Re. Lispy langs, agree about improving readability of the actual words by visually deemphasising punctuation. So many key presses to say so little; only traditional C syntax beats it out.
(Alas, nothing can ameliorate C’s swathes of needless line noise, except to stop making languages with C-style syntax. Baby steps.)
I don’t quite agree about structure editors—if there’s one thing 10,000 years of the written word should tell us it’s that the writing part needs to be flexible. Better to beg forgiveness than ask permission. It makes more sense for a code authoring tool to accept (and execute, within limits) the initial imperfectly constructed text, and from there negotiate edits and corrections with the author—a literal editor, if you like.
Another thing I’d like to see syntax highlighting do: provide multiple semantic views onto code beyond just crude grammatical categorization of words. Why can’t I have a view that colors words according to the library they came from, or any other high-level categorization? e.g. Show me all the words that have side-effects or perform “dangerous” operations. Or distinguish all the words relating to string manipulation, or accounting, or UI.
Embarrassing? Honestly, it’s painful the lack of innovation going on. While I’m gradually exploring some of this stuff in my own work, I really wish others would step up to the plate. I’m lazy and very slow, and would far rather smarter people do all the work so I can just enjoy the results.