This post did some rounds, and that is nice. However, I noticed that anyone who commented on it seemed no only pro-"Wings" but couldn't even conceive of why the population would want "Shells." And yeah, I expect most tumblr readers are pretty far from the world of union halls and floor stewards and labor contract negotiations. What they were fighting for doesn't exactly resonate in 21st century America.
And that's my fault, because I let the essay be too much about labor union politics in particular, and not the vibes and spirit of the Shells. So here is an addendum that puts the Shells argument in more relatable terms for the people who might read this.
&&&
The preceding essay limits its argument to employment — to people who clock in, who have managers, who can be fired. This is understandable. Employment is where the data lives. But the binary it identifies (the fortress of workplace protections versus the trampoline of social insurance) is doing something much larger than adjudicating between French and Danish labour markets, and if you only look at employment you will miss it entirely.
The actual question is: where does dignity live? Does it live in the work, or does it live in the society? And if you phrase it that way, you notice immediately that the question extends past the factory floor, past the office, past the unemployment office and the retraining programme, into every domain where a human being's sense of themselves is bound up with a skill they spent years developing.
Consider the illustrator.
Not the famous one. The one who draws product packaging. She spent four years in art school, another six building a client base, and she's good — not transcendently good, not Dürer, but good in the way that a competent plumber is good, which is to say she solves problems reliably and her work doesn't leak. She earns enough to rent an apartment and eat meals she didn't cook from a packet. She is, by any historical standard, a member of the middle class.
In 2024 her income dropped by about forty percent, because a machine learned to draw product packaging less well than she does, but so cheaply that her clients looked at the gap between her invoice and the subscription fee and decided the quality difference wasn't worth it.
The shells-versus-wings framework handles this cleanly in theory. The shell response: protect the illustrator's market position. Require human authorship for commercial art. Regulate the training data. Build a wall around the job. The wings response: let the machine draw the packaging, tax the productivity gains, and make sure the illustrator has healthcare, housing support, retraining funds, and enough income to survive while she figures out what to do next.
And here is where the fortress deserves more credit than the original essay gives it. Because the fortress does something the trampoline cannot: it tells you that you matter specifically.
The International Longshoremen's Association spent 2024 demonstrating exactly how powerful a well-built fortress can be. Forty-five thousand dockworkers walked off the job at fourteen ports along the East and Gulf Coasts, shutting down roughly half the country's container traffic for three days, and what they won was extraordinary: a sixty-two percent wage increase over six years, full protections against automation, and contract language so restrictive that port operators need union sign-off before introducing semi-automated gate scanners. Ninety-nine percent of the membership voted to ratify. Harold Daggett, the ILA president, looked at E-ZPass lanes and self-checkout scanners and saw the same thing: a job killed so that a company could save money. His members carry signs reading "Machines Don't Have Families." They are not wrong.
And the thing to notice about the ILA is that the fortress is not merely preserving income. It is preserving a world. Longshoremen earn well (a third of New York-area dockworkers make over two hundred thousand dollars a year with overtime), but the money is downstream of something else: a community organised around a skill, with its own internal hierarchies, its own culture, its own sense of who matters and why. The union does not just negotiate wages. It maintains a structure of meaning. When Daggett says automation "destroys lives and livelihoods," the second word is doing work the first word alone cannot. A livelihood is a life that coheres around a specific way of being useful. The word carries more than income. It carries the structure.
A Danish unemployment office replacing ninety percent of your income does not replace this.
The West Coast longshoremen's union, the ILWU, accepted automation in its 2008 contract. The ports of Los Angeles and Long Beach are substantially more automated than their East Coast equivalents. In Rotterdam, when two fully automated terminals opened, the feared job losses didn't materialise — the dockworkers moved to air-conditioned control rooms and operated robotic cranes remotely. The work changed. The workers didn't vanish. What vanished was a particular version of the work, and a particular culture built around that version, and whether you think that culture's preservation is worth the cost depends on whether you think the culture is the point or the people are the point.
The ILA has drawn a line. The line is holding. For now, fifty thousand families on the East Coast have their fortress, their wages, their identity, their world. The ports they work at are, by most measures, less efficient than automated ports elsewhere. The goods that pass through them cost more to move. The economy absorbs this cost because the ILA has enough leverage to make the alternative more expensive. This is the fortress working exactly as designed: protecting the people inside at a cost borne by everyone outside.
And it will hold for six years. The contract runs to 2030. After that, another negotiation, another strike threat, another test of whether the leverage still exists. The fortress is as strong as its walls, and walls require maintenance, and maintenance requires power, and power requires that the people outside the fortress continue to find the cost acceptable.
The illustrator does not have a fortress. She has no union, no contract, no ability to shut down half the country's visual communications by walking off the job. She has a portfolio and a client list and a set of skills that a machine now does worse but cheaper. The shells-and-wings framework offers her two options: legal protection (copyright reform, training-data regulation, mandatory human-authorship requirements) or social insurance (retraining, income support, healthcare delinked from employment). What neither option addresses is the thing she's actually losing.
Every professional middle class in history has been a group of people whose skills were rare enough to command a premium and complex enough to require years of acquisition. Athenian rhetoricians. Medieval notaries. Victorian engineers. The American professional-managerial class that Barbara Ehrenreich spent her career anatomising. What makes them a class, as distinct from a collection of individuals who happen to earn similar incomes, is that the skill itself organises their social lives. They train together, they credential each other, they develop internal hierarchies of taste and competence, they marry each other. The skill is the skeleton and the income is the flesh. You can lose flesh and survive. Losing the skeleton is a different kind of event.
When a technology renders a skill abundant — not obsolete, abundant, which is worse — it does not simply eliminate jobs. It eliminates the scarcity that made the skill a basis for social organisation. Printing did not eliminate scribes. It eliminated the scarcity of literate reproduction that made scribes a coherent social group with internal status hierarchies and guild protections and a shared identity. Some individual scribes did fine. The class of scribes dissolved.
The Langovin word for a person's trade skill — one of those constructions that doesn't translate cleanly into Common — is closer to "the shape your hands have taken" than to anything in English occupational vocabulary. It carries the connotation of permanence, of a deformation that is also a formation. A potter's hands are shaped by clay. A scribe's hands are shaped by the reed. The skill is something you have become. And the loss of its necessity is something closer to an ontological problem: you have become a shape that the world no longer requires.
There is a version of the anti-AI argument that almost nobody makes because it sounds embarrassing. It is not "the machine produces inferior work" or "the machine was trained on stolen data." It is "I am a person who draws, and drawing is how I know I am a person, and the machine draws worse than I do and it took my job anyway, and if my skill can be replaced by something worse then my skill was never what I thought it was, and I cannot bear that." The argument is closer to a theological one than a policy one.
The shells-and-wings binary is, at bottom, two proposals for what to do after a dissolution. The shell says: prevent it. Regulate the technology, protect the guild, maintain the scarcity artificially. The wings say: allow it, but socialise the costs, catch the falling workers with benefits and retraining. The shell protects the job. The wings protect the person. The skill as a form of life — the thing that made the person a person — falls through both nets.
One of them is still better than the other.
The ILA's fortress is magnificent and it is also, by the logic of the original essay, a monument to a specific moment of leverage. Daggett can shut down the ports because container ships need to be unloaded at specific places by specific people and the physical choke points cannot be routed around. The illustrator cannot shut down anything. A songwriter cannot shut down anything. A junior lawyer who used to draft boilerplate contracts cannot shut down anything. The fortress works when you have leverage, and leverage requires scarcity, and scarcity is precisely the thing that technology destroys.
Which means, for the growing class of skilled workers whose skills are being made abundant rather than obsolete, the fortress is somewhere else — behind a wall they cannot reach. They can fight for it, and some of those fights are worth fighting and some will be won. But the structural trend is toward abundance, and the fortress's power depends on the structural maintenance of scarcity. You can hold the line for six years. You can hold it for twenty. At some point the cost of the wall exceeds what the people outside it will bear.
The wings model has a different problem, but it is a more honest one. It says: we cannot prevent the dissolution, so we will catch the people who fall. It does not pretend to save the skill. It saves the person. And if you are the person — if you are the illustrator watching a machine produce worse versions of your style for a twentieth of the price, if you are the translator watching neural networks close the gap that used to be your livelihood — the wings model's honesty is cold comfort. It is telling you that your hands have taken a shape the world no longer needs, and here is money, and here is retraining, and here is healthcare, and we are sorry about the hands.
But the alternative — the fortress — is telling you that it will keep the world needing your hands by force. And that works until it doesn't. And when it stops working, you have no wings, because you spent the decades inside the fortress instead of building a society that could catch you.
What happens to a society that solves the income problem but not the skill problem? You get a society with a very rich top, a very comfortable middle sustained by transfers, and no professional class. No group of people whose daily activity earns them a specific social identity. No illustrators, no session musicians, no translators, no one who is specifically good at a specific thing in a way that other people need. Just consumers of varying comfort levels, and a small class of people who own the machines.
The Rehn-Meidner model's "security by wings" assumed that the worker freed from the dying firm would fly to a living one. That there would always be a living firm to fly to. That the skills released from one context would find purchase in another. What happens when the context is not a specific firm but an entire category of skilled work? When the living firm doesn't need you because it has a subscription instead?
The original essay is right that the trampoline can tear. Sweden in the 1990s proved it. But the fortress can become irrelevant, which is worse than tearing, because a torn trampoline is a political crisis that generates energy for repair, while an irrelevant fortress is just a ruin that the world has quietly walked around. The ILA is powerful now. The Luddites were powerful once. The scribes' guilds were powerful once. The question worth asking is whether the thing the fortress is holding can survive the next fifty years of technological acceleration, and the honest answer is probably not.
So you build the trampoline. You build it knowing that it does not solve the skill problem. You build it knowing that a society of comfortable people who do not do anything in particular is a society with a hole in it where meaning used to be. You build it because the alternative is a fortress that protects fewer and fewer people at greater and greater cost while the world outside it reorganises without them.
And then you try to solve the skill problem separately, with different tools, in a different conversation. Because that problem — the problem of what people are for when the things they learned to do can be done without them — is older than economics, older than policy. It has been asked in every language humans have spoken, and it has never been answered, and the fact that it is being asked again now, by illustrators and longshoremen and translators and lawyers, does not mean that this time we will answer it. It means that this time, like every time, we will have to live inside the question.
