Can lowlighting less important text improve readability? Are there alternatives?

Question

I think many experience the following: Sometimes I want to get the main message of a paper within 10 min then I am annoyed by any less important details distracting me. And sometimes I spend days or weeks to understand a certain part of a paper, where I would highly appreciate any further information on that aspect of the paper. Or I am just very curious if the authors got further results on a very specific aspect of their paper.

So I think good papers should be more interactive in a way that the reader can easily choose which parts to read and which parts can be skipped without missing the main message (or the introduction of the notation necessary to understand the main message).

I know about the following methods to achieve this goal:

1. less important stuff in the Appendix
2. Using remark environments to indicate that a certain remark provides additional information.
3. Using good section titels to let the reader choose which aspects they want to read.
4. Using footnotes to provide optional additional information that is specifically related to the content at the position where the footnote appears
5. Using lowlighted (i.e. grey / half transparent) text to indicate less important additional optional information.
6. Using smaller fontsize to indicate less important additional optional sections or paragrahps.
7. \underbrace, \overbrace to explain certain terms (that might be obvious to some readers)
8. \overset such as $a\overset{\text{Th. 1}}{=}b+c$ to indicate that a=b+c holds true because of Theorem 1, which might be obvious to some readers, but not to others.
9. (Writing something unimportant in parenthesis.)
10. There are probably more options...

I think all of these methods have their advantages and disadvantages and I think for each of them there exist scenarios where they can be very helpful for readability. I don't really understand why some journals and even Arxiv forbid some of them.

Question 1: What are the pros and cons of low lighting text by making it half transparent? Are there situations where it is helpful?

I think for example for papers in the intersection of mathematics and computer science grey symbols inside formulas can be useful to allow readability for both communities. E.g. a real valued random variable X is typically introduced as $X:(\Omega,\Sigma)\to(\R,\mathfrac{B})$ in mathematics literature, but this might unnecessarily confuse some computer scientists, because in this field of computer science the assumptions expressed by this notation are typically made implicitly without explicitly writing it down. So making :(\Omega,\Sigma)\to(\R,\mathfrac{B}) half transparent is understandable for both communities.

Therefore, we wrote our paper using lots of lowlighting and uploaded it to Arxiv a couple of months ago. Now, we wanted to update it fixing some typos and to furhter generalize the theory, but we got a mail, that highlighting and lowlighting is not accepted by Arxiv, because it is too unconventional.

Question 2: What can we do know? Is there an equally good alternative instead of lowlighting?

Related other questions: Is there a way to improve the readability of equations in research papers? Is highlighting important references a good idea?

Answer 1

Yes, there is an alternative to de-emphasizing the less important parts (which might possibly impair readability): Emphasize the most important parts. Something you can do is to use clearly outlined boxes with "take-home" messages, like in the example below:

Answer 2

I could fill a book with my (unorthodox) opinions about this. The arXiv is incredibly conservative about what it will accept, but I find it gratifying that you are thinking about this. Formatting makes a huge difference for a lot of people, due to disability, cognitive style, or just personal preference.

Something I imagine in my academic fantasy world, but have neither seen nor attempted myself, is to have papers posted in HTML format with customizable stylesheets. So the author would tag various sections as notational, of varying importance, etc., in very much the way you describe, and then the reader could load their own preferences about how they want that to be visually coded.

The old guard will probably go with print-like high-contrast plain vanilla, but I personally am strongly in agreement with you that visual cues reinforcing the organization of the paper would be extremely useful. However, those decisions are best left to the individual user.

Answer 3

Not a fan of smaller (inline) text or greyed text. You should be able to use structure, appendices, footnotes, abstract, conclusion, figure captions (left out of your list) as good enough keys to allow the time-pressed reader to skim to key parts and/or inspect details as needed. There is no need for the textual tools you propose given the other tools available.

I would avoid using greyed, or small, text as it is hard to read. Consider especially that some readers may be over 30 (start to get presbyopia). Also that some people may print the article for reference (faint type is a printing/copying problem, even worse than the eyestrain on screen).

Answer 4

Think of this as an engineering project. You would read the existing literature, and learn about best practices in a field, before even thinking of creating your own new solutions.

Do the same sort of web searching, book selection, and reading on formatting technical material for readability as you would if you were trying to contribute to a field of engineering you have used, but not previously studied.

While articles are often formatted by their authors, who are experts on the subject matter, textbooks may benefit from formatting by experts on readability. Select a few of the textbooks you have found most helpful, and study them again analyzing their formatting rather than their content.

I suggest looking at a few issues of Scientific American. Its economic survival depends on producing articles that can be enjoyed by anyone from a total novice to an expert. Many of their readers are experts in some of the fields they cover, but novices in most.

Most likely, after studying the field, you will find that simply following existing best practices will meet your needs.

If, on the other hand, you think you have an advance to contribute to the field of technical typesetting, test it carefully. Measure readability of different versions using experimental subjects who vary in age, eyesight, familiarity with the language, and subject matter skill. Of course, you may have a minimum level of subject matter and language skill.

Answer 5

This is admittedly a non-answer, addressing rather the background of the question.

Sifting through complex, dense, pedantic, and difficult-to-read texts is part of research in every field. Every field has bad writers, and every field has highly technical subfields where nothing makes sense without detailed formulas or frameworks. Every field has data collection, whether it's other people's (sometimes ancient) writing, noisy measurements, or blurry images. And it just takes time to filter out noise. Really, I would argue that 90% of any research effort is sifting and sorting and filtering until you find that nugget of truth to share with the world. There are tools to help with this-- thanks especially to progressions in natural language processing, machine learning, and data analysis-- but, at the end of the day, filtering noise takes time.

I think you would do well to improve for yourself how you personally sift through noise-- for example, there are many techniques for efficiently processing a research paper-- rather than investing time in trying to reinvent, well, writing.

By the way, papers are already interactive: it is always possible to e-mail the authors if you want more information. (Assuming they are still alive and active.) And if you want something that's really really interactive, make a video, applet, interactive code, etc.