Guardian article highlights observational biases in Knox investigators — LessWrong