[theoh]

I'm guessing he means "you wouldn't be able to read these wacky characters in isolation, but in a body of text you can infer which Latin character each represents".

It's not unlike the sense of "context" of Context Mixing: https://en.wikipedia.org/wiki/Context_mixing

In the case of a human attempting to read his text, the context could be considered to include semantic hints, knowing how many letters are in the current word, etc.

Edit: The paragraphs he refers to may be those rendered in italics in the preceding section of the article, but their content is not important in any case.

[saycheese]

Makes sense, reminds me of these (hoax) meme on mixing letters within words: https://www.mrc-cbu.cam.ac.uk/people/matt.davis/cmabridge/

Unclear how this font in question addressed any real issues though; interested in the topic though, since I'm not aware on any science-based font system.

[theoh]

Typeface design is a generally an intuitive, convention-based activity without much of a scientific theory behind it. (see http://www.typotheque.com/articles/the_science_of_typography )

The designer Phil Baines used a strategy of subtracting elements from each character to design a difficult-but-possible-to-read font in the 90s called You Can Read Me. It was actually inspired by an experiment by Brian Coe which was reproduced in a book from 1968 called The Visible Word by Herbert Spencer. You get the idea--the development of typefaces is quite a slow moving activity which owes a lot to the subjective judgements of idiosyncratic personalities, usually male. Over the last century there has been a minority of practitioners who strive for scientific rigor, though.

"[The Visible Word] is a major contribution to legibility studies and presents a summary of over one hundred years’ worth of investigations by one of the UK’s most influential typographers. Spencers extraordinarily detailed 24 page bibliography is testimony to his investigation. The visible word is part of a programme of research into the readability of print in information publishing. In this book legibility is explored with equal thoroughness and objectivity. Resulting in the fact that people read most easily the kind of lettering they are used to. Although this may seem obvious in todays comprehensive documentation of the topic, much was learned from his lucid demonstration." http://www.designers-books.com/the-visible-word-herbert-spen...

I'm not sure where to find "today's comprehensive documentation" of legibility and I suspect that it is a little tedious and empirical...

But as an example, here's some more up-to-date research+bibliography in this article:

http://gaultney.org/jvgtype/wp-content/uploads/BalanLegEcon....

[saycheese]

Typography to me injects bias that I largely see as artifacts of the past; your comment that people read lettering best that they know speaks to this.

Closet I've seen to research I more interested in was a grid of pixels randomly configured, displayed to the user for them to recall, hidden, them compared to what the user was able to recall. Study found that patterns on the borders of the grid were more likely to be recalled.

Thanks for sharing all the info and links, I look forward to digging into them.

[SomeHacker44]

I think this is exactly it.

Before reading these HN comments, I forwarded the article to my wife, saying effectively "I was amazed that I was able to read the paragraph of whacky text in Ivo font, even as, after reading it, I went back to look at the font and then realized I couldn't read individual words or letters!"