I ended up switching to Kakoune and Dance mainly for the multiple cursor support. I treat it like `sed`, but interactive. It feels miles ahead of other editors I know, with the exception of Helix, Vis, and maybe Ki (haven't played with it enough to tell).
> The basic premise of Kakoune, which Ki seems to share, is that it's more natural to move first and then command. That's not how my brain works:
I also didn't find this as a huge benefit, though some people do. That's why I'm sorry Vis (link in another comment of mine) is not more popular; I could far more easily recommend it to other people if it was as polished as Kakoune and maybe had a VS Code plugin.
---------
Now, for some things I don't fully agree with.
> I disagree with the premise (from the Fine Article) that vi/vim is in some way less coherent as a language. Languages don't really do 'coherent', they do 'expressive'.
I think Vim is definitely less coherent. The first example that comes to mind: it's easy to get to the first line with `:1<enter>` or `gg`, while getting to the last line is `G` for some reason (Update: TIL that `:$<enter>` also works, which makes a little more sense. But `:|<enter>` does not take you to the first line). Getting to the first column is `|` I think? I can never remember it, fortunately `<Home>` works.
Now, where I do agree is that, IMO, it's an open question whether making a modal editor's language more coherent than Vim's would make it more expressive or easier to learn. Vim's language is pretty good. My personal feeling is that Kakoune maybe gets close to Vim's bar, and maybe even exceeds it, but not by enough to make it a fundamentally better language (except for its multiple selection support, where it is fundamentally better).
My hope is that Ki might be better, or that its ideas might lead to a better language. I am not certain this is the case, but it should make the situation I described two paragraphs ago better, at least.
> Code golfing, I count 12 keystrokes in your example, and 10 in stock Vim.
To me, the number of actual keystrokes matters less than how much space they take up in my mind and how easy they are to think of (or, better, perform reflexively). If I typed faster than I thought (e.g. if I was transcribing a voice recording or translating), I might feel differently.