Selecting Leaders with Random Sampling and Standardized Testing — LessWrong