requirement to prop it up. No state has legislated that apples must fall. When "inevitable" has to be delivered by administrative mandate, procurement contract, and default setting, it has stopped being a prediction about the future and become a construction project in the present.
He gave the maneuver a name, nodding to Herman and Chomsky's Manufacturing Consent: manufactured inevitability.
That move — taking a sentence everyone is nodding along to and turning it over to see what's printed on the back — is what this man has been doing for seventeen years.
Michael Sacasas, who writes as L. M. Sacasas. He is not in Silicon Valley and not at a research university. He lives in Gainesville, Florida, where he is executive director of a small outfit called the Christian Study Center of Gainesville. He holds a master's in theological studies and was a doctoral candidate at the University of Central Florida working on technology and society, with a focus on Hannah Arendt.
In the summer of 2009 he started a blog called The Frailest Thing and kept it going for a decade. In February 2018 he started this newsletter, naming it in homage to two older books at once: Jacques Ellul's The Technological Society and Ivan Illich's Tools for Conviviality. As of August 2026 there are 201 installments. All of them are public. He has explained that he is trying to hold two principles at the same time: no one should be excluded because they cannot pay, and writers should be fairly compensated. In August 2021 Ezra Klein devoted an episode of his New York Times podcast to him and the subscriber count jumped; in 2024 he turned up on Vox's "Future Perfect 50" list alongside Billie Eilish, and mentioned in the newsletter that the accompanying portrait had taken all the grey out of his beard.
The novelist Robin Sloan once sketched him in a line he liked enough to keep in his own introduction: this is a man who lives on the back side of the hill in a very old house where a conclave of esoteric scholars occasionally gathers — historians, philosophers, philologists, at least one private detective — who come to the island on the ferry, traveling in twos and threes, whispering to each other in a dead language.
One more fact belongs here: from March to November 2025 he disappeared for eight months. The post that ended the silence was called "Out of the Wilderness." More on that below.
On 10 February 2018, the first installment of the newsletter opens not with technology but with being lost.
He says he has been thinking about finding and losing one's bearings. The unnerving part of being lost is not that you don't know where you are; it's that you don't know how to work out where you are. And we lose our bearings, he notes, usually because our vision has narrowed to the unfamiliar ground immediately around us. To recover, you look for a more distant landmark, or you consult a map, or — if you have the skill — you look at the stars.
In that same first issue he quotes a line from Alan Jacobs which, read eight years later, functions as the newsletter's constitution:
You can't understand the place and time you're in by immersion, the opposite's true. You have to step out and away and back and forward, through books and art and music, and you have to do it regularly. Alan Jacobs, quoted in "The Convivial Society, No. 1," February 2018
That is the root question in its first shape: we have lost the standpoint from which our own technological situation could be judged. Not for lack of information — information is overflowing — but for lack of anywhere to stand.
He tightened it over the years. By 2022 he was putting it as: the most important questions about technology are ultimately questions about what it means to be human. Walking out of the wilderness in 2025, he compressed it further:
the question of technology, pursued to any depth, simply becomes the question of the human. "Out of the Wilderness," November 2025
But how did he hit that question in the first place? There is a specific moment.
In late 2014 the veteran Silicon Valley observer Om Malik wrote a piece of self-reckoning that included this:
I can safely say that we in tech don't understand the emotional aspect of our work, just as we don't understand the moral imperative of what we do. It is not that all players are bad; it is just not part of the thinking process the way, say, "minimum viable product" or "growth hacking" are. Om Malik, quoted in "The Questions Concerning Technology," June 2021
Malik went on to say it was time to "add an emotional and moral dimension to products," by which he seems to have meant handling data responsibly and writing clearer terms of service.
What Sacasas seized on was the word add. His response: you don't need to add a moral dimension to technology; it is already there, and the only question is what its nature is. Langdon Winner had famously asked whether artifacts have politics and answered yes; by the same logic, artifacts have ethics.
The obvious next move would be "so we need a code of tech ethics." He didn't make it. In 2016 he sat down and wrote 41 questions — not principles, but questions you can put, one at a time, to any object at all: a phone, Zoom, a fitness tracker, a language model. A sample:
After the list went up in 2016, developers wrote to tell him they had printed it out and pinned it up at their desks. He reposted it in 2021, and Ezra Klein built the podcast episode around it — which is how most people find him today.
Notice the shape of all 41: not one of them asks whether the technology is good. Every one asks what it is making of you. The four ideas that follow all grow out of that turn.
First, the reversal: he used to be on the other side. As late as the summer of 2020 he wrote an entire installment premised on attention being scarce and therefore needing to be spent carefully. That is the field's default setting, and has been since Michael Goldhaber laid out the "attention economy" in Wired in 1997. Goldhaber's argument is clean: economics studies the allocation of scarce resources; information online is not scarce but overflowing; so what is scarce in cyberspace? Attention. Reading a profile of Goldhaber in February 2021, Sacasas wrote that "not that long ago, I would've pretty much assented to this whole line of thought."
But he happened to be reading Illich. After the burst of ferocious social criticism he published in the 1970s, Illich turned around and doubted his own method: the institutions he had savaged were riddled with faults and yet proved far more durable than he expected. His conclusion was that they were anchored at a depth that rabble-rousing could not reach, where a set of what he called certainties lay buried — assumptions we are barely aware of, which lend the current arrangement its air of inevitability. Chief among those certainties was scarcity. In 1980 Illich announced that he intended to write a history of scarcity. A history. Meaning: scarcity is not a physical property of the world but something invented at a moment and then internalized as common sense. The book never materialized.
Sacasas put that key into his own field's lock. The deduction takes two steps. First: every discussion of attention proceeds under the sign of scarcity. Second: to call a thing a resource is to grant that it can be extracted. If attention is a scarce resource, a competitive market for it follows inevitably — which means you are not criticizing the attention economy, you are writing its legitimation.
While I remain quite sympathetic to the spirit of this line of thought, it now seems to me that the framing of the problem is itself part of the problem. "Your Attention Is Not a Resource," April 2021
Then comes the proposition that makes people stop:
you and I have exactly as much attention as we need. ibid.
He knows the objection — there are only so many hours, and reading this means not reading that. So he adds a qualification, and the qualification carries the weight of the whole essay: you have exactly as much attention as you need at any given moment, provided that at that moment you also know what it would be good for you to do. This is not a pep talk. It moves the problem from "how do I ration an insufficient resource" to "what is actually the right thing to be doing now." His own example is working from home: the hard moments weren't the unfinished work, they were the children in the room. Sometimes the good thing is to keep working and sometimes it is to set the work aside — but either way he is not short of attention. He is short of judgment.
Eighteen months later he wired the idea into history. Working from Yves Citton and Jonathan Crary, he found that the "crisis of attention" is not a digital-age novelty at all; it runs back to the nineteenth century, and it was twinned from the start — one half the factory's need for workers who don't drift, the other half the need to advertise a glut of consumer goods. Having read across a hundred and fifty years of it, he distilled a pattern:
The devastating part of the cycle is that it successfully translates an environmental problem into a personal defect. You conclude that you can't focus, that you're weak-willed, that you need another app — and that self-reproach is step two landing on you.
For this he quotes the line from Illich that I'd argue is the master key to his entire body of work:
Contemporary man attempts to create the world in his image, to build a totally man-made environment, and then discovers that he can do so only on the condition of constantly remaking himself to fit it. Ivan Illich, quoted in "The Pathologies of the Attention Economy," September 2022
Two years on, in November 2024, the thread reached its end. He offered a larger analogy and said that if he were allowed one sentence, this would be it: resist the enclosure of the human psyche.
The enclosure of the English commons was the centuries-long conversion of land available to the many into a resource managed and extracted by the few. Two of the justifications are chilling to read now. One was that the commoners were rough, savage, lazy, with an inexplicable preference for sport over paid labour — so they needed managing, for their own good. The other, louder, was "improvement," meaning turning the land to profit; on that logic enclosure wasn't robbery at all, because the commoners made no profit off the commons and therefore had nothing worth taking. Sacasas's claim is that a structurally identical process is now running on the human mind. The hedgerows aren't hedgerows; they are the array of data-gathering tools and techniques. The justifications are the same two.
What makes the analogy detonate is that it reaches back and bites the 2021 proposition. Illich had pointed out that critics of enclosure mostly protested the injustice of the split — why should the profits go to the lords — while missing the deeper thing: the transformation of commons into resource was itself the loss, regardless of who profited. Accept the language of resources, extraction and value and you have surrendered the ground on which any other way of living could be built. Which means: if you say "attention is a scarce resource, so I'll budget it carefully," you are not resisting enclosure. You are surveying the land for it.
The analogy also yields a new definition of distraction, and it is the sharpest inversion he has written: idle land reads as waste to the encloser, and by the same logic a mind at rest, a psyche in a moment of silence, reads as waste too. The unceasing stream of notifications isn't a design failure. It is the extraction engine running properly.
This is the earliest of his themes and the one he has followed longest.
He began writing about technology around 2010 and quickly noticed a recurring trope: not an argument that a technology was good, but an announcement that it was unavoidable. By 2012 he had a nickname for it, borrowed from the assimilating cybernetic species in Star Trek and their slogan "resistance is futile" — the Borg Complex. He even drew up a list of symptoms, including: using "Luddite" ahistorically as a casual slur; equating caution or resistance with reactionary nostalgia; matter-of-factly announcing the bleak future awaiting those who refuse to assimilate.
Then he spent a decade collecting specimens. The funniest is Google Glass. In 2012 Evernote's CEO Phil Libin offered what he called a firm prediction:
in as little as three years from now I am not going to be looking out at the world with glasses that don't have augmented information on them. It's going to seem barbaric to not have that stuff. Phil Libin, quoted in "Resistance Is Futile: The Myth of Tech Inevitability," April 2021
Sacasas's comment: this firm prediction has not, as they say, aged well.
But the real argument doesn't rest on hindsight mockery; it rests on the history of technology. As the historian Thomas Misa put it, we lack a full picture of the technological alternatives that once existed, and of the decision-making that winnowed them down; we see only the results and assume, understandably but in error, that there was no other path to the present — the victors write the history, in technology as in war. So inevitability is a retrospective illusion: we tidy the past into a neat story and then project the story forward. And tech companies understand this habit perfectly well, and are prepared to exploit it.
In December 2025 the thirteen-year thread closed. When universities began writing AI into their graduation requirements, he saw that this was not one more specimen of the old rhetoric but its next stage — from asserting inevitability to constructing it. There is no better way to reinforce the myth than to stage AI's ubiquity so that its adoption becomes a fait accompli: in search results, in software updates, in features nobody asked for. It arrives first, and then you are the one who has to justify not having it.
And the real function of the maneuver is to launder responsibility:
The myth of technological and political and social inevitability is a powerful tranquilizer of the conscience. Its service is to remove responsibility from the shoulders of everyone who truly believes in it. But in fact there are actors. Joseph Weizenbaum, Computer Power and Human Reason (1976), quoted in "Manufactured Inevitability and the Need for Courage," December 2025
Sacasas adds one line underneath: it bears repeating.
But he doesn't stop at "so grow a spine." He does something more honest — he concedes that agency is unequally distributed. What a legislative aide can do, what a startup programmer can do, and what a single parent struggling to pay rent can do are not the same thing. The strategy of manufacturing inevitability works chiefly by obscuring the responsibility of those who have the most agency over the shape of things. Weizenbaum's answer was civil courage, with the specific note that its most arduous exercise is usually not in world-shaking events but in small contexts: fear for one's career, fear of those who appear to have power over us, fear of anything that might disturb the tranquility of an ordinary life.
This thread starts in a very small, private moment, and he came back to it four times.
December 2012, the Sandy Hook shooting. Like most people he learned of it by logging on to Twitter. And he felt the urge to say something — and then, crucially, recognized the urge not as duty but as temptation. Because of the position he was in:
If I am alone, I would feel no urge to speak… If I were in the presence of another, our silence would "say" enough… On social media, I am neither alone nor in anyone's presence. My silence would convey nothing, it would be meaning-less. To be silent is to cease to exist, to be unregarded, unnoticed. "The Convivial Society, No. 12," March 2019, recalling 2012
He returned to it twice more: in 2019, prompted by the Christchurch mosque shootings, and in March 2021, prompted by the shooting in Boulder, Colorado — that one landed on the day he finished writing, so he appended a coda saying something heavy: nearly ten years on, it seems we've not come very far, and he is still trying to say more or less what he was trying to say then.
But in the 2021 piece he pushed it to a new conclusion: it is impossible to be silent online. Not difficult. Literally impossible.
You will object that you do it constantly and it's called lurking. His reply is that we need to distinguish saying nothing from silence. Silence is felt; it is meaningful; it is not mere negation and can even be eloquent — and that kind of silence presupposes bodily presence. He notes a piece of linguistic evidence in passing: the word we landed on for being present without speaking is lurking, which carries an inherently sinister charge, because in non-digitized settings it means watching without being seen, a presence that refuses to materialize.
Then he produces the Illich story from 1926, which I think is the best story in the whole newsletter. Weeks after he was born, Illich was brought to his grandfather's estate on an island off the Dalmatian coast, a place that had barely changed since the late middle ages. And on the same boat, the island received its first loudspeaker.
Up to that day, all men and women had spoken with more or less equally powerful voices. Henceforth this would change… Silence now ceased to be in the commons; it became a resource for which loudspeakers compete… Unless you have access to a loudspeaker, you now are silenced. Ivan Illich, "Silence Is A Commons" (1982), quoted in "Impossible Silences," March 2021
From which Sacasas derives a line that could serve as a law of the internet: to give everyone a loudspeaker is to assure that no one can be heard. It is the cleanest available inversion of the claim that social media handed everyone a microphone — everyone can speak has never meant anyone is listening.
The same bodily thread explains a piece that looks unrelated. In April 2020, barely a month into lockdown, he used fifteen hundred words on the question everyone was asking: why is video-conferencing so exhausting? It became the most popular thing he has written (he confirmed as much a year later). The answer again lands on the body — "the problem with video-conferencing is that the body is but isn't there" — it resembles a face-to-face encounter more closely than any other medium, and for that very reason it frustrates us more, as the mind's non-conscious perceptual repertoire runs at maximum and never quite gets a grip.
Following the same line, in that 2019 issue he also updated Hannah Arendt's banality of evil in passing: the twentieth-century version came paired with the faceless individual swallowed by bureaucracy, whose complicity was clerical and silent, whereas today complicity "arises not from our being cogs in the machine but relays in the network… it is performative and a function of our self-expression."
Stack them and the claim is clear: in this environment you exist only by speaking; speaking costs something; and silence is not an available retreat.
This is what he most wants understood now, and it is the opening of his most combative piece of 2026:
Your AI is not a tool. It is an environment, and you are in it. "Your AI Is Not a Tool," June 2026
His target is the sentence almost nobody disputes: technology is just a tool, what matters is what we do with it. He says he is tempted to become "obnoxiously strident" about it, because — his words — we have just undergone a nearly 20-year, society-wide experiment showing that a so-called tool, a smartphone or a social platform, will malform even the most vigilant and virtuous user into its own image.
Where does the sentence go wrong? On a spatial relation. His distinction is so simple you wonder why nobody put it this way before:
I can pick up a tool and put it down, but the environment absorbs me into itself. ibid.
Better still, he didn't invent the distinction — he picked it up from Illich's late self-repudiation. Illich came to think his earlier work inadequate because he had not yet grasped that somewhere in the mid-twentieth century we had passed from the age of tools to the age of systems. You can stand over a tool; you cannot stand outside a system.
There is a fine irony here, and Sacasas is entirely aware of it: this newsletter is named after Illich's Tools for Conviviality, a book its own author later judged to belong to a superseded epoch. In a footnote he says: read it anyway, it remains valuable.
So how does an environment form a person? For the mechanism you have to go back to a piece from 2022, where he proposes a strange word for our relationship with technology: liturgy — the fixed set of rites, gestures and formulas that orders public worship.
His way in is an observation: some low-church congregations are convinced they have no liturgy at all, and are sometimes hostile to the idea. But they don't lack one; they have an unacknowledged one — the same predictable rhythms, cadences and formulations, and a calendar too, only with Advent and Lent traded for the most thoroughly commercialized holidays. And liturgy matters because it forms people more powerfully than anything said from the pulpit: it inscribes itself not on the mind but on the body.
His diagnostic is concrete enough to try right now: watch for the moment when your body is already in motion before you have thought about what you're doing. In your waking moments, do you find yourself reaching for a device before your eyes have opened? His own example is enjoyably merciless:
When I as a writer, for example, notice that my hands have moved to open Twitter the very moment I begin to feel my sentence getting stuck, I am under the sway of a technological liturgy. "Taking Stock of Our Technological Liturgies," September 2022
Then comes the counter-intuitive step. In such a moment we normally conclude that our willpower failed. From the liturgical view the opposite is true: the will did not fail, it worked beautifully — it has simply been trained to want something contrary to what you say you want. You could even call it a testament to the power of the will, acting in keeping with its training. Stretch the timescale and it holds: if I want to become a more patient person but have at every point adopted the faster, more efficient tool, I have been systematically conditioning myself to expect the world to answer instantly. That I turn out to be someone who cannot abide delay should not be a surprise. To show how well hidden the mechanism is, he borrows a line from Bourdieu:
The cunning of pedagogic reason lies precisely in the fact that it manages to extort what is essential while seeming to demand the insignificant. Pierre Bourdieu, quoted in ibid.
The movements of our fingers, the gestures of our hands, the posture of our bodies — all insignificant, until you realize you are etching the grooves along which your future actions will habitually flow. He closes that piece on a sentence worth keeping: by what we unthinkingly do, we undermine what we say we want.
One corollary lands hardest on the entire genre of advice about using AI well. Every such recommendation — use it critically, stay vigilant, remember to ask it to argue back — presupposes that you can keep doing what you once did. He names this fallacy "Sacasas's corollary to Pascal": Pascal held that the error of the Stoics was imagining a person could do always what they could do once, and we are facing machines that demand unceasing cognitive and ethical vigilance and are designed to wear that vigilance down. He also quotes a reader who watches it run in fast-forward inside a company of 600,000 people, in what I think is the most piercing sentence written about AI in the last two years:
The teams using AI most carefully are the ones losing the ability to tell a good option from a merely safe one. The malformation doesn't skip the diligent. It recruits them. A reader (Joroan Lazaro), quoted in "Your AI Is Not a Tool," June 2026
His prescription is not media literacy. What's required, he says explicitly, is not literacy in a cognitivist mode but the training of perception in an ascetical mode — a project he inherits unfinished from the late Illich, who wanted to gather people to think about what tools do to our perception rather than what we can do with them. And Illich ran that line to an unexpected destination: reclaiming the senses is the fundamental condition for renouncing the technique that sets up a definitive obstacle to friendship.
What travels best from a long-running writer is usually the method, not the conclusions. His has four parts, all portable.
One: widen your temporal bandwidth. The rule was laid down in issue one, and he later gave it a name borrowed from Pynchon: temporal bandwidth, the width of your now. The more you dwell in past and future, the thicker your bandwidth and the more solid your persona; the narrower your sense of now, the more tenuous you are. The paradox being that the harder you stare at the present, the less of a grip you get on it. So his first move on anything new is always to step backwards — to a loudspeaker landing on an island in 1926, to nineteenth-century factory discipline, to Plato complaining about writing.
Two: be a pearl diver, not an innovator. He says it flatly: "I do not prize originality." He understands his work as a project of recovery — bringing Ellul, Illich, Weil, Mumford, Arendt, McLuhan and Ong back into a contemporary conversation. The best description of the method he found in Arendt on Walter Benjamin: like a pearl diver who descends not to excavate the sea floor but to pry loose the rich and strange, because what sinks and dissolves also crystallizes, surviving as "thought fragments" that wait for someone to carry them up into the world of the living. He calls the sentences he brings up amulets — worn, not cited, because they change how you see. The one he has worn longest comes from Wendell Berry, and he says it is perhaps the briefest statement of his philosophy of technology: "We live the given life, and not the planned."
Three: use questions, not conclusions. The 41 questions are the purest sample. He has explained why: he sees his writing chiefly as a way of thinking in public, and where his own perspective shows through he intends it as an invitation to consider a way of being rather than a prescription of behavior. He is also fond of a form he calls fragments — a loosely structured list of associated reflections gathered around a problem — because the form "invites rather than forecloses further thought."
Four: offer three options, not two. Faced with any technology his frame is embrace, negotiation, refusal, and he is explicit that more often than not we ought to be negotiating. It matters because technology arguments are almost always compressed into a binary, and one of his most durable contributions is insisting that the middle setting exists.
He confessed the biggest problem first. In 2022 he devoted a whole installment to the question readers ask him most — what is to be done — and opened it like this:
I get some variation of that question a lot, and I confess that I don't ever feel as if I give an adequate answer. "What Is To Be Done? — Fragments," August 2022
His defence has two layers. First: the line between technology and culture, economics or politics is merely heuristic, so the answer can never be "fix this technical problem and all will be well"; and many of our difficulties are functions of the human condition, which cannot be solved, only better abided — indeed some of our troubles may stem precisely from treating the human condition as an engineering problem. Second, he quotes the Illich scholar Simon Ravenscroft: Illich's late refusal to offer political answers, preferring to promote friendship and hospitality, was itself a kind of politics. (Note: Ravenscroft's essay is a reply to George Scialabba's charge that Illich is insufficiently useful. I have not read Scialabba's original, so I report only the bibliographic fact and do not paraphrase his argument.)
Whether that defence holds is for the reader to judge. But three things look to me like genuine soft spots.
The formation thesis is close to unfalsifiable. "The tool is reshaping you even when you believe you are using it carefully" is structured so that any counter-example gets absorbed: you say you weren't changed? That is exactly what being changed looks like. His evidence is mostly of three kinds — anecdote, introspection, and the insight of several dead authors — and its persuasive force depends on your already sharing his account of the person. He occasionally cites empirical work (the paper on LLMs amplifying false memories in witness interviews, say), but that work isn't his and isn't load-bearing.
The reading list determines the conclusions. He returns to the same people — Illich, Ellul, McLuhan, Arendt, Weil, Mumford, Berry — and makes no secret of his theological training. That makes certain conclusions close to predetermined: benevolent limits, the given rather than the planned life, hospitality, asceticism, friendship. If you don't accept that anthropology — if you think the human capacity for self-transformation is itself a good, or that limits have no intrinsic value — a great deal of his argument loses its force at once. He will not supply you the missing premise, because in his system those are the premises.
He has admitted being scooped. He published "Your Attention Is Not a Resource" on 1 April 2021; three weeks later, in the links section of the next issue, he ran Dan Nixon's 2018 Aeon essay — titled almost identically — and wrote "Dan Nixon beat me to it." I count that as evidence of his honesty rather than a failing, but it does establish something: his most widely circulated proposition was not first said by him. He doesn't mind at all, having told us he doesn't prize originality. But if you came for an original thinker, the truth is that his originality lies in synthesis and in analogy — enclosure, liturgy, manufactured inevitability — not in being first to a fact.
And one thing he was not immune to himself. On 20 March 2025 he published an essay on lethargy — a word whose root is the Greek lēthē, forgetting, which is also the root of alētheia, truth, so that truth is literally un-forgetting. His argument was that to be immersed in the digital stream is to drink from the waters of Lethe, and that we drink far too long each day. At the top of the piece he mentioned he had two or three posts in the works and the pace might pick up.
Then he disappeared for eight months.
The post that came on 25 November was "Out of the Wilderness." He wrote that emerging from a wilderness was the faithful description of those months, that what had gone wrong was "the labor of thinking well and writing well," and that it may have involved the emerging psychodynamics of a post-literate society. In the same piece he set out the thesis he now means to pursue: the arc of artificial intelligence bends toward demoralization — burnout society has phased into the demoralized society.
You can read those eight months as a refutation of his whole account: even the man who studies this professionally couldn't hold the line. Or you can read them the other way: he didn't hide it, he turned it into the next subject. As for which reading is right, I'd rather leave here the line he himself quoted that November — the one a son says in King Lear to his blind father, who believes he has just thrown himself off a cliff and wants only to die:
Thy life is a miracle. Speak yet again. King Lear, quoted in "Out of the Wilderness," November 2025
He had picked that amulet up only a year earlier, in November 2024, and written it into the piece called "Amulets." A year later it was what led him out of the wilderness.