Article of Interest: Most arXiv papers contain information never meant to be shared

 

Image by geralt from Pixabay

An interesting read from Nature. A study by Pennekamp et al. (2026) (which is currently in the preprint stage) looked at over 2.7 million entries in the arXiv preprint server to see how many records may have unintentionally shared sensitive information. The authors of the study found that an astounding 88% of records appeared to have shared such information in their associated LaTeX files (being a software used to typeset documents), "from arguments between co-authors and to-do lists acknowledging weaknesses in the text, to passwords, GPS coordinates that can reveal a researcher’s home address, and application programming interface (API) keys — strings of characters that function like passwords for programmers."

As discussed in the feature from Nature, the accidental sharing of information via LaTeX files is not a new phenomenon. One study by Apruzzese and Fass (2026) even uncovered comments such as "WTF does this mean?" in their analysis of arXiv preprints (being a comment I think most academics can empathize with😄). 

The Nature article cautions researchers to be careful to remove sensitive information from files before submitting them to preprint servers like arXiv, as such information can be difficult (if not impossible) to efface once posted, and can be potentially discovered by large language models.

Happy reading! ☕

There really is a journal for everything

Thinking about trying to get published but feeling anxious about the process?  Why not start by submitting your work to the Journal of Universal Rejection?

From the journal's home page:

The founding principle of the Journal of Universal Rejection (JofUR) is rejection. Universal rejection. That is to say, all submissions, regardless of quality, will be rejected. Despite that apparent drawback, here are a number of reasons you may choose to submit to the JofUR:

• You can send your manuscript here without suffering waves of anxiety regarding the eventual fate of your submission. You know with 100% certainty that it will not be accepted for publication.
• There are no page-fees.
• You may claim to have submitted to the most prestigious journal (judged by acceptance rate).
• The JofUR is one-of-a-kind. Merely submitting work to it may be considered a badge of honor.
• You retain complete rights to your work, and are free to resubmit to other journals even before our review process is complete.
• Decisions are often (though not always) rendered within hours of submission.

In Biology Publishing Shakeup, eLife will Require Submissions to be Posted as Preprints

"When major biomedical research funders launched the open-access journal eLife in 2012, they hoped it would push biomedical publishing to take full advantage of the internet’s power to share results freely and instantly. In the ensuing years, the open-access model has caught on. And more and more biologists have shared work in online preprint servers such as bioRxiv and medRxiv before undergoing peer review.

But those changes are not enough for Michael Eisen, a biologist at the University of California, Berkeley, and the journal’s editor-in-chief since 2019. This week, eLife announced it will only review manuscripts that have been posted as preprints. And all peer reviews will be made public, including those for manuscripts the journal rejects. Eisen sees the changes as the next logical step in the evolution of the preprint, he told ScienceInsider."

Read the whole story.

Recommended reading by Barbara Ruggeri who felt this was a significant development for peer review and academic publishing.  

The Scientific Paper is Obsolete. Here's what's Next ...

 

Well ... Not quite.

This article, wiritten by James Somers, came out in The Atlantic in April 2018, but I only just recently ran across it via PocketWorthy.

I thought it was a fascinating read, considering a big part of our bread-and-butter as health science librarians is still journal articles.    The mode may have shifted to digital formats but they are still flat 1-dimensional articles, standard pdfs.  But they could be so much more ...

Back to Somer's article:  

"This is, of course, the whole problem of scientific communication in a nutshell: Scientific results today are as often as not found with the help of computers. That’s because the ideas are complex, dynamic, hard to grab ahold of in your mind’s eye. And yet by far the most popular tool we have for communicating these results is the PDF—literally a simulation of a piece of paper. Maybe we can do better."

He introduces us to software called Wolfram Computational Notebooks (Mathematica) and makes a case for the next thing after journal articles being computational essays that are interactive and dynamic, allowing the reader to "read it and run it:"  

The idea is that a “paper” of this sort would [be dynamic ...]—interactive diagrams interleaved within the text—with the added benefit that all the code generating those diagrams, and the data behind them, would be right there for the reader to see and play with. “Frankly, when you do something that is a nice clean Wolfram-language thing in a notebook, there’s no [BS] there. It is what it is, it does what it does. You don’t get to fudge your data,” Wolfram says.

To write a paper in a Mathematica notebook is to reveal your results and methods at the same time; the published paper and the work that begot it. Which shouldn’t just make it easier for readers to understand what you did—it should make it easier for them to replicate it (or not). With millions of scientists worldwide producing incremental contributions, the only way to have those contributions add up to something significant is if others can reliably build on them. “That’s what having science presented as computational essays can achieve,” Wolfram said.

It sounds like Mathematica hasn't quite been adopted like the open source option has ...  Jupyter is an open source version of computational notebooks.  Somers compared the two computational notebooks thus:  Wolfram's uses the quintissential cathedral approach vs Jupyter's open source bazaar approach.  Wolfram's is the proprietary software, carefully constructed, and Jupyter is wide open and freely available and adaptable.

So far, pdfs are still the standard format for publishing.  Journals don't seem to be requiring computational essays as supplemental data along with article submissions -- or are they?

This sounds absolutely fascinating to me. Are any HSL Librarians using this software?  Either Wolfram's or Jupyter?  Is it available at your institutions?  Are you helping researchers manage their data with this type of software?  Have you seen any of these computational essays?  [They give an example in the article, but I wasn't able to get much out of it ...]  Where are they stored and published?  How are they made available?  Is NLM using these computational essays in the biomedical field?  I'm curious to know more ...

What's your experience with computational essays?  

Should I ask MLA for a CE class on them so we can all learn more?