Neo-Luddism

This is a primer on neo-Luddism and on the two-hundred-year career of the insult it is built from.

Before you accept a label for somebody, find out who benefits when it sticks.

tl;dr: A Luddite today is somebody who is against new technology for no good reason. The word is used to end arguments. The Luddites of 1811 were skilled textile workers whose demands were about who controlled the terms of their work, down to who was allowed to do the job. Soldiers broke the movement. Over the next two centuries the account drifted, in every genre that retold it, into a story about men who hated machines. That picture turns a dispute about power into a personality flaw: resistance gets called irrational, then ridiculous, then dangerous, while the demand goes unanswered. German weavers rose against the same industrial squeeze in 1844, and German culture remembers their hunger. The English version was a choice, and Germany made a different one. Neo-Luddism is the attempt to take the word back, from Chellis Glendinning’s 1990 manifesto to students today who name their clubs after it. Around what the press now calls the AI backlash, the same escalation is running again. The worry underneath it stays impersonal: many Americans fear what AI does to jobs in general; few workers expect to lose their own. The open question is whether the next round of arguments about AI turns on wages, ownership and control, or stays a conversation about who is afraid of the future.

Table of contents

• What happened in 1811
• How a movement became a punchline
• The German counter-memory
• Taking the word back
• Picked up from below
• The work the word does
• What the fear explains
• The third wave
• A word worth arguing over
• Related concepts and further reading

What happened in 1811

The term has fascinated me since 2013, when I found myself defending the machine breakers in a re:publica talk about the end of work.¹ A political movement with specific demands was defeated militarily and then converted, over the following two centuries, into a name for a personality flaw. The history of the word “Luddite” is a history of power, and the conversion was useful, is still useful, and is running again right now.

The machine breaking comes first because the usual account gets the sequence backwards.

From 1811, in the textile districts of Nottinghamshire, Yorkshire and Lancashire, skilled workers began breaking frames in nighttime raids. They worked inside an old apprenticeship system that regulated who was allowed to do the job and to what standard. The new machinery arrived in the middle of that system and was used to get around it: cheaper goods of worse quality, made by people who had never served their time, sold by owners who had found a way past every rule that used to constrain them. Wages fell.

The machines they broke were the ones being used to cut them out. The fight was over who set the terms of the work. A smashed frame was a message aimed at a production regime. Brian Merchant’s Blood in the Machine returns names, workplaces and strategic judgment to people usually remembered as a mob. The shortest version I know runs, “They didn’t want ‘zero technology,’ they wanted to feed their families.”²

What happened next usually gets edited out. Machine breaking was made a capital offense. The government put more soldiers into the English Midlands against the Luddites than Britain had fighting Napoleon in Spain. Men were hanged. The movement ended because it was crushed by the state on behalf of the mill owners.²

In the standard story the Luddites are swept aside by the tide of progress. Later accounts treat their disappearance as proof that technological advance defeated an obsolete worldview. A worldview cannot be hanged. People can. The industrial future that later looked inevitable was imposed by soldiers, which is a thing inevitability does not normally require.

How a movement became a punchline

The word survived its people, and it changed jobs.

Open a dictionary today, and “Luddite” means somebody opposed to new technology, usually with a suggestion of stupidity attached. Steven E. Jones tracked how that meaning took hold. His Against Technology reads the two centuries after 1811 as a cultural history of an idea, following the label through novels, poems, cartoons, films and political speech and showing how it hardened into a permanent category: the anti-technology person, the man shaking his fist at the future.³

Jones does not hand us a culprit, and the honest version of this claim says so. No pamphlet announced the new meaning, no decade can be pinned down as the turn, and no author is on record profiting from it. What the material shows is a meaning that drifted in one direction across every genre that retold the story and never drifted back. Drift like that can look like nobody’s doing, which is precisely what makes it hard to argue with. The English record cannot settle on its own whether the same events had to be remembered this way.

Notice what the category does. It takes a dispute over wages, standards, ownership and control and relocates it inside the head of the person making the argument. The Luddite in the dictionary has no demands, because a demand would have to be answered. He has a temperament. You do not negotiate with a temperament; you wait for it to pass, or you explain the benefits more clearly, or you feel a little sorry for the man who cannot keep up.

The label also moves the dispute in time. The person objecting becomes the one standing in the way of the future. The mill owner becomes the future’s local representative. Accept that chronology and resistance is late by definition, because the arrival of the machine now counts as the event, while the wage cuts, the loss of standards and the broken customs are downgraded to reactions. Cause and response quietly swap places.

Military defeat also took away the movement’s ability to argue about its own memory. The name then traveled without the demands that had given it meaning, which is what made it so widely usable: by the twentieth century a person could invoke the Luddites in a fight about office computers, television or the internet while knowing nothing whatever about apprenticeship rules or the wages of a textile district. The label became portable precisely because history had been emptied out of it.

The German counter-memory

The question of whether those events had to be remembered as a war on machines has an answer. It comes from Germany, where the same insult sits on top of an entirely different memory.

Germany had its own machine breaking in the same period, with 1830 among its peak years.⁴ The event that stayed in the culture is the Silesian weaver uprising of 1844, when weavers ruined by falling wages and the competition of mechanized production attacked the premises of the merchants and manufacturers who set their prices, wrecking the works. The historical scholarship reads the uprising as a social conflict about hunger, wages and dependency, in which some machinery got destroyed along the way.⁵

The literary memory went a different way. Heinrich Heine wrote “Die schlesischen Weber” in the same year. The weavers in his poem curse God, the King and the Fatherland while they weave the shroud of old Germany. Not one line is about a loom. Gerhart Hauptmann’s play Die Weber in 1892 staged the same events as a social tragedy of impoverished families. The authorities treated it as politically dangerous, which is a reasonable judgment about a play in which the poor are right.⁵ England remembers a war against machines. Germany remembers hunger.

Two countries, one industrial process, two different national stories about what the workers were angry at. That is the counterfactual the English record could not supply. Nothing inside a broken frame required the English verdict to win.

Which makes the German present absurd, because “Maschinenstürmer” works here exactly as “Luddite” works in English. It is the standard move against anyone who raises a question about a technology, and German scholarship has said so out loud: reviewers of the historical literature on the Maschinensturm note that friends of progress have long misused early industrial machine breaking “zur Diskreditierung jeder Technikkritik,” to discredit any and all criticism of technology.⁴ The word takes the German memory of hunger and converts it back into the English caricature at the moment of use. German skepticism about nuclear power, genetic engineering and now AI gets written up as a national neurosis, the famous German Angst. The content of those arguments, which is usually about risk and control, stays unexamined behind the diagnosis.

Taking the word back

Reclaiming an insult is an old political technique, and the Luddite version starts, in print, in 1990.

Chellis Glendinning published “Notes toward a Neo-Luddite Manifesto” in the Utne Reader that year, the earliest documented appearance of the term.⁶ Three of her claims founded the position. Neo-Luddites are people who like human making and human culture. All technologies are political, carrying the interests and values of whoever built and deployed them. And technologies have to be judged by their social, economic and ecological effects, since judging them by personal convenience is how a society ends up in a place it never chose. Her program went a great deal further than most people who use the label today would follow: she called for dismantling nuclear, chemical, genetic, television and computer technologies. The analysis and the program can be separated and mostly have been.

Kirkpatrick Sale supplied the history five years later. Rebels Against the Future (1995) recovered the Luddites as pragmatic defenders of skilled work and local community, then pushed on to a larger target, arguing that industrialism itself, and not only its capitalist form, has been catastrophic for people and for the living world. Sale also performed the position with a showman’s instinct. At an Utne Reader event at Town Hall in New York in the mid-1990s, he brought an IBM PC on stage and destroyed it with a sledgehammer, which took him two solid blows.⁷ The stunt guaranteed attention. It also handed the caricature a perfect photograph.

In Wired, the same year he bet Kevin Kelly a thousand dollars that by 2020 the world would see currency collapse, serious war between rich and poor, and converging environmental disasters. On the last day of 2020 the appointed judge gave the bet to Kelly, who noted that his win was “a squeaker.”⁸ Sale’s diagnosis of industrial costs deserves to be argued with on its merits. The bundle of forecasts deserves its own record. Reclaiming a slur creates no exemption from evidence.

The movement’s own definition arrived at the Second Luddite Congress in Barnesville, Ohio, in April 1996, which described neo-Luddism as “a leaderless movement of passive resistance to consumerism and the increasingly bizarre and frightening technologies of the Computer Age.” Stephanie Mills was there. The wider milieu was the American bioregional and simple-living scene.⁹ One demarcation matters for the definition and can then be dropped: anarcho-primitivism, John Zerzan’s line, rejects agriculture, domestication and civilization as such. That is a separate and more radical position. Neo-Luddism is the narrower thing, a politics of which technologies get built and on whose terms.¹⁰

So what does the position hand you to work with? Langdon Winner asks what a system is doing to the people inside it, and his answer is to switch it off for a while. He coined “epistemological Luddism” in Autonomous Technology in 1977: unplug or dismantle a technical system “in order to create the space and opportunity for learning.” Winner himself has since deflated the phrase, describing what he had in mind as a tiny, modest experiment, a person separating from a technology they depend on for about a week and noticing what becomes visible.¹¹ His durable contribution is “Do Artifacts Have Politics?” from 1980, where he showed that arrangements of power can be built into an artifact itself so that the political question arrives with the object and does not wait for the question of use.¹² Sociotechnical Imaginaries covers the institutional version of that insight, and No future is neutral the general form.

Gavin Mueller puts the question of who gains back into labor terms. Breaking Things at Work (2021) treats workplace technology as an instrument of class power, arguing that much of what is called “automation” mainly extends managerial control, deskilling and surveillance. His definition of Luddism as a political stance: worker autonomy, technology as a site of struggle, refusal of production for its own sake, open antagonism toward the way capital shapes machines.¹³ Marx’s fragment on machines sits behind this, with machinery as congealed social relations.

Ivan Illich asks what a better arrangement would look like, the question criticism of this kind stands accused of ducking. His answer in Tools for Conviviality (1973) is a test you can apply to a single tool: it is convivial when it enlarges what people can do themselves, and the measure runs through its scale and through who is allowed to use and repair it.¹⁴

Picked up from below

In 2021 a group of Brooklyn teenagers led by Logan Lane started meeting with flip phones and paperbacks and called themselves the Luddite Club. The New York Times went back in January 2025 and found Lane at Oberlin, still chairing an organization then in the final stages of becoming a non-profit, with campus chapters spreading across a list of American universities and a few abroad.¹⁵ The chapter count varies by source. The naming is documented: these students chose the word for themselves.

The wider Gen Z Luddite wave is partly a media construction. A Harris Poll in 2024 found around half of Gen Z respondents wishing X or TikTok had never been invented, which measures regret about particular platforms. The Economist and CNN have both amplified the trend story and then qualified it themselves.¹⁶ The clubs are real, the platform regret is well documented, and the headline supplies a scale the organizational data cannot yet carry.

So the reclamation comes in two waves, running in opposite directions. Glendinning claimed the term from the top down, through a manifesto, in a magazine. Teenagers claimed it from the bottom up, by adopting it as a club name. Neither wave stopped the other use: the same word is a weapon in the hands of people dismissing an argument and a banner in the hands of people making one.

The work the word does

Watch the label in action and you can see three moves that hold each other up.

The first move declares the outcome settled. AI is coming; the transformation is inevitable. Adapt or be left behind. Sacasas gave the reflex its name in 2013, the Borg Complex, after the Star Trek line about resistance being futile. I was using it against tech journalists and venture capitalists from a conference stage the same year.¹ Mar Hicks has shown what claims of inevitability accomplish: they transfer power to whoever is building the thing declared inevitable, since a future that is already fixed cannot sensibly be argued with, only prepared for.¹⁷ I have worked through the psychology of this at length in Reactance and future narratives. It needs careful handling here. The inevitability rhetoric does provoke people. It also came second. The interests came first, and the rhetoric is how the interests present themselves in public.

The second move follows the first automatically. If the outcome is settled, then opposing it can only be a failure of understanding. The person opposing it gets a description where an answer would go. That is the whole function of the label. Call somebody a Luddite and you have said, in one word, that their objection is a mood, that they are behind, that history has already ruled. The description is doing the work of an argument without ever taking the risks of one. Using it costs nothing, because two hundred years of cultural sediment mean nobody has to explain the joke. The insult is efficient because the movement it names was already defeated once, so the word arrives with the verdict pre-attached.

The third move is the one that does the real damage, and it is the reason the second one still has teeth in 2026. Ted Kaczynski, the Unabomber, attacked industrial technology and modern organization in his manifesto, and he is routinely called a neo-Luddite. Kirkpatrick Sale was genuinely part of that public episode, having been asked by the FBI in 1995 to assess the manuscript. There is no evidence that Kaczynski was influenced by Sale’s work, and the neo-Luddite label has stuck to him anyway.¹⁸ Its usefulness is obvious: it lets anybody who wants to skip an argument reach past ridicule for something heavier. The New Statesman, writing sympathetically about technology critics, called this the Unabomber trap and warned that radical dissent tipping into violence discredits every legitimate criticism standing nearby. Both parts hold. The trap is real, and it is also being set deliberately whenever a data center zoning objection gets discussed in the same paragraph as anti-tech extremism.

Irrational, then ridiculous, then dangerous. The label alone runs the first two; the third arrives with Kaczynski. Each step catches whoever escapes the one before. None of them engages the demand.

What the fear explains

People who resist a technology are not motivated purely by material interest, and a story in which they are would flatten them just as thoroughly. Affect drives them too. The sharpest instrument for it comes from a different corner of German thought.

Günther Anders, in Die Antiquiertheit des Menschen, described the Promethean gap: the widening distance between what we are able to produce and what we are able to imagine, comprehend and take responsibility for. The gap is structural, a mismatch between the reach of our production and the reach of our understanding. It describes the position of anybody standing in front of a system whose behavior nobody in the room can fully account for. Out of that gap Anders drew a feeling. Promethean shame is the sense of one’s own inadequacy in the face of one’s own products, born and fallible against something manufactured and precise. It is doing plenty of work in the current moment, with the writer watching a model produce serviceable prose in four seconds.¹⁹

The shame settles nothing. A frightened worker can be accurate about deskilling. A nostalgic parent can be right about an exploitative platform design. An angry resident can have precise figures on water and power. Emotion shapes the form resistance takes. The substance of the demand stays open to checking. Treating affect as the explanation repeats the original trick in a more sympathetic voice: politics becomes psychology.

The third wave

After the machine breakers and the manifestos of the 1990s, the third round of resistance got its media label, “AI backlash,” before it got a program.

“AI backlash” became an established media category during 2025 and 2026, gathering labor disputes, copyright fights, data center opposition, safety campaigns, consumer aversion and outright vandalism into one story. Brian Merchant is one of its most visible interpreters. His reading is the neo-Luddite one: what is being resisted is a set of business models that concentrate power and devalue work, and the useful question about any technology is who it serves and who it squeezes.²⁰

The record holds wins and defeats. Organized labor set the template in 2023, when the WGA and SAG-AFTRA contracts established consent, disclosure and compensation for AI use. Local opposition to data centers has won as well: Loudoun County in Virginia refused the Belmont rezoning in March 2024 and ended by-right approvals a year later; the Dutch Council of State struck down the Zeewolde zoning plan after Meta had paused the project, and Ireland ran a de facto connection moratorium around Dublin for years.²¹ Elsewhere it lost outright. Memphis, where xAI’s Colossus runs gas turbines in a Black, low-income neighborhood, produced protest, over two thousand public comments and no legal victory at all.²² Saline Township in Michigan voted 4 to 1 against rezoning farmland for a campus serving OpenAI and Oracle in September 2025. The developer and the landowners sued two days later over exclusionary zoning, and a consent judgment in October let the project proceed while the township took fourteen million dollars for fire services, farmland preservation and a community fund.²² The refusal ran through the channel the law provides and the project is going ahead anyway; in 1811 it took soldiers. PauseAI organizes around catastrophic risk with dozens of local groups, and the Authors Guild passed five thousand certified titles under its human-authorship label by March 2026.²³

The surveys show a worry that is broad, regional and impersonal.

In the United States, the share of adults more concerned than excited about AI rose from 37 percent in 2021 to 50 percent by June 2025, with 10 percent more excited than concerned.²⁴ Edelman finds trust in AI at 32 percent in the United States against 72 percent in China, and the Stanford AI Index records a global rise, from 55 to 59 percent, in the share of people saying AI offers more benefits than risks.²⁵ The backlash is concentrated in rich Western countries. It cannot stand in for a global mood.

The gap runs between the collective and the personal question. Pew found 56 percent of the American public extremely or very concerned that AI will eliminate jobs. Ask workers about their own job, and Gallup found in February 2026 that 18 percent think it likely that theirs disappears to automation within five years.²⁶ The anxiety is collective, held on behalf of a society, at a distance from the individual life. It travels well as a mood. Behavioral evidence is thinner than the survey record: labeling a product as AI-made lowers purchase intentions in experiments, most clearly for high-stakes goods. Those experiments measure stated intentions, and actual opt-out behavior does not appear in the record.²⁷

There is a coherent discourse about the backlash and only a partly coherent movement underneath it. Artists, unions, parents, county residents, safety campaigners and a violent fringe are using overlapping language for very different reasons. Some want compensation, some want a moratorium, some want their kid off TikTok, and some want the substation somewhere else. Treating that as a single front is a mistake that both its promoters and its opponents find convenient, and the opponents get the better deal, because one bad actor anywhere in that coalition can be used against all of it.

A word worth arguing over

Permanent Underclass and this primer are two halves of one thing. That one describes a future being announced from above, a story about mass economic redundancy that arranges expectations and does political work while it circulates. This one describes what happens to the people who answer back. A future gets asserted, and the objections get classified.

The classification is cheap, fast and mostly invisible. Nobody has to conspire. A word does it. To be heard, a technology critic needs an argument, evidence, a proposal and a demonstration that they are not the kind of person the word describes.

There is a version of technology criticism that solves this problem by giving up the fight in advance. It talks about attention, wellbeing, design ethics and healthier habits; avoids the question of ownership; and is rewarded with a place at the table, which is my reading of the trajectory in Center for Humane Technology. The Luddite position, in Mueller’s sense, keeps the question of who controls the machine at the center and accepts being called names for it. Meanwhile, the people most confident about technological inevitability are quietly building their own way out of the future they describe, which is the subject of Exit strategies as collective surrender and, in its more ideological form, of Dark Enlightenment.

What follows from all of this is a habit. When an objection gets classified, recover the thing being fought over. Which technology, deployed by whom, under which rules, and with what effect on other people’s capacity to act? Then ask what remains available: negotiation, refusal or redesign. None of that settles whether the critic is right, and that is the point of it. It restores the conditions under which anybody could find out. A joke cannot do that work for us.

The Luddites wanted the machinery kept out of the hands of people using it to hollow out a trained trade, and they wanted to go on feeding their families. That was an answerable demand. It was never answered; the men who made it were hanged, and then their name was handed to the incapacity to understand progress. Two hundred years of usage have made “Luddite” a word for people who lost, which is a useful thing to remember about who gets to name things.

The students in Brooklyn picked the word up anyway. They are not a mass movement, and their flip phones will not decide anything about AI. But somebody has to go first on a thing like this, and taking a slur and wearing it is an old way of making a fight visible again. Whether it works depends on something none of us can see yet: whether the next round of arguments about AI gets to be about wages, ownership and control, or whether it stays a conversation about who is afraid of the future.

Related concepts and further reading

The theoretical foundation of the historical correction is No future is neutral: every image of the future has beneficiaries, and any technology arrives with an arrangement of power already inside it. Reactance and future narratives covers the Borg Complex and what happens when people are told their future is settled. Cultural Lag and Cultural Lead follows the wandering narrative of automation anxiety, which has been announcing the end of work since the steam engine. Sociotechnical Imaginaries gives the institutional frame for the politics built into artifacts, and Critical Futures Studies is the methodological home of all of it.

On the sources: Merchant’s Blood in the Machine is the best entry point to the history, Jones’s Against Technology to the history of the label, and Mueller’s Breaking Things at Work to the politics.²³¹³ Glendinning’s manifesto takes ten minutes to read and lays out the three claims that founded the position.⁶