[ Question ]

Best way to write a bicolor article on Less Wrong?

by Viliam1 min read22nd Feb 20218 comments


Site Meta
Personal Blog

I am planning to write an article that would use two colors to distinguish between "map" and "territory", kinda like in The Neverending Story. In other words, instead of:

The sentence 'snow is white' is true if and only if snow is white.

I want to write something like:

Snow is white if and only if snow is white.

In general, the quote marks have the advantage of unlimited nesting, but I don't need it now; my article will have exactly two levels, not more. On the other hand, I feel that the bicolor version may be easier to perceive instinctively; the words referring to the map and the territory are not only surrounded by marks, they look differently by themselves. Also, quote marks are generally used for various purposes (quoting, irony, etc.), so authors emphasise the "this refers to map, as opposed to the territory" usage by not saying merely "X", but "the sentence X", "the words X", "the text X", etc. By removing such words, even my example here is shorter.

So, how to do this technically?

  • If I understand it correctly, Markdown doesn't support colors, but you can replace Markdown editor with Rich text editor in user settings, and the Rich text editor supports colors. Am I right?
  • Alternatively, instead of colors, I could use italics or bold font, just like I did in the example here. Not only is this simple in Markdown editor, but as an advantage, if some blind people read this, their software would probably pronounce those words in a different voice, making the article more accessible. (Is this true? Can anyone confirm?) The minor disadvantage is that I will not be able to use the italics or bold font in the usual way in the article, but I think I can live with that. Though, as you see, I love using italics, so I would probably use bold font for the map-words.
  • (For the sake of completeness, using CAPITAL LETTERS would also be an option, but it is an ugly option I would rather avoid.)
  • And maybe more of the above could be combined, e.g. the words referring to the map could be written using italics and different color. That would make them even more stand out of the remaining text. Is it worth the extra effort?

Is there another option I missed here?

Which option would you prefer? I care about everyone's opinion, but the perspectives of colorblind or blind readers are especially valuable, because using a wrong option might completely ruin the article for them.


New Answer
Ask Related Question
New Comment

5 Answers

You could use a monospace font, the same way that Douglas Hofstaedter distinguishes strings of a formal system in Godel, Escher, Bach. It's poetically appropriate because Hofstaedter was trying to solve the same problem you are: use typesetting to set apart map and territory. 

I personally find colors to be distracting and nonstandard ways of carrying such information.  I like Samuel Marks's idea of using monospace fonts for strings or formalizations as opposed to concepts.  I also VERY MUCH prefer talking about strings vs concepts than map vs territory, as the territory isn't actually available in any text, only different levels of map (in this kind of document; there are other conversations where it makes perfect sense to use map/territory).

Depending on the topic and text, you might also consider physical separation rather than typography as the differentiator.  A column for one level of narrative, alongside another column for a different (more objective) level of description.  Or more humorously, an OOC injection between paragraphs like

Narrator: Dagon thinks self-reference is fun.  Not everyone agrees.

But even more than any specific, I'd love to see some experimentation.  Write a sample paragraph, and try out the 3-4 leading styles.  If you're into it, write a preprocessor that generates multiple styles from the same input, and let readers choose their preference.  


If you ever want to do anything particularly weird in an article, you can send me plain HTML via the Intercom and I insert it into the post directly (after doing some basic sanitization). This will make the post usually admin-only editable (if you used any HTML features that are admin only), but works well-enough, and I've done this a few times for articles that really wanted to use color (Beth's AI Safety Debate writeup was one that comes to mind here)

Markdown also supports underline by surrounding the text with single underscores, though it looks a bit like hyperlink. I don't know off the top of my head if LessWrong supports that feature...
if this text is underlined, it does (Edit: looks like LessWrong interprets that as another way to do italic, likely to avoid the hyperlink confusion)

There are also other plain-text options you might consider.

  • You could use grouping characters other than quotes (parentheses, square brackets, angle brackets, braces) to distinguish map and territory.
    • You can also use unconventional grouping characters, like literally any punctuation marks that aren't reserved by markdown. Just be sure to test your choices first. I've had some unexpected results involving the $ character on LessWrong in the past.
  • There are also several styles of quote marks that might be useful for this kind of differentiation. Guillemets may be especially nice here as you could point them »in for map« and «out for territory».
  • You could explicitly tag map and territory statements in a number of ways, as long as you write a forward explaining what you're doing. This kind of tagging carries the advantages of being unambiguous to the reader, easy to remember, and almost certain to be rendered by screen readers. Downside is that they tend to be a bit clunky.
    • You could borrow namespace conventions from any programming language. e.g. map::"thing the map says", territory::"what's really out there"
    • I've seen fanfic enthusiasts compare works with a similar operator:
      e.g. map!"thing the map says", territory!"what's really out there"
    • You could borrow function/method syntax from programming as well. This has the advantage of already being solved for multiple parameters, in case you need something like that.
      e.g. map("thing the map says"), territory("what's really out there", "closely related thing")

My first thought (just upon reading the title) was to use the green hyperlink as one color, but that sounds clunkier than using bold or italics, so I'm not sure it's the best way to go. I would find the bicolor notation very helpful, fwiw, but I'm not colorblind and I don't use a screen reader, so I have no input there (and no idea how a screen reader would process a hyperlink).

3 comments, sorted by Highlighting new comments since Today at 10:55 AM

(LessWrong dev here)

Not an answer to your question, but FYI we deliberately don't support colors in our standard editor, because the naive way to implement would result in copy-paste errors when people copy their google doc over to the LessWrong editor. (i.e. I believe it pastes in text as "explicitly black" rather than "non-colored" and links as "blue", overwriting LW's default green links)

I think there are occasional places where colored text would be preferable but it's unfortunately tricky to make it usable-when-actually-important without screwing things up a lot of other times.

It's probably been brought up before, but colors are also not a particularly accessible feature for those who cannot perceive them. There are some easy (partial) solutions [1] that can result in this being less of an issue, but simply not supporting colors is certainly the easiest. Monospace is supported in markdown using the code ticks; perhaps it would be useful to be able to explicitly specify serif, sans, and script fonts as well? (Perhaps we already can and I just haven't noticed?) With the four font types, times three formatting options (bold, italic, normal), we'd have 12 available "colors" that should all be pretty easy to differentiate. I suspect that many would already be more than it's easy for most readers to keep track of anyway, so it should be more than enough.

[^1] Allowing bright orange and dark blue provides a high-contrast color set that can be distinguished by most individuals with various types of color-blindness. Standard and inverted are also easy for most sighted persons to see. Still, that even leaves out those without sight unless their screen readers are set up to detect those signals.

As a note, I see new LessWrong posts first in my RSS reader, which tends to ignore most formatting choices. It would be helpful to know at the start of a post that I should click through to the original page in order to get the best experience if you decide to use formatting to mark different concepts within the text.