The Customer Rage Index Goes Up, and Cory Doctorow Reaches for Stephen King

There is a particular genre of tech writing that has hardened into something close to a movement over the past three years, and Cory Doctorow's Pluralistic sits at its center. The word he coined, "enshittification," has escaped the blog and entered ordinary speech. His latest post, "The world has moved on," takes the framework somewhere darker, borrowing the elegiac language of Stephen King's Dark Tower novels to argue that the rot in modern products and services is not a vibe but an end-state. It is worth reading, and it is worth pushing on.

The spine of the piece is not literary, though. It is a data point. The 2025 National Customer Rage Study, which has polled a representative panel of around 1,000 American consumers every three years since the mid-1970s, reports that people are encountering more problems with the things they buy, that those problems are more severe, and that more of us now actively want revenge on the businesses that wrong us. The Guardian's Heather Timmons used the study to anchor a new investigative series on the general feeling that everything has gotten worse, and the reader response was large enough that it became a story in itself.

The trend is real, and the survey is the strongest evidence for it

Start with what is solid. A longitudinal survey running across five decades is rare and valuable precisely because it controls for the thing skeptics raise first: maybe people always thought the present was worse. The Douglas Adams line Doctorow opens with captures the suspicion neatly. Anything invented after you turn 35 feels like an affront to the natural order. If the data simply tracked the aging of the panel, you could dismiss the whole thing as nostalgia wearing a lab coat.

But a fresh sample every three years is not the same respondents getting older together. The trend line points up across cohorts, which is the part that should give even a committed skeptic pause. Something measurable is happening to the experience of being a customer, and it is not reducible to one columnist getting cranky in late middle age. Doctorow, to his credit, says as much, and explicitly disowns the "old man yells at cloud" reading of his own work.

Where the AI angle is genuinely sharp

The most useful new idea in the essay is not about antitrust. It is about bureaucracy, and it deserves to be lifted out and examined on its own.

Drawing on David Graeber's The Utopia of Rules, Doctorow points out that paperwork has an asymmetry problem. In an analog office, a form costs the institution something to process. That cost acts as a brake. Nobody invents a fourteen-page form lightly, because someone on the institution's side has to read all fourteen pages. The web already broke this symmetry, since a form is cheap to deploy and the burden falls almost entirely on the person filling it out.

His claim is that generative AI removes the last bit of friction. With "vibe coding," meaning the practice of prompting an LLM to scaffold an application rather than writing it by hand, any mid-level manager can spin up a form plus its processing and analytics back-end in an afternoon. The result, he argues, is that a task that once needed one short form now spawns ten long ones, each built by a different person asking for slight variations on the same information. He calls it "Kafka-as-a-service," the democratization of bureaucracy.

This is a real mechanism, and it connects to something practitioners already complain about. When the marginal cost of producing a process drops to near zero, the volume of processes explodes, regardless of whether any of them are good. The counterpoint developers will raise is that vibe-coded back-ends are notoriously brittle, and a form whose processing script silently corrupts your data is worse than no form. Doctorow agrees, and treats that brittleness as part of the harm rather than a reason to doubt it. The unanswered question is whether better tooling closes the gap. The HTML-first argument he links to suggests that thoughtful, fast, respectful forms are entirely buildable. They just require the form-giver to do real work once instead of offloading it onto everyone else forever. That is an incentive problem, not a technology problem, which complicates the story of AI as the villain.

The part where consensus is doing a lot of work

Here is where a reader who admires the reporting can still get off the train. Doctorow's explanatory frame is total. The customer rage survey, the canceled flights with no gate agent, the chatbot that emails an out-of-date PDF, the John Deere products that suddenly degrade, all of it traces back through a single causal chain to the dismantling of antitrust enforcement, and from there to what he describes as a conservative project to install a hereditary aristocracy. He cites Corey Robin's The Reactionary Mind and Frank Wilhoit's much-quoted line about in-groups the law protects but does not bind.

This is a coherent worldview, and large parts of it are well supported. Antitrust enforcement did weaken substantially from the 1980s onward, the Bork-era "consumer welfare standard" did narrow what counted as illegal, and market concentration in many sectors is documented fact. The reporting on understaffing as a deliberate transfer of value from customers to owners, and on the destruction of "process knowledge" when experienced workers are replaced by contractors and bots, lands because it is specific.

The weaker move is the totalizing one. When every grievance in modern life, from healthcare denials to airport lines to immigration forms, gets folded into one ideological origin story, the framework starts to explain too much to be falsifiable. Some product degradation is monopoly power. Some is the ordinary cost-cutting of firms in competitive but low-margin businesses. Some is the genuine difficulty of running complex systems at scale. A reader who wants to understand why their bank's app is bad is not always helped by being told it is downstream of a fifty-year war on the rule of law. The categories deserve to be kept separate, if only so the strong claims do not get dragged down by the loose ones.

There is also a sentiment question the essay sidesteps. Customer rage is partly about products and partly about expectations, which technology itself keeps ratcheting up. People angry that a chatbot cannot resolve a billing error are angry at a baseline of instant resolution that did not exist when the survey began in 1976. That does not make the anger wrong. It does mean the rage index measures the gap between expectation and reality, and both halves of that gap are moving.

The prescription, and its honesty

What sets the piece apart from most enshittification commentary is the refusal to offer consumer advice. Doctorow flatly says you cannot shop your way out of a monopoly, that shopping is not politics, and that the only real levers are collective: joining a union, supporting groups like the Electronic Frontier Foundation, getting involved in the unglamorous local politics that movements like Moms for Liberty have used to real effect. For tech workers specifically he points to Tech Solidarity, the Tech Workers Coalition, and in the UK the United Tech and Allied Workers.

This is the most defensible part of the argument, because it follows from the diagnosis rather than undercutting it. If the problem is structural, individual virtue cannot fix it, and any honest writer has to resist the urge to end on a tidy list of apps that respect you. Whether you accept the full Dark Tower cosmology or read it as one writer's metaphor for institutional decay, the call to organize rather than to optimize your purchasing is the rare conclusion that does not flatter the reader into thinking the problem is theirs alone to solve.

The essay arrives as Doctorow tours for The Reverse-Centaur's Guide to AI, out this month from Farrar, Straus and Giroux, and ahead of The Post-American Internet, a sequel to last year's Enshittification. The framework is clearly hardening into a multi-book project, which is its own kind of risk. Movements built around a single explanatory word tend to get better at applying the word and worse at testing it. The customer rage data is a gift to anyone making this argument, because it is external, longitudinal, and hard to wave away. The discipline worth keeping is to let that data say what it says, which is that the experience of being a customer is measurably worse, and to treat the grander claim about why as a hypothesis still being argued rather than a verdict already in.