> Full justification in monospace through word choice (which is what the Ares Luna article is about) is bonkers, but I can't see that happening for TV dialog.

> Full justification in monospace through word choice (which is what the Ares Luna article is about) is bonkers, but I can't see that happening for TV dialog. We're going to see more of this once diffusion language models mature. They won't match the exact length but will sure choose words to fix the set number of tokens.