The rejection email lands, and the immediate reflex is to salvage the work, to find a new home for it before the ink dries. The Reddit thread captures this moment perfectly, asking the question that lingers in the back of every researcher's mind: how much of that reviewer feedback do you actually honor? The honest answer, as the comments suggest, is nuanced. You are not a vending machine for revisions. You are the author, the one who understands the paper's core contribution, and that means you have the right, and the responsibility, to filter every critique through your own judgment. This is not about being defensive; it is about being strategic with your limited time and energy, especially when the ICLR deadline is looming.
The thread's focus on novelty and significance critiques is where this gets interesting. When a reviewer says "the contribution is incremental," they are not always wrong, but they are also not always right. Sometimes, they are reacting to a framing issue rather than a fundamental flaw in the work. They might be missing the forest for the trees, or they might be applying a standard so high that even a solid, incremental step forward looks small. The pressure of a close deadline amplifies this tension. You can either spend your time trying to add more experiments to satisfy a skeptical reviewer, or you can spend it sharpening the narrative, making the existing contribution clearer and more compelling. The former is a gamble that may not pay off; the latter is almost always a wise investment. We have seen this dynamic play out in other contexts, such as the discussion around NeurIPS Acceptance Raises Questions About AI Review Justifications, where the disconnect between scores and final decisions highlights the subjective nature of the process.
The most practical takeaway here is that you should treat reviewer feedback as a menu, not a mandate. You are allowed to decline the dishes that do not appeal to your taste. If a suggestion would take the paper in a direction that you believe is wrong, or if it would require you to compromise on the core narrative, then you should not implement it. The goal is not to create a paper that pleases every reviewer; it is to create a paper that is a stronger, more honest version of your original idea. This selective approach is not about arrogance. It is about curation. It is about understanding that your time is finite and that you should spend it on the changes that will have the most impact. As you navigate this process, remember that you are not alone in this struggle. Many researchers are exploring similar challenges in their own work, as seen in practical applications of Exploring Real-World Computer Vision: Deployments, Edge Models, and Current Challenges, where the gap between theory and practice is a constant theme.
The real skill is in the triage. You have to read a critique and ask yourself a simple question: "Does this reviewer understand what I am trying to do?" If the answer is no, then their suggestion is less useful. If the answer is yes, then you should listen more carefully. This is not about being dismissive; it is about being efficient. The pressure of the deadline forces you to prioritize. It forces you to focus on the changes that will most improve the paper, not just the ones that are easiest to make. The final paper should be a reflection of your best work, not a patchwork of other people's opinions. The most important question you can ask yourself is not "What did the reviewers say?" but "What do I need to change to make this paper the best version of itself?" And that is a question only you can answer. The next time you sit down to revise, consider that the most valuable feedback might be the feedback you choose to ignore.