The discussion focuses on GitLab’s new rate limits and their impact. Several commenters praise GraphQL as ideal for LLM-driven agents because it lets callers request only the fields they need, reducing token and quota use (bob1029, sockbot, mattkrick), while others warn GraphQL can still be expensive server-side and can trigger heavy backend work (philipp-gayret, iamEAP). People flagged the 60 requests/hour per IP rule for unauthenticated users as too low for schools, offices, or ordinary browsing (lateatdesk, rkagerer, MeetingsBrowser). Technical points about rate-limit behavior, Retry-After/429 responses, token-bucket semantics, caching, and the limits of client-side optimization are raised (mplanchard, sandeepkd, nijave), and one commenter asked for a test endpoint to validate client implementations (mschuster91).
Many attribute the policy change to LLM scraping and bot saturation and see it as a defensive move (bearjaws, sparkling, Jaxan). Proposals to pay or reward open-source authors for scraped traffic get cautious praise as potential alignment but also warnings that bots would game kickback schemes (cush, aprentic, latexr). Several voices fear this is the start of paywalls or usage-based pricing that will harm discoverability and open contribution (MeetingsBrowser, swatcoder, Frieren). Others criticized the announcement’s tone or suspected it was AI-written (demibabs, Jeremy1026, svachalek).
Summary generated by AI from the linked article. hn.today is not affiliated with Hacker News or Y Combinator.