> We have this (reasonably) rigid definition that grants the appearance of specificity. But there are almost surely UPFs that are fine and there are probably non-UPFs that cause the same problems as UPFs.

Not "almost" or "probably"; both are certain: Any description of reality fails on those standards. Always, there are exceptions and edge cases: 'We have this (reasonably) rigid definition that grants the appearance of specificity. But there are almost surely X that are fine and there are probably non-X that cause the same problems as X.'

To meet your standards, we would need a list of every food and its status - and that couldn't be created without impossibly extensive research.

That's why humans use imperfect models and categories - reality is too complex. The most specific, usable category seems to be 'ultra-processed foods'.

> So what is the benefit of saying "avoid UPFs" over "avoid junk food?"

Good question. Both categorizations will be flawed, but maybe UPF is a better fit or easier for people to understand.