AI-generated volume is straining review, code and content alike

"How the Claude Code team does code review" (50122951) got a pile-on. Commenters tied it to a GitHub repo with 5000 issues, about a third of them random regressions of earlier behavior. Another reacted to the figure of 30 thousand changes merged per day with a flat "no, just no." The review process is the story, and people are not convinced it holds up at that speed.

The same worry hit content. An Ask HN asked for a flag for AI-generated articles (48886741). Commenters said voters often cannot spot generated text, detectors are unreliable, and HN's new guideline against posting generated text covers comments but not obviously submissions. A Cliff Stoll thread (48037336) about being declared dead by AI slop adds to it, and so does Anthropic cutting Claude Code subscriptions off from third-party harness OpenClaw because such tools put "outsized strain" on its systems (47633396). Users countered that they were just using the capacity they paid for.

The pattern: generation is cheap, and verification, moderation and capacity are now the costly part. Quality control has become the scarce resource in code, content and compute.