Cory Doctorow's 'The Claude Delusion' Tops Hacker News With 90 Points

Cory Doctorow published a new essay on his Pluralistic blog titled 'The Claude Delusion,' and it climbed to the front page of Hacker News, where it accumulated 90 points and 154 comments by the time it was logged. The piece, dated September 21, 2026, sits at the URL pluralistic.net/2026/09/21/sunsetting/, a path that signals the post's central concern. Doctorow has spent years writing about platform enshittification and the ways technology vendors degrade their own services after locking in users.

The headline borrows the name of Anthropic's Claude family of models, but the argument appears to be broader than any single chatbot. The word 'sunsetting' in the permalink points to a familiar Doctorow theme: companies that sell access to a product and then retire, throttle, or reshape it once customers have no easy exit. In his framing, AI assistants are a subscription like any other, and the people paying for them are renting capability rather than owning a tool.

The comment thread on Hacker News ran long, with 154 replies against 90 upvotes, a ratio that suggests the post provoked more argument than agreement. That pattern is common for Doctorow's work, which tends to draw both readers who share his critique of concentrated platform power and engineers who think he overstates the case against large language models. The discussion URL, item 49787765, keeps the full exchange archived alongside the original essay.

Anthropic's Claude line has become one of the most visible commercial AI products, competing with OpenAI's GPT models and Google's Gemini across enterprise and consumer tiers. If Doctorow's claim is that access to these systems can be withdrawn or downgraded at the vendor's discretion, that is a governance question as much as a technical one. Enterprises that build workflows on top of a hosted model have no guarantee the model, its pricing, or its terms will look the same in two years.

What the post does not settle, and what the thread likely spent its 154 comments chewing over, is whether the answer is regulation, open-weight alternatives, or simply treating AI subscriptions like any other rented software with a cancellation risk. Doctorow's track record suggests he will keep pressing the ownership argument. The next signal to watch is whether AI vendors respond to the criticism by committing to longer support windows for the models their customers depend on.