Blue Orange
Gone to Texas
If a game design does not do what it intends or claims to do, it is a bad design. If a game is confusing or difficult to read or otherwise inaccessible, it is badly written and/or laid out. These are real metrics. They have degrees in game design for a reason.
I'd argue the exception is that if a game design doesn't do what it intends or claims to do, but is good for something else, it then becomes a good design for that purpose.