The goal of the rebuttal/response process is to help ensure that paper submissions receive a fair and accurate assessment:
Rebuttals will be text only and of a maximum 4000 characters. An effective rebuttal focuses on the major issues raised in the reviews. These should be addressed before responding to minor points in the reviews. By prioritizing and focusing on the major concerns, and by grouping multiple reviewer comments that generally pertain to the same issue into a few major categories, you are demonstrating that you understand the general issues of primary concern to reviewers.
It is recommended that you summarize or paraphrase the issue before you address it, and clarify to which comment(s) you are responding. While the room for rebuttal is limited, if properly utilized by condensing the response down to the essentials, this is an effective way to show that you understood the reviewers' concerns and have valid answers to the questions raised in the reviews, or to establish that certain reviewer comments were incorrect or unsubstantiated.
An effective rebuttal addresses reviewers' criticisms by explaining where in the paper you had provided the requisite information, and perhaps further clarifying it.
New/additional experimental results in the rebuttal are not allowed, and breaking this rule is grounds for automatic desk rejection. It is, however, allowed to amend the presentation of existing results. Please note that the final paper decision will be made based on the content of the submitted manuscript and not any promised changes in the rebuttal. If accepted, authors are not allowed to make substantial changes to the fundamental content of the paper, including experiments, data, and analysis.
In the case that a reviewer requests specific, additional experiments or results as part of the rebuttal, Area Chairs are instructed to ignore those requests and to not penalize rebuttals which adhere to the rebuttal guidelines. Authors may, however, address such concerns primarily based on the content of the submitted paper. For example, authors may want to acknowledge suggestions for further experiments and consider them for future work. Authors may also decide to refute a request by making a case why a requested experiment is not necessary to judge their presented contributions.
Note that the lack of certain experiments or missing comparisons can be valid concerns listed under the paper's weaknesses and may have informed the reviewers' scores.
Authors are reminded that links to external material are also not allowed to be included in the rebuttal. This includes links to code repositories. In the rebuttal, authors can indicate their intention to make code available after acceptance.
Finally, a good rebuttal is polite. On the other hand, if you feel you have received a review that was not courteous, or made incorrect or unsubstantiated arguments that you can succinctly refute, you should point this out.