Can Peer Review Keep Up with Super‑Prolific Authors Publishing 100 Papers a Year?

In 2018 Ioannidis and colleagues introduced the term “hyperprolific author” for researchers who publish more than 72 papers per year, roughly one paper every five days. They identified over 9,000 such names in Scopus, a group that grew 2.5‑fold from 2001 to 2016.

Five years later Gemma Conroy reported that the 2023 update from Ioannidis’s team showed a four‑fold increase in authors publishing over 60 papers annually, with some fields now seeing researchers publishing 50, 100, or more papers each year.

A dedicated study of AI conference trends (2014‑2023) covering NeurIPS, AAAI, ICML, ICLR, IJCAI, CVPR, ICCV, EMNLP, ACL, KDD, and ACM CHI counted 87,137 papers and found that in 2023 NeurIPS and in 2024 CVPR each had more than 250 authors who published over five papers at a single conference in one year; CVPR even recorded individual authors with more than 20 papers in a single year.

The only major conference that limits single‑author output is IJCAI, which imposes a strict cap on the number of papers a single author can submit per year.

Two forces explain the AI‑era surge: large‑scale collaborations required for big‑model training, benchmark suites, and cross‑institution labs, and the incentive pressure described by Goodhart’s law—when publication count becomes a hard metric for hiring and promotion, the metric itself ceases to be a good indicator.

Submission volumes have exploded: ICML 2026 received 23,918 valid papers (up from 12,107 in 2025) with a 26.56 % acceptance rate; NeurIPS 2025 saw 21,575 submissions, and AAAI‑26 attracted nearly 29,000 papers from over 75,000 authors. In response, many top conferences introduced author‑level caps ranging from 7 to 25 papers per year, discarding excess submissions by order of submission number.

Reviewer scarcity has led some reviewers to rely on large language models (LLMs). ICML disclosed that 497 papers were withdrawn in 2023 for violating its LLM‑use policy after watermark detection flagged them, illustrating how AI tools are being used on the review side.

Journal publishing has also expanded dramatically: open‑access publishers such as MDPI grew from 14 journals in 2000 to 487 today, and Frontiers Media to over 220. This proliferation fuels “paper‑mill” operations; a recent study found 2,100 retractions in the past two years explicitly linked to AI‑generated content and another 2,300 retractions tied to paper‑mill activity.

Retraction Watch’s co‑founder Ivan Oransky warns that AI‑based screening cannot solve the overload, arguing that the only real remedy is to reduce the sheer volume of papers requiring peer review. Midway’s proposed remedy is to recognize and financially compensate peer‑review labor, thereby aligning incentives with the true value of reviewing.