When you disagree

Sometimes a reviewer is wrong, or asks for something unreasonable. You're allowed to disagree. How you disagree decides whether the editor agrees with you.

The approach

1. Acknowledge the concern behind the comment.
2. Explain your reasoning, with evidence: data, citations, an extra analysis.
3. Make a change anyway where you can, often a clarification, so the next reader doesn't have the same doubt.
4. Stay polite and factual. The editor is the judge; write for them.

Example

ExampleR1.4: "The authors should use a transformer model instead of a CNN." Reply: "We understand the reviewer's interest in transformers. Our goal is a model that runs on low-cost phones offline; at our model size, CNNs remain faster on such devices (see benchmark in Supplementary Table S2). To address the concern, we also ran a small vision transformer of similar size: it achieved similar accuracy (−0.4 points) but was 2.3× slower on the target phone (new Supplementary Table S3). We have added this reasoning to Section 3.1 (lines 150–158)."

What to avoid

• "The reviewer clearly didn't read the paper."
• Ignoring the comment.
• Arguing at length without new evidence.

When it's truly out of scope

Say so politely: "This is an interesting direction, but beyond the scope of this paper; we have mentioned it as future work (line 430)."

RememberDisagree with evidence, not emotion: acknowledge the concern, show your reasons, and clarify the paper anyway.