The idea that personal genomics might render insurance irrelevant makes some logical sense. The only problem is that it oversells the science of prediction in biology, and underestimates the role of randomness in disease outcomes. … Yes, a non-trivial minority of people will find out that they have a high risk of developing a given disease early on enough in adulthood that the acturial tables imply that they’re uninsurable. But even those with “clean” results should probably still purchase some insurance plan to protect against ‘tail risk’. People who don’t smoke do get lung cancer, and people without a family history of heart disease and cancer do get heart disease and cancer.