The ICLR rebuttal season brings a familiar tension into focus: the gap between what reviewers ask for and what a paper can reasonably deliver in a few days. One researcher, facing scores of 3/3/2, is wrestling with a short review that demands extensive new measurements. Their draft has grown long, and they worry about readability. That concern is more perceptive than they may realize.
This isn't just about one rebuttal. It reflects a deeper pattern in peer review that we have explored before. When reviewers focus heavily on novelty metrics, as discussed in When Theory Takes a Backseat: ICLR Reviewers Focus on Novelty, authors often feel pressured to add layers of new results rather than clarify what is already there. The result is a bloated rebuttal that buries the core contribution under fresh experiments. The researcher's instinct to polish and revise the paper itself is the smarter move. Promising additions for the camera-ready version signals responsiveness without sacrificing the rebuttal's focus.
There is a practical lesson here that applies well beyond ICLR. Reviewers evaluating a rebuttal are time-constrained and human. They will scan, not study. A document that reads like a miniature paper, dense with new tables and methodology, works against its own purpose. The goal is not to prove you can generate more data. It is to make the existing argument impossible to ignore. That means leading with the clearest version of your response, not the most exhaustive one. If a reviewer cannot grasp your main point within sixty seconds, the length has already cost you.
The researcher also asks whether others are making promises for the camera-ready version or revising the paper now. This question touches on strategy that our piece on Navigating Peer Review for Original Classical ML Research at Top Conferences addresses directly: committing to specific improvements for the final version is often more persuasive than attempting to preempt every critique in the rebuttal window. It shows you understand the gap and have a plan to close it, rather than pretending the gap does not exist.
The specific detail to watch here is how the community resolves this recurring tension between thoroughness and readability. If the norm shifts toward longer, more data-heavy rebuttals, the signal-to-noise ratio drops for everyone. If authors learn to be surgical, reviewers may reward clarity over volume. The researcher's draft is too long today, but their awareness of that problem puts them ahead of most. The next step is trusting that a tighter, more readable rebuttal is not a compromise. It is a competitive advantage.