Nobody ever fought the machine

The demand sheet the Korean Metal Workers' Union brought to Hyundai Motor, before the three-day partial strike that began on July 13th, contains one item that gives the whole thing away. Alongside the larger base pay increase, the performance bonus, and a profit-sharing formula tied to the company's annual earnings, the union asked that the retirement age be lifted from 60 to 65. The strike has been reported as a revolt against humanoid robots, and Hyundai has indeed committed to putting Atlas units built by Boston Dynamics, its own subsidiary, into its Georgia plant from 2028. But a union that wanted the robots stopped would have asked for the robots to be stopped. This one asked for five more years of work before they arrive, formal negotiation before deployment, income protection as automation expands, and a cut of the proceeds. The comparison it keeps reaching for is Samsung, where workers extracted enormous bonuses out of the semiconductor windfall, on the argument that if the people making the chips are sharing in the boom, the people making the cars ought to as well.

That is a bargaining position. It is also, almost exactly, the bargaining position taken by the men who gave machine-breaking its name.

A framework knitter in Nottinghamshire in 1811 — a stockinger, in the trade's own word — made hosiery on a knitting frame in an upstairs room of his own cottage, sometimes in a row of houses whose top floors had been knocked through to fit more frames in. He did not own the frame. He rented it from a hosier, one of the merchant capitalists who put the work out, and the hosier charged the rent whether or not there was work that week, because the rent was how he earned his return on the frame. Often the knitter was paid not in coin but in goods redeemable at the hosier's own shop. He owned his labor and his house, rented his machine from the man who bought his output, and sold that output back to him at a price the same man set.

Breaking even

What the stockingers objected to in 1811 was not the frame, which had been in the trade for two hundred years. It was the wide frame, which turned out several stockings at once on a broad sheet of fabric that was then cut up and seamed — cheap goods, made without the shaping that had been the skilled part of the job and the profitable part of the piece rate — and it was the practice of putting men on those frames who had never served the seven-year apprenticeship the trade's charter required. In December 1811 the knitters and the hosiers held public negotiations over wages, piece rates, and frame rents. The talks failed. The frame-breaking followed.

It had begun that March at Arnold, outside Nottingham, where a crowd destroyed sixty-three frames belonging to the hosiers who had made themselves, in the phrase of the time, the most obnoxious to the workmen. They chose which machines to break by which master they had the quarrel with. Over the next two years they broke perhaps a thousand frames out of some twenty-five thousand in the trade, and left the rest standing. Eric Hobsbawm, a historian of industrial Britain, called the practice collective bargaining by riot, and the phrase has held up for seventy years because it describes what was happening: men with no legal standing, no union, and no strike fund, using the one asset they had, which was that the capital was in the room with them.

What settled it was not an argument. Parliament made frame-breaking a capital felony in 1812 and roughly twelve thousand troops went into the Midlands and the northern counties. At a Special Commission convened at York Castle in January 1813, sixty-four men were tried, seventeen were hanged — for rioting, for burglary, for the murder of a mill owner who had promised to ride up to his stirrups in Luddite blood — and twenty-five were transported to Australia. One detail from the muster rolls outlives the rest: a significant share of the militia raised to garrison the frames were themselves weavers, thrown out of work by the strikes of 1807–08. The state hired the displaced to guard the machines that had displaced them, and it worked.

The economists' verdict — that the stockingers were wrong, that mechanization lifted everyone eventually, that the fallacy bears their name — is true on a timescale no human being inhabits. Robert Allen, an economic historian, named the half-century after 1790 Engels' pause. Between 1780 and 1840, by his accounting, British output per worker rose 46% while the real wage index rose 12%; the profit rate doubled and the share of profits in national income expanded at the expense of labor and land. The compensation did arrive. Between 1840 and 1900 output per worker climbed 90% and real wages climbed 123%, and wages have tracked productivity in the industrial economies more or less ever since. A knitter who was forty when he took a hammer to a wide frame would have been sixty-nine the year the pause ended. He was right about his life and wrong about his country, and the word "fallacy" collapses two separate claims into one.

Hold harmless

Which is why the most instructive figure in this history never broke anything. Harry Bridges ran the International Longshore and Warehouse Union, the dockworkers' union on the American west coast, from its founding in 1937 until 1977. The job his men did was moving cargo by hand: gangs of them down in a ship's hold, stacking mixed freight piece by piece, packing the space by eye and by muscle, a week of work to turn a vessel around. The shipping container was going to end all of it, and Bridges could see he would not stop it. In October 1960 he signed the Mechanization and Modernization Agreement with the Pacific Maritime Association and traded the union's work rules away — the shipowners could containerize without interference — in exchange for guaranteed income, pension money, and no layoffs among the fully registered men. The membership shrank. The men who remained got rich.

Then, in 1963, he went back to the table and demanded the employers mechanize faster. Marc Levinson, in his history of the shipping container, records Bridges telling the PMA's negotiators: "We intend to push to make the addition of machines compulsory. The days of sweating on those jobs should be gone and that is our objective." The ship lines hesitated, because the machines cost money. The union was demanding robots and capital would not pay for them.

Bridges understood what the stockingers understood and what the Hyundai union understands. The fight has never been about whether the surplus gets created. It is about the split, and the split is set by leverage.

Read two centuries as a ledger of what each protest cost the party across the table, and the arc is not encouraging. Frame-breaking cost the British state a capital statute, twelve thousand men under arms, and a garrison in the Midlands for the better part of two years. The M&M Agreement cost the shipowners a permanent claim on the gains, paid out for decades. The Writers Guild strike of 2023 ran a hundred and forty-eight days and won contract language keeping AI out of literary material. And in 2025 Anthropic settled with book authors whose pirated work had trained its models for $1.5bn, against a valuation now around $965bn as the company prepares to go public. The largest copyright settlement the technology has yet produced comes to roughly a sixth of a percent of the defendant's paper value, entered as a line item.

Each rung is cheaper than the last for the party that owns the machine. The reason is not that capital has grown more ruthless. It is that the protesters have less to withhold.

Breaking frames works when the capital sits in a cottage down the lane, owned by a man whose name you know. A strike works when you stand between the machine and its output. Hyundai's union can close a factory in Ulsan, and a union that can close a factory gets a conversation about retirement ages. The New York Times cannot close anything. Its only point of contact with OpenAI is that OpenAI read its archive, and so on July 9th the paper and the New York Daily News went to a Manhattan federal court and asked for sanctions, alleging that OpenAI had lied about its ability to search its own systems for their articles. Two protests against machines, four days apart. One party can stop production. The other can file a motion.

The copyright suits are frame-breaking conducted by people who cannot reach the frames. The grievance underneath them is disintermediation — the chatbot answers the question and the reader never arrives — but disintermediation is not a cause of action, and copying is. So the fight gets routed through the one legal relationship that happens to exist between a newspaper and a laboratory that shares with it no employees, no contract, and no physical space.

There is a bargaining table, of course, and some publishers have already sat down at it: News Corp, Axel Springer, the Associated Press, all of them now licensing archives to OpenAI for cash. But look at who is in those chairs. The deals are struck by the owners of the archives, not by the people who filled them. No reporter at any of those titles was consulted about the price of their own back catalogue, and no reporter will see the money. What the licensing market prices is the training data. Labor is not a party to the transaction, and was never going to be, because labor has nothing the buyer needs.

Nearly every machine before this one was installed within reach of the people it displaced. The frame was in the cottage. The container crane stood on the dock the longshoreman had worked for thirty years. The model sits in a building the translator, the paralegal, and the junior illustrator have never seen and could not find, and there is no master to bargain with because there is no relationship to bargain over.

Hyundai's workers are among the last who can still make the argument the old way, by standing where the machine needs them not to be standing. They are using it to ask for five more years. Engels' pause ran fifty.