We are currently reproducing several papers in a specific sub-field of computer science under different experimental conditions. When we submit the paper, we will be required to declare people who may have conflicts of interest. Most of the original authors of the papers we are reproducing are aware that we are doing this work.
Should we declare the authors of the original papers that we are reproducing as conflicts of interest?
Arguments in favor:
- Reviews may not be blind, as they know we are doing the work (review is double-blind)
- We may be "stepping on some toes" by showing that their results do not hold up when reproduced under more realistic circumstances. Conversely, as Wrzlprmft notes, we may be biasing others in favor of accepting, if we show their results to be correct.
- They are best suited as reviewers. If we name all of them as conflicts of interest, we will have excluded more or less every expert in the field, thus making it much less likely that we will get a reviewer familiar with the field
Is there an established guideline for this kind of situation (in CS or other fields)?