The Logical Fallacy of Generalization from Fictional Evidence — LessWrong