That's my thinking as well: just re-brand string theory as math. It's not as though string theory will be testable anytime soon, so the criteria of verification is just hurting the field.