Bryan Cantrill writes from the reader’s side of the page: too many people are putting their name to pieces that are clearly LLM-authored, and he has two pointed questions for them. Do you think readers cannot tell? Or do you think readers do not care?

His answer to the first is that anyone who reads broadly can tell immediately — “the hand of the LLM is so clear it’s as if the writer’s intellectual fly is open.” The second question is where the argument gets teeth.

  • Of 668 developers in Cynthia Dunlop’s survey, 78% said they “stop reading immediately” upon detecting an LLM
  • 71% also said they “avoid the author in the future” — a cost that outlives the piece
  • 98% preferred an author’s own imperfectly written piece over an LLM-polished one, which is a preference for authenticity rather than polish
  • These are not fringe respondents: active readers on social media are the people who repost and promote writing

The deeper harm is epistemic. When a piece has had substantial LLM assistance, readers no longer know what is real and what is not. Cantrill’s formulation, from Oxide’s RFD 576, is that using an LLM to write voids the social contract between writer and reader — readers should not have to labour to understand a sentence the writer did not work to create.

His strongest move is refusing to rest on taste. For a public piece, the only purpose is to serve the reader, so if readers abandon you, you have actively repelled the audience you were trying to attract. Using an LLM to write therefore becomes, in his word, simply ineffective.

He reaches for the spam wars as the analogy. Spam was supposed to kill email in the early 2000s; instead filtering improved until the economics collapsed, and today being labelled spam can destroy a domain and a brand. Cantrill asks whether LLM-authored text has reached the same inflection point — and explains why the answer depends on detection accuracy, and why he spent a long time hunting for a tool with an acceptable false-positive rate before Oxide made Pangram-clean public writing a policy requirement.

The prescription is narrower than a ban. Do not have an LLM write your piece, but do use it as an editor — LLMs are, he grants, superlative editors. Or, if you genuinely believe prose generated from your prompt is fine, spare everyone the wasted cycles and just publish the prompt.