One other project that I really love was presented at the 2021 Sigbovik: Fontemon, created by Michael Mulet. It's a full-blown choose-your-own-adventure game in a font. Truly insane. Here's a short video showing the basics of how it works: https://www.youtube.com/watch?v=VY1hDQMeU3A
It's interesting that mapping random noise to characters via OCR generally produces valid perl... but I always hated how they phrased the title of this experiment since it's obviously bullshit. Essentially, a good interesting experiment made less interesting by a sensationalist title.
Perl programs are, by definition, text. So "paint splatters are valid Perl" implies that there's a mapping from paint splatters to text.
Do you have a suggested mapping of paint splatters to text that would be more "accurate" than OCR? And do you really think it would result in fewer valid Perl programs?
“Feeding garbage to OCR” is a really boring way of generating text. I was assuming it would be something more interesting, like creating a symbolic representation of the splatters and generating text from that. Using OCR is basically piping /dev/urandom to perl and seeing what happens. The fact that they’re valid perl programs is worth a laugh but the generation method is totally uninteresting.