I like the approach that Sean Carroll describes in this episode about being a science advisor for a movie. Rather than telling the director that their time travel theory is wrong, you treat the screenplay as data and concoct a theory to fit. Much better for collaboration.