You assume A) that the set of interestingness values is both finite and discrete and B) that it's cardinality is extremely low. Both things can be remedied by defining a more comprehensive kind of interestingness that lets you compare more articles.