Not dumb to me, you are saving yourself from a lot of quoting headaches.
If the text does not needed to be edited by a human, then I prefer ascii separators. Easy parsing in shell scripts.
Otherwise unicode is ok, I guess, because not every text editor can handle ascii separators correctly.
As I mentioned on Matrix, I did this once a long time ago, and it worked (though it looked silly). A lot of the criticisms boil down to a lack of support and convenience, which are things that can come in time; really, I think it’s unfortunate that they’re using the symbols because the characters don’t behave well right now. It’s not like any existing editors support USV anyway.
Just for fun, I tried it with the raw control characters and added a bit of Emacs configuration to make them use the symbols only for display purposes: