FAST '23 Double-Blind Submission and Reviewing Guidance

Blinding is not intended to be a great burden for reviewers or authors. After reading this page, if you are uncertain about how to anonymize your submission, please contact the program co-chairs, at fast23chairs@usenix.org, for guidance well in advance of the submission deadline.

General Information

Double-blind reviewing has been shown to reduce bias in the reviewing process. Double-blind means that authors don't know who is reviewing their papers and reviewers don't know the authors of a paper. FAST (and most conferences) are adopting this policy because we don't want a reviewer's attitude toward a submission to be affected, even unconsciously, by the identity of the authors. Similarly, reviewers' identities are blinded from authors to encourage honest and high-quality feedback to authors.

It is important to note that the program committee chairs know the identity of both reviewers and authors. This ensures that the process is fair and that inadvertent issues don't arise. For example, chairs may intervene to ensure that a reviewer doesn't recommend rejecting a paper due to preprint work (e.g., on arXiv) that the reviewer didn't realize was by the author of the submitted paper.

Policy for Regular Papers

The paper review process will be double-blind. Authors must make a good faith effort to anonymize their submissions, and they should not identify themselves either explicitly or by implication (e.g., through the references or acknowledgments). To refer to your previous work, write as if it was written by a third party. Do not say "reference removed for blind review." Supplemental material must also be anonymized. Submissions violating anonymization rules will not be considered for review.

Policy for Deployed-system Papers

Deployed papers must follow the double-blind policy above in that the authors need to be anonymized. The key difference is that the product or company described in the paper need not be anonymized.

In evaluating whether to deanonymize a product or company, consider whether this information is necessary to understand the work. For instance, one could probably understand how a sorting library works without knowing it came from Palm, Inc (i.e., the deanonymization does more to potentially bias the reader than clarify the work). On the other hand, a paper on optimizing storage for a Palm Pilot may be difficult to understand if the authors were deliberately vague on the details of the device and its capabilities.

FAQ for Authors

Q: What are the basics of ensuring anonymity?
A: The key to remember is that the goal isn't to make your identity completely undiscoverable across the entire internet, but rather ensure the paper is written in a way that plausibly could have been written by someone else.

There are several items you can easily do for every submission:

  • Cite all related work in the 3rd person
  • Use care in naming of files in your submission or supplemental material
  • Remove any identifying names (e.g., acknowledgements, funding sources)
  • Do not link to any site or webpage that will give away your identity

Q: Is it OK to post my submission (or any details about my submission) on arXiv, department webpage, Facebook, a blog, Twitter, git, or similar venues while it is under submission?
A: This is a very delicate subject, but this goes back again to having both authors and reviewers make a best effort to avoid discovering author identity.

First, we advise reviewers to complete a preliminary review of a paper prior to any search for related work. Although reviewers may eventually find your technical report or arXiv paper, they will already have taken notes and created an opinion of the work before knowing the authors' identities.

Second, we ask authors to make a reasonable effort not to "push" deanonymizing information to PC members out-of-band. For instance, it is inappropriate to email draft work directly to PC members for feedback, or post to mailing lists, social media, or any other forum that may potentially include PC members. In contrast, it is acceptable to post your work on a site such as arXiv, since it may then be found via a Google search but would not readily appear in the feed of a PC member where they would see your work without actively searching for it. It is OK if reviewers find your deanonymized work through their actions once they have done a preliminary review, but they should not receive this information passively, such as via email or their social media feed.

Q: How should I cite my prior workshop paper that my FAST '23 submission extends?
A: Work that extends a previous workshop paper by the authors should discuss the extensions upon the previous workshop paper in the submission using an anonymous citation. Further, the authors should upload an anonymized copy of the workshop paper along with their submission.

Q: How should I cite my prior conference paper or online material?
A: The most important point here is to never avoid a citation to ensure anonymity. Always cite related and past work using the following guidelines:

  • Always cite in the 3rd person
  • Do not provide links to your own online content. If needed, authors may submit anonymized supplemental material along with the paper submission. In case of issues uploading supplemental material, please contact the program co-chairs.
  • If you need to reference another submission to FAST '23 on a related topic, reference it as follows: "A related paper describes the design and implementation of our file system [Anonymous 2022]." with the corresponding citation: "[Anonymous 2022] Under submission. Details omitted for double-blind reviewing."

Q: How do I ensure the anonymity of my submission that is based on a well-known system or code in a public repository?
A: This is a situation where clarity takes precedence over anonymity. If the work simply cannot be understood outside of the context of its larger project, that context should be included. For instance, Windows creates processes with a very different API than Unix; a paper on process creation would be very confusing if it were not clear the target platform was Windows. On the other hand, it may not be necessary to understand the work to know whether the work was done at Microsoft, or by whom.

Code in a public repository should never be an anonymity barrier: if anyone can download and modify the code, then, in principle, anyone could have written a paper that derives from that work. There is no reason to hide that one modified an open source project. Do avoid drawing attention to particular commits if the work has already been upstreamed, or inadvertently using published code to deanonymize the work. It is ok to mention in a submission that code has been accepted to an open source project, and trust the reviewers not to look at commit logs.

As always, please contact the program co-chairs if you have any doubts or need clarifications.

Q: Is it a problem if a PC member has seen previous details of my submission (e.g., poster, WIP, invited talk)?
A: In addition to collecting conflicts of interest from authors and reviewers, we expect reviewers to opt-out of reviewing any work they cannot objectively review as part of review bidding. For example, if a reviewer remembers seeing a poster of the same work and believes that knowing the identity of at least one author will affect their review, we expect the reviewer to decline to review the paper. This approach aims to strike a balance between facilitating early feedback to improve the work, and minimizing bias in the review process.

Q: How is the conflict of interest policy enforced with double-blinding?
A: We rely on authors and reviewers accurately listing conflicts of interest in the submission system. The program co-chairs can see the authors' names and conflicts and will review listed conflicts of interest for accuracy. The submission system ensures that reviewers are not assigned papers with listed conflicts.

FAQ for Reviewers

Q: Is it ok to Google a paper name? If so, when?
A: Do not search for details (e.g., title, citations) of a paper until you have done a preliminary review. Since knowing authorship may substantially bias the review, it is best to form as much of an opinion of the quality of the work as possible before searching. In general, we expect all reviewers to make a good faith effort to not learn the authors names during the review process. That said, searching for related work the reviewer may have missed is a common part of "due diligence" and perfectly acceptable as part of the review process.

Q: Is it ok to Google the paper area or general topic?
A: Yes, it is fine to learn about the general topic before and during the review of a paper.

Q: I learned the author's name(s) or affiliation...what should I do?
A: The double-blind policy is not meant to be a substitute for the conflict-of-interest policy. The real guideline for whether a PC member can review a paper is based on their determination of whether they can be fair and unbiased in their review AND the conflict-of-interest policy. Although every effort should be made to avoid discovering author identity, knowing the author names is not a problem as long as the review is fair and unbiased.

Please err on the side of caution and contact the program co-chairs if you are concerned that learning an author's identity may affect your review of a paper, especially if you learned the identity prior to a preliminary review.

Q: A paper gives away their identity explicitly or implicitly. What should I do?
A: Assuming it is not a Deployed Systems paper, please contact the program co-chairs with your concern.

Q: The authors anonymously cite a previous workshop paper on which this work is based, can I find the workshop paper and read it as part of my review?
A: Authors are asked to include an anonymized version of the workshop paper in their submission to facilitate the reviewing process and remove the need for reviewers to search for the (deanonymized) workshop version. As stated above, do not seek out any citations, including a previous workshop paper, prior to first completing an initial review. FAST 23's policy is that material that was previously published at a workshop (e.g., HotStorage), and not another conference or journal, may be included in a FAST submission without prejudice. However, the submission should still have a substantial new contribution above the previously published work, and we expect reviewers to compare the two documents.

Q: How is the conflict of interest policy enforced with double-blinding?
A: We rely on authors and reviewers accurately listing conflicts of interest in the submission system. The program co-chairs can see the authors' names and conflicts and will review listed conflicts of interest for accuracy. The submission system ensures that reviewers are not assigned papers with listed conflicts.

If you suspect you have an overlooked conflict of interest with a paper you are assigned to review, please contact the program co-chairs as soon as possible.