ICLR submissions top 50,000 as researcher calls for per-author paper caps

burny_tech · x · 2026-09-20

Roei Herzig notes that every few years the field agrees peer review is broken, yet submissions keep doubling — ICLR just passed 50,000 papers. He proposes capping how many papers one author can submit per cycle, arguing nobody has more than 5 conference-worthy ideas in a single cycle. The number highlights the explosive growth of AI research output and the mounting strain on the reviewing system.

Related event: ICLR 2026 Submissions Top 50,000 as AI-Generated Papers Overwhelm Peer Review(12 posts)→

Original post →

More from AGI Musings

AGI Musings channel →