Eliezer Yudkowsky comments on Error detection bias in research