I'm not convinced this is an improvement. I don't think someone trying '!' in an f-string expression happens often enough to worry about that specific error message.

And a generic "optional specifier" isn't great. If we're going to do this, it should reflect the actual thing that's present: a conversion specifier, a format specifier, or debug specifier. And "optional" doesn't add anything, since it's actually present in this string.

I'll give it some more thought. I can't come up with an error message I actually like. The focus still needs to be on the empty expression.