it's layered
one layer (the fuzzy dark one one back) is for the human ...
... the layer in front (the thin lines) is what the LLMs tend to focus on, if trained on text scraped in the decoy font
mind you, it's not super human legible lol ... and I, a meat human, tend to focus on the thin-line robot layer myself 😅
but it's a fun concept
item #2 in my latest "Linkfest" newsletter, free to read/subscribe here: buttondown.com/clivethompson/archive/linkfest-47-a-decoy-font-reverse-minesweeper-and/
