A leading AI research conference will only review your paper if you help review someone else's

The ACL Rolling Review is preparing to require every paper to come with a qualified reviewer. Each author would be capped at 20 total submissions and 5 first-author (the lead author, named first in the byline) papers per cycle. Any submission without a registered reviewer would be put into a lottery for whatever spare capacity remains. The change is the centerpiece of a new "Sustainable Reviewing Policy" the ACL Rolling Review announced this week.

ACL, the Association for Computational Linguistics, runs the leading annual conference for natural-language-processing research and a family of co-located venues. The Rolling Review, or ARR, is the year-round review pool that feeds them. Under the new policy, each submission must "pay" for itself by supplying a qualified service contributor, either a reviewer or a chair. A qualified reviewer, per the official @ReviewAcl account, is someone with at least three papers in the ACL Anthology, the public archive of the field's accepted work. If no contributor is attached, the paper goes into a lottery for whatever reviewing capacity is left over after paired submissions have been matched.

The policy arrives after years of warning signs that the field's reviewer bench is being outrun by its submission volume. Major AI and natural-language-processing conferences have spent several cycles stretching reviewer pools, expanding rosters of area chairs (mid-tier reviewers who oversee clusters of papers), and shrinking the time reviewers have to respond. The Rolling Review already requires authors to commit to reviewing as part of its annual reviewing-workload requirement and has been layering reviewer-volunteering and incentive changes into the 2025 and 2026 tracks. The May 2025 Peer Review Committee Report on incentives framed reciprocity as a precondition for the system to keep working.

A small number of researchers and labs account for a disproportionate share of submissions at major NLP venues, and the 20-paper, 5-first-author ceiling targets that concentration directly. Papers without a registered service contributor are not guaranteed a review; they are entered into a capacity lottery that may yield nothing. The policy also opens a path for designated non-author contributors, who can vouch for a paper in a manner the post compares to arXiv's endorsement system, where an established researcher must sign off before a paper can appear. A mentorship track is promised for contributors who are not yet qualified to review. Accounts that repeatedly submit or endorse low-quality work, or otherwise abuse the system, are flagged for penalties or bans.

Peer review is unpaid, distributed labor, and that labor is unevenly distributed across institutions. The new model favors authors and labs that already have a bench of qualified reviewers in their networks, and it places a higher barrier in front of early-career, independent, and globally under-resourced researchers who do not. The Reddit thread surfacing the policy captured one researcher's reaction that the change is "a bit of gatekeeping"; the small comment set underneath is closer to a starting point for a discussion than a verdict. Both readings, the policy as a community's attempt to price the hidden labor of peer review into publishing, and the policy as a way to concentrate publishing access in already-resourced hands, are supportable from the announcement itself.

Two things are still unresolved. The X post frames the policy as a proposal and promises an update to the official ARR site "shortly"; any claim that the 20-paper cap or the lottery is final should wait for that page to land. The definition of a qualified reviewer (three ACL Anthology papers) is taken from a separate standing @ReviewAcl guidance tweet, not from the policy post itself, and the enforcement language for "abuse" is loose. The next concrete signals to watch are the official ARR site update and the operational guidelines in the acguidelines.md repository, which will determine how the lottery is run and how the cap is enforced when submissions open for the next cycle.