I think the basic premise of multiple cursors editing as a preferred model (which is what Kakoune does) is flawed. Why? Because large-scale scale changes are the exception, not the rule. By emphasizing multiple cursors, Kakoune actively makes it harder to do the normal thing people do in a text editor: jump around a file and make specific edits to individual lines. Instead, a bunch of normal movement commands (such as forward and backward search) leave extra cursors in your wake, forcing you to press an extra key to dismiss them otherwise you'll get erroneous edits that might not even be visible from where you are in the buffer. This is not a good thing.
The other basic premise that Kakoune addresses (that vim commands lack visual feedback) is not even an issue for experienced vim users who type commands quickly from muscle memory. The edits happen when the command finishes and you don't get distracted by intermediate results. For global substitution, you can always append the `c` flag to add a confirmation prompt to each substitution. That alone obviates much of the motivation for Kakoune's model.