I like to talk of 'correctness' (as in the objective quality) and 'rigidity' (subjective; generally better for larger programs and programs that need to be correct, but potentially worse for prototyping and flexibility).

Ultimately, what I care about as a programmer is, if I write/refactor/tweak some code in this language, how many weird edge-cases are ruled out or handled for me? How many unit tests do I have to write, to ensure this myself?

Python is relatively bad at that, even though it's technically memory-safe, as the post mentions. The dynamic typing, for example, means some refactoring mistakes will just not show, unless you've got close to 100% integration test coverage.

So, yeah, I feel like this whole "memory safety" buzzword is lost on pretty much everyone working in a garbage-collected language, even though many of them would be extremely glad about a more correct and/or more rigid language.