--- name: brycewang-stanford/acl-review-process source: https://app.decimal.ai/s/brycewang-stanford-acl-review-process@1/SKILL.md source_sha256: f229c32542b1 --- # ACL Review Process Use this to model where an ACL paper actually is and who decides what next. ACL decoupled reviewing from the conference: ARR produces the review package; the conference consumes it. Confirm current mechanics at aclrollingreview.org/reviewing, since cycle structure has changed repeatedly (most recently to 10-week cycles in 2025). ## The pipeline, stage by stage 1. **ARR cycle intake**: submission on OpenReview to a dated cycle (e.g., ARR January 2026); checks for format, Limitations, checklist, anonymity. 2. **Review**: at least three reviewers score and write reviews. 3. **Author response + discussion**: authors reply; reviewers discuss and may update reviews (see `acl-author-response`). 4. **Meta-review**: the action editor / area chair synthesizes reviews into a meta-review with a recommendation. Notably, ARR senior area chairs do *not* make accept/reject recommendations — that authority sits with the venue. 5. **Commitment**: authors choose whether and where to send the finished package — ACL, or another \*ACL venue accepting that cycle. 6. **Venue decision**: the conference's senior area chairs and program chairs read reviews + meta-review and decide: main conference, Findings, or reject. 7. **Ethics branch**: papers flagged during review go to an ethics committee, which can require revisions or recommend rejection on ACL Code of Ethics grounds independent of technical merit. ## Who decides what | Actor | Controls | Cannot do | |---|---|---| | Reviewer | Review text, scores, discussion updates | Accept or reject anything | | AC / action editor | Meta-review, review-issue handling | Bind the conference's decision | | Author | Response, revision, where/whether to commit | Change reviews after the cycle closes | | Venue SACs + PCs | Main/Findings/reject from the ARR package | Re-run the reviewing itself | | Ethics committee | Ethics-based conditions or rejection | Judge technical contribution | ## Outcomes and what they mean - **Main conference**: full acceptance with a presentation slot. - **Findings of ACL**: peer-reviewed, Anthology-published acceptance for solid work that missed the main-program bar; typically without a main-stage oral. Citation-wise Findings papers perform respectably — treat it as a real publication decision, not a consolation to appeal. - **Reject at commitment**: the ARR reviews remain attached to the paper's history; a stronger resubmission through a later cycle is the designed path. ## Reading your review package strategically - Weigh the meta-review over individual scores: it is the document venue chairs actually read. - A package with one detailed negative review and a meta-review that endorses the objection rarely survives commitment — fix the objection in a new cycle instead of shopping the package to another venue unchanged. - Conversely, mediocre scores with a meta-review noting resolved concerns can be worth committing: the venue decision is made on the whole record. - A meta-review score at the floor triggers the resubmission constraint: within six months, only a wholesale revision avoids desk rejection. ## Venue-shopping within the ARR ecosystem The same review package can be committed to exactly one venue at a time, but a rejected or withheld package can target a later conference (EMNLP, EACL, NAACL, AACL) whose commitment window accepts that cycle. Sequence deliberately: each conference publishes which ARR cycles it accepts, and reviews age — a package built on a two-cycle-old baseline set gets punished at commitment. ## Anatomy of one 10-week cycle ```text week 0 submission deadline (AoE) weeks 1-2 desk checks; reviewer + AC assignment weeks 3-5 reviewing week 5 reviews released -> author response (a few days) week 6 reviewer discussion; reviews may be updated weeks 7-8 meta-review written ~week 10 package complete -> commitment window for eligible venues ``` Offsets drift by cycle; the January 2026 cycle ran Jan 5 → reviews by early March → ACL 2026 commitment March 14. Always re-derive from the current dates page. ## Reading reviewer language for what it predicts - "The contribution is unclear" from two reviewers predicts a meta-review hedge — fix framing in the response, not just individual complaints. - "Missing citation X, Y, Z" alone rarely sinks a paper; the same list attached to "novelty over X is marginal" usually does. - A confident low-score review with factual errors is the best target for a review-issue flag plus a surgical rebuttal. - Uniform middling scores with short reviews often mean poor reviewer fit; at commitment, venue chairs weigh the meta-review's own reading more — make sure the response arms the AC with the paper's best case. ## Confidentiality and conduct boundaries - Review content is confidential to the cycle; quoting reviews publicly (even anonymized) before decisions violates ARR expectations. - Contacting reviewers or guessing identities is out of bounds; all communication flows through the OpenReview thread and the AC. - Authors are bound by the ACL code of ethics throughout — including in how the author response characterizes reviewers' competence. - Deadlines inside the cycle (response, discussion) are short and rarely extended; plan around them rather than requesting exceptions. ## Output format ```text [Stage] cycle-intake / under review / response / meta-review / commit-window / venue decision / ethics review [Package strength] [Commit recommendation] commit to ACL / hold for revision / target sibling venue [Ethics exposure] [Next actor to influence] ```