I've heard it said that global continuous internal symmetries must be broken in any reasonable theory of quantum gravity. What's the precise version of this statement, and the reasoning behind it? (I know how to show this is true in string theory, but apparently there are more general arguments.)