The calendar is a cruel taskmaster for researchers this year. When the author notifications for NeurIPS 2026 land on September 24th, and the ICLR deadline follows on the 25th, the community is left with a choice that feels less like strategic planning and more like a fire drill. The original poster's frustration is valid, especially when five of six reviewers ignored their rebuttals. That silence is not just annoying; it undermines the entire purpose of the discussion phase. If the conversation is supposed to refine our work, what happens when half the room refuses to speak?
We see this as a symptom of a broader tension in how we evaluate research. The pressure to have a backup submission is real, but it creates a perverse incentive. You are not improving your paper for the sake of science; you are polishing it to survive a rejection. This is where our take diverges from the "hustle culture" of academia. We would tell that researcher, and anyone else in this bind, to resist the urge to treat ICLR as a safety net. Instead, use the 24 hours between the notification and the deadline to have an honest conversation with your co-authors. If the reviews were unaddressed, that is a signal about the process, not necessarily the quality of your work. A rushed submission to ICLR, built on the heels of a NeurIPS rejection, rarely produces your best thinking.
This connects directly to the challenge of Clean Data Starts With Catching AI Slop Before It Skews Your Model, where noise in the input corrupts the output. Here, the noise is the compressed timeline and the unresponsive reviewers. Just as a sentiment model suffers when you filter out genuine reviews, a paper suffers when you prioritize meeting a deadline over incorporating substantive feedback. You are making a trade-off between speed and depth. The real question is whether you are building a research program or a publication list. If it is the former, then skipping this ICLR cycle is a legitimate choice. If it is the latter, you are playing a game where the rules are rigged against thoughtful revision.
We also see a parallel in Exploring Real-World Computer Vision: Deployments, Edge Models, and Current Challenges, where practical constraints often dictate theoretical elegance. In that post, the focus is on optimizing for mobile phones, not just maximizing accuracy on a benchmark. Similarly, the constraint here is not the quality of your ideas but the logistics of the review process. You have to decide if you are optimizing for the ideal outcome or the realistic one. Our honest take is that the community needs to push back on this scheduling crunch. It is not normal for the discussion phase to be this long, and it is not healthy for the AC and reviewer phases to feel like a black box. If you are preparing a backup, do it because you have genuine new insights, not because the calendar forced your hand.
The concrete point to watch is the tone of the notifications themselves. If reviewers who ignored your rebuttals suddenly offer detailed feedback in the final decision, that tells you the process is performative. If they remain silent, you have learned something valuable about where to invest your energy. We would advise you to draft your ICLR submission now, but not to submit it. Let it sit. The most empowering move is to give yourself permission to skip a cycle when the conditions are not ripe. The research will still be there in December. The question is whether you will be ready to defend it on your own terms, not on a schedule set by a conference deadline.