Paper: Testing ecological models — LessWrong