Preprints and copyright for visualization papers

Cody Dunne

2024-06-04

TL;DR: I recommend the CC-BY 4.0 license for papers when possible; Apache 2.0 for code; and OSF for hosting. See the IEEE VIS journal paper template for an example including the OSF Preprint and OSF Project I added. But publishers and funders create varied and sometimes incompatible license and hosting restrictions.

Note: I'm taking off my hat as an Open Practices Chair (OPC) for IEEE VIS 2022–2024. None of the following represents the position of IEEE or IEEE VIS. But VIS also recommends CC-BY for papers when possible, Apache 2.0 for code, and using OSF for hosting.

Table of contents

• The Context
• Plan S/cOAlition S
• Wiley CGF and EuroVis Full Papers and STARS in CGF
• EuroVis Short Papers in the Eurographics DL
• ACM CHI
• IEEE Generally
• IEEE TVCG
• IEEE VIS
• IEEE Loophole for Preprints?
• What does the future hold?
• arXiv vs. OSF vs. other repositories for preprints and supplemental material

The Context

Several of us visualization researchers recently discussed copyright issues for visualization papers. I'd like to summarize and bring up related policies from other publishers and funders. Please correct me if you see anything wrong!

Thank you, Tamara Munzner, for starting this conversation; Zhu-Tian Chen for pushing back on my blanket recommendation for CC-BY 😊; and Dominik Moritz, Lonni Besançon, and Alexander Lex for their contributions. This started as a Slack post but quickly grew unwieldy—now it is my inaugural blog post here! (I also have a very outdated blog at codydunne.blogspot.com.)

This conversation took place in the context of a broader push toward open practices in the visualization community. On the IEEE VIS Open Practices page, we wrote "Visualization research is better communicated and acted on if it is freely accessible to the research community, practitioners, and the general public. Likewise, there are steps as a community we can take to make sure that our work is transparent and that the claims in our papers are scrutinizable." Our goal was to "make work at VIS (1) reproducible, (2) replicable, and/or (3) extensible by future researchers." I encourage everyone to read that page and the links therein.

My summary of our conversation follows... Note that there are 6 variants of CC-BY*, hence the * I use. (I'm ignoring CC0 which preserves no rights.)

Plan S/cOAlition S

You get CC-BY* if your funder and publisher can agree—If your work is funded by cOAlition S funders, you must release your papers under an open license like CC-BY*. See the IEEE VIS 2024 Plan S guidance for details, and the map in Lonnie Besançon and my 2022 paper DyStopia: Into a potential future of IEEE VIS under Plan S for an overview of countries with cOAlition S funders.

Wiley CGF and EuroVis Full Papers and STARS in CGF

You get CC-BY* by paying; otherwise, Wiley's license—You would pay $3,710 USD. Per EuroVis guidance, you may be able to get Plan S compliance depending on your funder. Submitted (not even peer-reviewed) articles can be published on preprint servers including arXiv and OSF with added copyright text after acceptance. Accepted versions can go on arXiv and OSF after a 12-month embargo and with added copyright text. Without paying, on arXiv you'd use the arxiv.org perpetual, non-exclusive license; on OSF No License and provide copyright details.

EuroVis Short Papers in the Eurographics DL

You get CC-BY* for free—See EuroVis guidance and Eurographics DL guidance. On arXiv and OSF you could choose CC-BY*.

ACM CHI

You get CC-BY* by paying; otherwise, ACM's license—You can pay individually $1000 USD ($700 for ACM members) or have the corresponding author's institution pay up to $95,000 USD each year for "read + publish" access to the ACM DL. You can see if your institution is included in this list and map. You only get Plan S compliance by paying. See the CHI guidance. Without paying, on arXiv you'd use the arxiv.org perpetual, non-exclusive license; on OSF No License and provide copyright details.

IEEE Generally

You get CC-BY* by paying; otherwise, IEEE's license—IEEE's conference policy Zhu-Tian Chen pointed us to and the journal policy agree. Without paying, once the paper has been accepted, all Preprint, Submitted, and Accepted articles online must be updated to comply with IEEE's policies. You can not share the Final published article online. Note that IEEE only allows preprints on arXiv, TechRxiv (OPC doesn't recommend), authors' personal website, and authors' employers' websites. The OPC—including Lonni Besançon, Michael Correll, Laura Garrison, and me during 2022–2024—have worked for years to get OSF approved for preprints 😠. IEEE also specifies the text that should go on Preprint, Submitted, and Accepted. Note that even your thesis must be updated with credits to IEEE for any figure or text from your paper! Without paying, on arXiv you'd use the arxiv.org perpetual, non-exclusive license; on OSF (if you don't mind violating IEEE's policy) No License and provide copyright details.

IEEE TVCG

You get CC-BY* by paying; otherwise, IEEE's license—See IEEE Generally above. TVCG is a hybrid journal, which means its article processing charge (APC) is $2495 USD + tax. The APC can be reduced or waived for low and lower-income countries. You also get a 5% discount for IEEE members. Without paying, on arXiv you'd use the arxiv.org perpetual, non-exclusive; on OSF No License and provide copyright details.

IEEE VIS

You get CC-BY* by paying; otherwise, IEEE's license—See IEEE Generally above. I don't know what the APC is. Does anyone have any experience paying for open access?

IEEE Loophole for Preprints?

There is a possible loophole in the IEEE conference and journal policies:

Please note that once the paper has been published by IEEE, preprints on locations not specified above should be removed if possible.

Emphasis mine. If you preregister the preprint as a PDF on OSF, it is immutable! You can't remove it on your own, and you can't update the text to include the copyright statement. But the regular (and better) OSF preprint service lets you withdraw the preprint. Your mileage may vary...

What does the future hold?

All this guidance may change soon! Here are some upcoming changes I'm aware of:

• Plan S funders are cutting funding for transformative journals such as IEEE TVCG, which is a hybrid journal. Will TVCG be legacied in, and for how long? Will authors covered by Plan S be prohibited from publishing in TVCG if it doesn't become gold open access?

• The U.S. NSF has a Public Access Initiative:

  by 2025 peer-reviewed publications and associated data arising from federally funded research be made immediately and freely available upon date of publication.

  TBD what that entails regarding APCs.

• The U.S. NIH is also working on a policy that will take effect 2025-12-31. Details TBD.

arXiv vs. OSF vs. other repositories for preprints and supplemental material

Authors currently use arXiv, OSF, IEEE Xplore, IEEE Dataport, GitHub, GitLab, Dataverse, Databrary, institutional repositories, homepages, and lab pages for sharing papers and supplemental material. The OPC came up with these evaluation criteria back in 2022. Papers & materials repositories should:

1. be freely accessible to enable future researchers from all backgrounds,
2. provide a persistent identifier to avoid link rot and enable citation tracking,
3. allow for post-submission updates to correct errors and address omissions while keeping the version history,
4. have a long-term sustainability plan and are otherwise resilient to financial or corporate pressures, and
5. have immutable versioning for preregistrations.

The OPC recommended OSF for supplemental material. We cautioned against using solely IEEE Xplore and Dataport, GitHub & GitLab, or institutional repositories / homepages / lab pages. We encouraged hosting multiple mirrors of material.

The OPC recommended OSF or arXiv for preprints. Both meet criteria 1–4. Additionally:

arXiv PROs:

1. Supports conversion to future document formats. arXiv requires LaTeX users to upload the source code. Most of their other good reasons to upload source don't apply to modern LaTeX usage, such as our VIS templates.
2. Widely used in CS.
3. Fast to load.
4. PDFs load in the browser window reliably.
5. Indexed by DBLP.

arXiv CONs (IEEE VIS provides a guide to using arXiv that describes some of these):

1. Everything you upload, including commented-out LaTeX, is available publicly forever! This burned some Microsoft researchers. You just need to go to Download, Other formats, then Download source.
2. Many LaTeX users often face problems getting arXiv to compile. Here are some common mistakes. I have had papers require slow and repeated intervention by arXiv staff to compile or simply waive the source requirement and let us upload a PDF. Other visualization researchers have mentioned this issue. Adding to the frustration is that arXiv sometimes doesn't have the latest LaTeX packages and that it doesn't run BibTeX for you.
3. There is a file size limit of 10 MB unless you email their administrators.
4. There is no supplemental material hosting.
5. New users face an endorsement requirement and delay. arXiv requires endorsement from existing researchers, which can be hard for newcomers to our community to get. IEEE VIS provides instructions for using arXiv.

OSF PROs:

1. You can simply upload any PDF! There are no LaTeX compilation issues to troubleshoot.
2. File size limits are much higher than arXiv: 5 GB per file and 50 GB per project.
3. You can host all your supplemental material, preregistrations, etc.
4. You can link your preprint with supplemental materials directly. See, e.g,. our OSF preprint on differential privacy and scatterplots which has PUBLIC DATA and PREREGISTRATION links at the top. Likewise, our OSF project has a link to the preprint near the top.
5. Much less chance of information leakage without the source included.

OSF CONs:

1. Doesn't support conversion to future document formats.
2. Not widely used yet in CS.
3. Slow to load and somewhat janky interface.
4. PDFs download rather than showing in your browser (at least for many people?).
5. Not indexed by DBLP.
6. OSF requires pre-moderation on preprints (but not on projects). This means that you get your OSF URL & DOI, but the preprint isn't visible until a short while later when a moderator approves.

Which to use? I prefer OSF. But you don't have to pick one—you can also put your preprint on both!

Thanks if you read this far. You are a rockstar 💪!