ICLR 2027 Caps Authors at 20 Papers; DeepMind Researcher Responds With Irony

The organizers of ICLR 2027 have set a new rule: no author may be listed as a co-author on more than 20 papers in a single cycle. In response, a Google DeepMind researcher launched a satirical petition, turning a technical regulation into a public discussion about co-authorship culture in the academic AI community.

How the new ICLR limit works

ICLR 2027 introduces a hard quota: a maximum of 20 papers per author per conference cycle. The mechanism has two stages. At the abstract deadline, authors who exceed the limit receive an automatic warning. If the quota is still violated by the full paper deadline, the organizers begin rejecting papers at random until the author's count fits within 20.

"If by the full paper deadline an author is still listed as a

co-author on more than 20 papers, the conference will randomly reject some of the papers until the quota is met," the ICLR 2027 rules state.

Key parameters of the new policy:

• Quota: no more than 20 papers per author per ICLR 2027 cycle
• Warning: issued automatically at the abstract deadline
• Penalty: random rejection of papers if the violation persists at the full paper deadline
• Scope: any position in the author list, not just first or last author

ICLR 2027 is the first of the leading AI conferences to introduce a hard numerical ceiling on co-authorship. Submissions to NeurIPS, ICML and ICLR grow every year, and so does the number of researchers at large labs who are listed as co-authors on dozens of papers in a single cycle.

Why the DeepMind researcher responded with irony

The Google DeepMind researcher answered the rule not with an objection but with a satirical petition. According to Jiqizhixin, the document is written in a satirical vein: it does not demand that the limit be scrapped, but mocks the prevailing co-authorship culture. The number 20, as an official ceiling, sounds enormous in itself, and that is precisely the point of the joke.

The question of co-authors' real contributions has long been on the academic AI community's agenda. When one person is listed on 30–50 papers in a single conference cycle, a natural question arises: how deeply is that person involved in each of them? The status of "senior author" or "corresponding author" does not always mean a substantial intellectual contribution; sometimes it amounts to a formal presence in the author chain. The academic community has discussed "hyperauthorship" for several years, but in AI it has become especially pronounced as large labs with hundreds of employees have grown.

What this means

ICLR 2027 sets a new precedent for academic AI conferences. If the limit proves effective, NeurIPS, ICML and other venues may adopt similar restrictions. For most authors the change will go unnoticed; for "hyperproductive" co-authors at leading labs, it is a direct signal to reconsider how they participate in papers.

Frequently asked questions

What happens if an author exceeds the 20-paper limit at ICLR 2027?

The organizers will automatically send a warning at the abstract deadline. If the quota is still violated by the full paper deadline, some of the author's papers will be rejected at random until their count drops to 20 or below.

Why 20 papers specifically, and not fewer?

The ICLR organizers have not publicly explained the choice of this particular threshold. The DeepMind researcher's satirical petition effectively asks the same question: if 20 is already the ceiling, does that mean 19 papers in a single cycle counts as an acceptable norm?