After describing some joint work with Menezes in which isogenies are used to show that conventional wisdom about parameter selection might sometimes be wrong, I'll shift gears and make some comments on how easy it is to get things badly wrong in cryptography. I'll illustrate by giving a brief survey of some of the many misjudgments I've made over the years.