The ICML 2027 call for papers reads like an admission that the conference review system has reached its breaking point. We think the proposed two-stage review process is a sensible, if imperfect, attempt to address the submission explosion without sacrificing rigor. The key innovation is not the first-stage cut itself, but the checks and balances around it, requiring senior area chair confirmation for any proposed rejection and providing authors with a short justification. This is a far cry from the opaque desk rejections we have seen at other venues, and it deserves a closer look.
The most practical change for researchers is the reciprocal reviewer requirement. Tying authorship volume to reviewing duty is a direct response to the systemic imbalance we have navigated in an era of expanding research teams. Under the new rules, each author can be on at most one paper with no eligible reciprocal reviewer among its coauthors. This closes the loophole where senior researchers add their names to papers with interns or undergraduates who cannot serve as reviewers, offloading the burden onto the committee. If you have three or more submissions, you must serve on the program committee. The math is simple: if you want to submit, you must review. We have previously examined why top AI conferences reject papers before reading them, and this reciprocal requirement addresses the root cause rather than just the symptom.
Our primary concern mirrors the one raised in the community: a single area chair reading a paper quickly could miss a novel contribution. The senior area chair override helps, but it is not a guarantee. The February 23 justification deadline is a double-edged sword, it provides transparency, but a short rejection notice does little to guide authors toward improvement. We also wonder how the 20-submission cap per person will interact with the new CVPR 2027 author enrollment form requirements, as conferences increasingly shift administrative burdens onto authors to manage volume. The specific detail to watch is how the reciprocal reviewer rule affects large, interdisciplinary collaborations where only one author has the expertise to review for ICML. That single "free pass" per author is generous, but it will force teams to think strategically about who gets listed on each paper.