4

It is common, especially in Computer Science fields, to see papers that are under review at conferences and are also available as preprints (e.g. arXiv).

However, I am uncertain about the potential advantages or disadvantages of this approach.

  1. Does uploading a paper to ArXiv during the conference submission process have any significant impact on the review process?
  2. Is it generally considered a good practice to pre-publish on ArXiv, or is it more advisable to maintain a blind submission until the conference review is complete?
  3. Are there specific scenarios or types of conferences where pre-publishing on ArXiv is more beneficial or discouraged?

I also saw this question:

Is publishing on arXiv early a good idea?

and, this question:

Double blind review allowing arXiv submissions

But I think what I'm asking is something different. I want to know if it is advisable (or acceptable) in different situations to arXiv the paper, especially if you are submitting it to a conference in which reviews aren't blind.

2
  • 2
    I'm very surprised this hasn't been asked before... I suggest we generalize this to a canonical question for "Given X, Should I Put My Paper on Y," where here X = "submitted to conference" and Y = "arXiv"?
    – user173920
    Jan 14 at 1:50
  • Note that some CS conferences allow Preprint submission even if reviews are double-blind.
    – Coder
    Jan 14 at 21:10

1 Answer 1

3

1. The impact of preprints on the review process

A preprint is a non-peer reviewed publication that is often to be peer reviewed in the future and because arXiv requires the use of real names, if a reviewer were to search the title of the submitted paper (often discouraged by reviewer policy when double blind) and finds the preprint, or otherwise accidentally stumbles upon it, then this would break the double blind policy by de-anonymizing the identity of the authors or contributors of the paper.

Related arXiv policies:

Once double blind no longer holds in that case, the reviews know who the authors and contributors are, which brings in all its potential for bias in the review process. This bias is what blinding tries to avoid for a more technical and unbiased review.

There may be reviewer policies that prohibit such searching for such knowledge, however that is more of an honor system than an enforceable rule. Also, a reviewer's technical review may benefit from conducting a brief literature review, and the preprint (or technical report) could accidentally be discovered during that review.

The brief literature review can be beneficial, for example, whenever the expert conducting the review needs to confirm with evidence that this work provides (or does not provide) a novel contribution to the current state of the field. Or, they may need to verify or better understand niche specifics being cited from other works.

2. Are preprints good practice? or maintain blind submission?

The "good practice" answer depends on the cultural norms of your field of study. Far more important is whether the publishing venue that the paper is submitted to has a policy that allows preprints or allows active efforts to promote social media publicity of the submitted paper or its content.

If preprints or publicity is not explicitly stated to be allowed in the publisher's policies, then send them an email to confirm this before doing so, especially if the review process is to be double blind, or otherwise intends to keep the paper's authors and contributors anonymous.

3. Specific scenario where preprints were discouraged

Wikipedia has a list of academic publishers' policy on preprints, including some who do not accept preprints. Below, I briefly discuss one of the most popular machine learning conferences that recently prohibited the active publicity of preprints.

CVPR 2021-2024 Publicity and Media Restriction

The Conference on Computer Vision and Pattern Recognition (CVPR) is an example of a conference having discouraged and even setting rules against sharing your submitted article before acceptance or rejection. It was referred to as a "Publicity / social media policy" or "media ban" and was based on a motion passed in the CVPR 2015 PAMI-TC meeting. CVPR 2021 seems to be the first case of it enforced without some loophole. CVPR 2022 strengthened the policy and CVPR 2023 continued with that policy. This policy seems to also be in effect for CVPR 2024.

"Papers submitted to CVPR must not be discussed with the media until they have been officially accepted for publication. Violations of the embargo will result in the paper being removed from the conference and proceedings".

The social media ban does not prevent any non-peer reviewed preprints, such as arXiv, or making a public project page, but prohibited the publicity of the submitted or preprint paper or any project page, in particular within a specified "silent period", for example, for CVPR 2022 it was originally "projected to be from 10/19/2021 to 03/02/2022".

On that same linked page under header faq there is more details including reasoning.

Authors must not:

  • Talk to the media about their work as "in submission to CVPR".
  • Make any posts to social media or elsewhere that can be linked to a specific CVPR submission (e.g., mentioning the title of the submission or details and content).

Authors may:

  • Talk about their work in an academic presentation without saying it's submitted to CVPR.
  • Submit to arXiv without mentioning CVPR.

The purpose was to preserve the anonymity of a submitted paper's authors and contributors as much as possible and avoid biases introduced from the reviewers' finding out whether a large well-known company, organization, or individual was involved in the paper.

Benefits of preprints

Non-exhaustive benefits of preprints:

  • You get the work out there faster than waiting for finished peer-review for people to see, learn from, and possibly use or build upon.
  • You can get informative feedback from others this way to help improve the quality of the work.
  • You have date-time publicly documented evidence of your contributions from the preprint. Useful for whenever you need or want to prove "I was first." If that matters.

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .