Quoting Kyle Kingsbury
Kyle Kingsbury, writing in Jepsen's results for ParticleDB 1.1.4, raises a concern about how large language models affect confidence in software and its documentation. His argument, as quoted by Simon Willison, is that LLM use weakens trust not only in docs and code but in the organizations that produce them, since the markers engineers once read as evidence of a mature system have become easier to fake.
Kingsbury frames the problem as a pair of practical questions for maintainers: what can they do to keep software correct and documentation accurate in the LLM era, and once that work is done, how can they show users it was done? The quote appears in Jepsen's analysis of ParticleDB version 1.1.4.
This suggests a shift in how developers evaluate projects. Signals that were cheap to trust, such as polished prose or confident-looking examples, no longer reliably indicate care. A likely trade-off is that maintainers will lean more on verifiable evidence, like reproducible tests, independent audits or Jepsen-style fault-injection results, to distinguish real quality from generated polish.
Why it matters: maintainers and product teams that ship docs or code alongside LLM output face growing skepticism from engineers who read it, and may need to make verification and provenance explicit. For developers choosing dependencies, the practical effect, in my reading, is more weight on tests, audits and reputation than on presentation.
Based on reporting from the original publisher. Visit the source for full context and later updates.
Publisher excerpt
More broadly, the use of LLMs erodes trust in documentation, in software, and in the organizations which produce them. Perhaps engineers will begin to treat any new software project with suspicion—the signals we used to rely on as markers of a mature system are now easier to counterfeit. [...] In the age of LLMs, what can maintainers do to ensure correct software and accurate documentation? Having done that work, how can they demonstrate its quality to users? — Kyle Kingsbury , Jepsen results for ParticleDB 1