A sane policy would be that there's a process for verified accounts to change their name, and that process is that some human at Twitter sees the account's current name and new name and rejects changes that aren't obviously still "the same person." A policy like this could prevent these kind of "pretend to be Vitalik or Elon or whoever" schemes without actually having to review the verification itself.

So you can still allow hashtags and emoji and so on being stuck onto names, still allow "spooky" Halloween names, but preserve some kind of value from the verified checkmark.

I don't know how many verified users there are but this doesn't seem like it would necessarily be onerous. Of course, it's still more onerous than just doing nothing about it is.