Hello everyone! A quick FYI: as I just posted in my fellow report for last week, I created and merged a PR to (try to) disable automated reviews in the Django repo, and also updated the contributing docs accordingly. This was done in agreement with the Triage and Review team.
I’m posting here to raise awareness about two things:
- this is now in effect in the repo [1], and
- automated and unreviewed content posted directly to pull requests or tickets is generally not helpful.
Expanding on the second point: this is not a stance against automation or LLM-assisted work in general. Some contributors use tools locally to explore ideas or draft changes, and that is acceptable. The issue arises when automated reviews or LLM-generated feedback are posted directly to pull requests or in Trac tickets. In practice, these tend to add noise and increase the review burden on maintainers and contributors who must sort through them.
Django’s development process intentionally prioritizes careful review, long-term stability, and well-understood changes over perceived velocity. Automated comments can create a sense of activity without necessarily reflecting the project’s conventions, historical context, or design trade-offs.
For that reason, meaningful human review remains essential to maintaining the framework’s quality and reliability, and to avoiding [the accumulation of] comprehension debt [2].
See example of a nullified comment at Fixed #36784 -- Enabled Script and Media classes to render custom attrs. by nessita · Pull Request #40 · nessita/django · GitHub ↩︎
Thank you @nanorepublica for teaching me the term ↩︎