x
Optimizing Multiple Imperfect Filters — LessWrong