There is some interesting discussion on the LaTeX stack overflow page about the challenge of detecting and preventing rivers during the typesetting process:

https://tex.stackexchange.com/questions/4507/avoiding-rivers...

https://tex.stackexchange.com/questions/29049/how-to-define-...

And if you really want to get into it, there is a rather detailed paper by Alex Holkner: https://gwern.net/doc/design/typography/tex/2006-holkner.pdf

Avoiding rivers becomes a rather non-trivial optimization problem. In Holkner's paper he found that it took ~1 minute just to typeset 1200 words. Some of his experiments took more than six hours to complete.

Futility Closet had a cheerful, wholesome podcast for many years until abruptly ending it without much explanation. Glad to see they're still busy. Anyone know what happened? I always wondered.

Since Christian Bök’s Xenotext was recently discussed here (https://news.ycombinator.com/item?id=43974005), check out “The March of the Nucleotides”: https://x.com/christianbok/status/1064741910948995072

In this case the edges of the river represent DNA base pairs: A on one side has always a T in front of it; the same with C and G.

> Because they’re distracting, these artifacts are generally discouraged

...

> (squint to see it)

This seems rather contradictory?

WebKit is addressing this problem.

https://webkit.org/blog/16547/better-typography-with-text-wr...