ICML Review Process and Guidelines

The crack appeared on April 8th with a question that sounds procedural until you sit with what it actually means: how are ICML 2026 reviewers submitting evaluations without completing the mandatory acknowledgement step? That acknowledgement is not bureaucratic decoration. It is the moment a reviewer formally confirms they have read the conference guidelines before sitting in judgment on someone else's year of work. Whether the cause is a platform failure or deliberate corner-cutting, the outcome is the same — researchers received evaluations from people who may not have agreed to the standards governing those evaluations. By April 14th the conversation had widened into a broader question about area chair guidance, with a community-run "qq thread" filling the vacuum left by unclear official communication. Then on April 16th something stranger surfaced: scores that had increased and then decreased, with no obvious explanation attached. A reviewer changing their mind is normal. Scores moving without explanation or visible discussion is a different thing entirely. The frustration then jumped conferences — an ECCV thread in early May documented what one poster called "stupid reviewer behavior," and by May 19th another ECCV-specific complaint had emerged about reviews showing no modification timestamps, making it impossible to tell whether a reviewer had engaged with author rebuttals at all. Two conferences, the same pattern, the same community doing the monitoring that the conferences themselves appear not to be doing publicly.

If confirmed, here is what this means. The academic publishing infrastructure that determines careers, funding, and the direction of research fields is operating with accountability gaps that the community is patching together through Reddit threads. That is not a small problem dressed in procedural language. Researchers who receive contradictory or unexplained scores have no formal mechanism to understand what happened to their submission, and area chairs operating without clear published guidance are making consequential decisions in an informal fog. The second-order effect is corrosive: when the process feels arbitrary, the most experienced researchers — the ones with options — start routing their best work elsewhere, toward workshops, preprints, or rival venues. The conferences that fail to demonstrate process integrity do not collapse immediately. They hollow out slowly, retaining the brand while losing the substance.

Watch for any official ICML or ECCV communication addressing reviewer compliance or score-change transparency — that would either confirm the community concerns or reframe them entirely. Equally telling would be silence: if neither conference acknowledges these threads, that itself becomes part of the story.

NewsHive monitors these sources continuously. All signal titles above link to the original reporting.

Intelligence by NewsHive. Need help navigating what this means for your business? Contact GeekyBee →