Genuine question (and not an exercise in snobbery ;p), I wonder if many people's seeming disfavour for ns-qualified keywords comes from lack of (properly setup or understood?) IDE tooling. With cljr-slash (https://github.com/clojure-emacs/clj-refactor.el/blob/50d2d8aad5e0bd8002173b300f8419d72ceab7af/clj-refactor.el#L1968 ), I type ::foo/ and the right :require will be inserted in my ns declaration, and the requiring will be actually performed into the runtime that's an instantaneous op, and after that, tab completions for ::foo/ will be 100% accurate I can imagine that if lacking this tooling, using ns-qualified kws could be more tedious.

I don't find auto-completion/hinting to be any obstacle at all to using ns-qualified keywords, to be honest. My set up is Atom/Chlorine with its bare-bones completion (without Compliment). Stu Halloway says he doesn't use auto-complete/hinting at all in his editor setup -- and he seems to use ns-qualified keywords just fine 🙂

As I'm converting code at work from clojure.java.jdbc to next.jdbc I'm working more and more with (table)-qualified keywords and I like it. I'm even changing some old code that still uses c.j.j to use the :qualifier option to add a table-qualifier to the column keywords 🙂