From the abstract, it sounds more like a toolkit for "Interactive / Hypermedia" papers. The paper itself is still dead.

I was hoping more for "Living" as in "active, uncertain, will grow over time".

A toolkit for expressing the research from beginning to the end - the state of the world as you understand it, highlighting the key uncertainties and experiments, and mechanisms for viewing the history.

As a motivating example for the type of "living research paper" that I'm thinking of, think of long-running software design decisions a la the implementation of `async/await` in rust.

> I was hoping more for "Living" as in "active, uncertain, will grow over time".

This is, I think, the very antithesis of what research papers are for.

The point of a publication is that at some point in time, you wrote down your thoughts and process and results and submitted that writing (and possibly a related artifact) to a committee who evaluated it and decided it was Good.

To have a publication that updates over time is just... it doesn't work. What if it stops being Good? What if you screw something up that invalidates the results? Also, do you just never publish a new thing? How do people learn about the recent changes if not a new publication? Do all researchers now need to subscribe to RSS feeds of every project they've been interested in?

That sort of stuff is what blogs are for — or, honestly, CVs. But a singular publication needs to be frozen in time, or else it honestly loses its value.

You forgot the most important point. Once, you publish the final version it can be referenced. If you keep changing (unless minor corrections), then no one else can reference the work.

Research literature is like an ongoing conversation. Picture a social media site, where each paper is a comment. You don't want a comment (paper) to be edited, because any replies or references to that comment would lose their context.

Just publish dated-/versioned drafts of your paper, like arxiv?

This can be nicely achieved with https://manubot.org, which produces a manuscript (a (possibly freely accessible) HTML, a PDF and a DOCX file) from a set of git-tracked Markdown files.

There are people doing a bit of this work with methods journals and pre-registering hypotheses.

This is very nice. It's a good tool for academics wanting to engage with the work of other academics.

I think, however, the biggest issue with scholarly communication right now has to do with how inaccessible the language and techniques tend to be for the average person. Of course, a highly technical paper needs all those details. But I believe we should be looking at requiring authors to maybe make an "extended abstract" sort of deal when submitting articles.

More in line to what you see from news sites who would be reporting on your article, but in a way that will obviously be more accurate; as journalists might overlook important details. The language we use matters so much.

Lots of journals have a plain language abstract requirement. I think the primary issue is a lack of open access everywhere and for profit journals being seen as better than society journals.

What is missing is innovation in the space traditionally occupied by textbooks. We need new forms for digesting recent results and making them useful for answering the kinds of questions asked by people with different levels of training in a field.

Interesting, always curious to see things from the UW data lab!

Would be nice to have this supported as a custom format in Quarto [1], which supports some interactivity in documents, but doesn’t have the same focus on it as this.

[1]: https://quarto.org/

I'd be perfectly happy with having papers as Jupyter Notebooks