AAAI-20 Multiple Submission Determination Process
Some authors may argue that the submissions are distinct papers. But if distinct papers (by overlapping authors) claim to be making (or appear to be making) some of the same contributions, this is still problematic. Only one published conference paper should get credit for the novel contribution. As a result, our reviewers would not be able to properly assess the novel contribution of the paper, because it is not clear to which submission it belongs. Hence, it does not make sense to have our reviewers review the paper in its current state. This also applies (1) to cases where it is left ambiguous whether something is a new contribution of the paper or not, because our reviewers would still not be able to judge the paper properly; and (2) to cases where multiple papers on a larger project are simultaneously written, in a way where the contribution of each is “sliced too thin” (trying to make too many papers out of one project); if it seems too likely that the reviewer would think that one of the paper’s contributions is the larger project as a whole, and likely would not have been impressed if the larger project is already considered known, then we don’t think the reviewers can evaluate the paper well. Generally, if the designated evaluators have trouble seeing the difference, then probably so would the reviewers, and we can’t expect them to be able to evaluate the submission properly on its own.
Our process for these cases is as follows. Whenever a concern is raised about the similarity of two submissions, at least three people inspect the concern, and if they all agree that the simultaneous submission is too problematic, we reject the paper and inform the other conference. This is necessarily a subjective process, which is why the process is conservative.
We would like to emphasize why it is ethically wrong to submit the same, or highly overlapping, papers to multiple conferences at the same time, for at least the following reasons.
- The community is overwhelmed with reviewing tasks, and authors submitting multiple versions of the paper at once (or the same contribution across multiple papers) are claiming a disproportionate share of the reviewing pool. As a result, many reviewers waste time discovering the same issues with the paper/contribution. Also, there are fewer relevant reviewers available to review the papers of other, well-behaved authors.
- Authors engaging in this behavior attempt to give themselves an unfairly high chance to have a paper accepted, thereby also unfairly reducing the chances of other, well-behaved authors.
- There is a chance of both submissions being accepted. In this case, the authors would either have to withdraw one of them, which hurts one conference’s ability to plan and wastes the effort they put into assessing the paper; or publish the same results multiple times, which is clearly wrong; or attempt to change one of the papers to something that is no longer what was reviewed, which is severely wrong.
Multiple submissions also contribute to a cynical attitude among the people involved. Therefore, these actions hurt the AI community, and science in general. Authors engaging in such behavior are giving themselves, and the subcommunities of which they are a part, a bad reputation. We hope that these authors will correct this by acting in a responsible way in the future.