3 ms·

It’s time that someone starts a thing similar in appearance to GitHub, but for science (datasets, images, calculations, scripts) and then, if journals required

It’s time that someone starts a thing similar in appearance to GitHub, but for science (datasets, images, calculations, scripts) and then, if journals required it, it just might get traction and make fraud science easy to spot. Edit: found this article which says everything I wanted to say but couldn’t put into words. https://slate.com/technology/2017/04/we-need-a-github-for-academic-research.html https://slate.com/technology/2017/04/we-need-a-github-for-ac... Add another aspect here that LaTeX is a bit outdated in 2024 (I know that’s controversial! Sorry) and that we can do a lot better for digesting and displaying information than A4 sheets of paper, for example responsiveness, audit/comment logs/references to individual paragraphs/revision logs, and the ability to click figures and see underlying data or high resolution copies. This would be great in a web-based editor medium. Also the ability to “fork” a paper would be fantastic. And to automatically track and generate references, then roll it up as back/forward reference analytics for the authors so they can see impact.

jahewson 2y ago
> Add another aspect here that LaTeX is a bit outdated in 2024 (I know that’s controversial! Sorry) I met Leslie Lamport seven or eight years ago and asked him what a completely modern LaTeX might look like. He replied “well, we won’t be using PDFs in twenty years” and so it would need to be something completely different. Something interactive, with depth. Remembering, of course, to focus on quality content first and quality presentation second. In a world with LLMs, this question becomes ever more interesting - why write a literature review if one can be generated?
- aetherspawn 2y ago
  I think the important thing is capturing information in basic blocks (text, images, etc) and having the flexibility to reflow it later for any modern presentation mode, be it ingestion by LLM, listening to it, or just rendering it on desktop, mobile. Translation and a11y is another important consideration here.

jahewson 2y ago

> Add another aspect here that LaTeX is a bit outdated in 2024 (I know that’s controversial! Sorry) I met Leslie Lamport seven or eight years ago and asked him what a completely modern LaTeX might look like. He replied “well, we won’t be using PDFs in twenty years” and so it would need to be something completely different. Something interactive, with depth. Remembering, of course, to focus on quality content first and quality presentation second. In a world with LLMs, this question becomes ever more interesting - why write a literature review if one can be generated?

aetherspawn 2y ago
I think the important thing is capturing information in basic blocks (text, images, etc) and having the flexibility to reflow it later for any modern presentation mode, be it ingestion by LLM, listening to it, or just rendering it on desktop, mobile. Translation and a11y is another important consideration here.