Concerns About Increasing Appendix Lengths in AI Conference Papers
The post discusses the growing trend of longer appendices in conference papers, which are becoming essential parts of the paper rather than just supplementary material.
Why it matters
This discussion highlights an important issue in the AI research community regarding the role and expectations of conference papers versus journal publications.
Key Points
- 1Appendix lengths in AI conference papers (ICML, NeurIPS, etc.) are increasing significantly
- 2Appendices are often required to include more experiments and details that cannot fit in the main 8-10 page limit
- 3This makes the appendix a mandatory part of the paper, going against the concept of the main pages standing on their own
- 4The author believes that papers requiring 25+ pages should be published in journals, not conferences with strict page limits
Details
The post raises concerns about the growing trend of longer appendices in AI conference papers. The author observes that appendices are often required to include additional experiments and details that cannot fit within the typical 8-10 page limit for conference papers. This effectively makes the appendix an essential part of the paper, rather than just supplementary material. The author argues that the original purpose of page limits in conferences was for the main paper to stand on its own, with the appendix containing only secondary information. However, if the appendix is now required to include a significant portion of the work, then the natural place for such a long paper should be a journal, not a conference. The author questions whether this trend of increasing appendix lengths has become the new normal in the field.
No comments yet
Be the first to comment