Extensive or summarised background? -- You don't want extensive background material padding every paper out with several more pages, so once you're familiar with 'the literature' (the key background details), often you can quickly skim and get an idea of the baseline of the paper just from specific citations
Citation formats? -- some citation formats are better than others...I usually publish in journals using IEEE/numeric, which can be denser but requires flicking to the Bibliography...names can make it easier to recall the gist of a paper if it's a famous one like the 'Hinton' or whatnot, but can really pad out content if you have strings of names inline
Papers too short/page limits? -- You don't want pages and pages of content, otherwise the paper is likely contributing _more than one idea_, and so is less focused (and also kind of dilutes the review process). If a paper contributes A and B, but B is weak, do you still let it through if A is great? If the contributions are distinct, you can easily have an A paper and then request some more work for B.
2-column? -- 2-column and dense can make it easier to quickly jump from section to section, or get more information on a single page...so when reading in print format (or digitally with full-page view), you can easily see what was said earlier on a page, whereas 1-column tends to have much larger font so less content-dense.
I don't think we are at optimum, but I don't think ANY new format that's been proposed gives such a noticeable benefit to some of these areas as to overtake. Fully digital with hovers and links and stuff may be useful, but would completely regress when printed out for reading and annotating.
Having a different style of the paper will not magically make it easier to understand. Focusing too much on the format instead of the content is counter productive.. And having somewhat standardized sections allows an academic to jump to juicy parts with enough experience.
With Adobe PDF, you double click on one column, it zooms onto it perfectly, you double click again, it goes full page.. Navigation is super easy, the zoom feature perfectly fit the column to my screen.
Of course tts is a different beast, but.. Try to read science in general with tts.. It doesn't make sense.. And you need time to process each sentence and put it into context.. So even with reflow, tts would be a nightmare for science content.
Epub is great for content that is mostly texts (Novels,…). As soon as there are objects that need to keep their relative distance to each other (equations, diagrams), I go for PDF. It's also not easy to get good typesetting for epub files. I believe that typesetting gives you a visual map that provides an additional memory boost. Doing focused reading on a phone is a no-no for me. I use my tablet for PDFs and my e-reader for epubs.
We don't like reading these newspaper-column style papers any more than lay people do.
Scientific paper formatting is because they are still, generally, actually paper-published. 2 column layouts, for instance, are actually quite readable on paper. Online articles tend to have single column styles.
There is also very little "fiddling" in academic publishing. Journals/publishers tend to pick a template and styling system and funnel everything into that single style regardless of content. They are usually thinly staffed and want to enforce aesthetic standardization article to article to reduce typesetting workload and to ward off any claims of favoritism.
For instance, a typical review process in Nature or Cell takes around 12 months these days. Lots of waiting, back and forth messages, and waste. If you get rejected, switching from one journal to the other implies massive changes to format and rewriting, as the length constraints could not be more different.
Computer Science has a much more dynamic publication culture.
In an ideal world, every publication would be more similar to a Git repository or a website that contains all the code for the statistical evaluations and, of course, the raw data that was collected initially The code would document why a certain sample was excluded etc. Of course, the results wouldn't be presented in dry tables or bar charts. There would be interactive visualizations that support the understanding of the data and results. All this would be integrated into the text. Of course, such a text would have real hyperlinks to other works without the need to go to a reference section and look up a publication.
If a publication would fail to replicate, the original authors or their academic offspring would add a remark that this work is superseeded by some other work.
Needless to say that all this has been proposed already in publications ironically in PDF. Also noteworthy that Berners-Lee proposed the WWW 35 years ago, which basically implements this.
Completely unreadable on different format screens and makes accessibility impossible and therefore narrowing the field of possible readers even lower.
We've had HTML since what, the beginning of the WWW? How are we still using PDFs in 2024? Completely absurd.
Two column format is not typical in mathematics; here is an example of one of my papers, with formatting is highly typical for my field:
https://arxiv.org/pdf/2404.00541.pdf
I'd say that I do value aesthetics; I like this look, for the same reason that I appreciate HN's look and feel. It is simple, functional, and to the point.
Personally I dislike most of the trends in web design. The web feels increasingly designed by marketers who apparently want to measure or guide how I browse their sites -- rather than by someone who just wants to provide me useful information in an attractive and organized fashion.
For example here is the current website for the University of Wisconsin's mathematics department
and here is the 2004 version:
https://web.archive.org/web/20040619085511/http://www.math.w...
I prefer the old one.
I don’t think this discussion is applicable to just format/layout though. The writing style of academic papers is also anachronistic/archaic. I feel like most are written in an unnecessarily awkward way that impedes understanding. It’s almost an attempt to elevate the “complexity” of the content by choosing difficult writing styles, overusing jargon, and under-explaining. I’ve seen this style of writing previously defended in the same way as the aesthetics, that it is a matter of the content being “technical” or “scientific”. But if you look at any public facing post from companies that write deeply technical blogs or other such things, you’ll see there are definitively better alternatives to communicating deeply technical content.
These norms need to be broken. I think a paper should be given more weight if it is easy to consume. This means free access, nicer designs, simpler language, more transparent explanations.
You'll notice that GPT-4 Technical Report uses single column https://arxiv.org/abs/2303.08774
So some variants exist. They print well. Colours are more annoying to print.
The one that I don't like is many venues have a 10 page max, so some papers end up getting golfed down to 10 pages and become impossible to read.
Any academic who wants to be read publishes in a more readable version of their research.
I'm not sure how, but sometimes a well-formatted paper gets published Example: